2021-09-07 16:49:16 @asajeffrey Just seems like the more respectful default for a collaborator is to assume experience until they reveal not having it 2021-09-07 16:35:54 @asajeffrey Collaborating with anyone with a PhD, I'd just assume those skills because I feel like it's insulting to view people with PhDs as still existing within some arbitrary hierarchy 2021-09-07 16:31:29 @tekknolagi Chief Latke Officer 2021-09-07 16:30:37 @asajeffrey This kind of toxic title-and-time-based respect is what I hated most about being a software engineer, and it's really painful to think people still hold to it even after I finish my PhD 2021-09-07 16:29:25 @asajeffrey Sorry, two undergrads during grad school, on their own work, in addition to folks I mentored on my own project. Also, I'm PI on a grant I was a key player in and I'm coordinating significant parts of the collaboration between institutions 2021-09-07 16:27:40 @asajeffrey Hm, OK, I think I'd have a problem collaborating with anyone who views me as inexperienced at mentoring (I advised two students during undergrad, and run an international mentoring program), leading project teams (I lead all of my thesis work and am leading 3 projects now), etc 2021-09-07 16:16:40 RT @hardmaru: “The brain is able to use information coming from the skin as if it were coming from the eyes. We don’t see with the eyes or… 2021-09-07 16:13:18 @tekknolagi Senior Principle Tater 2021-09-07 15:58:39 @asajeffrey Out of curiosity, do you end up viewing academics similarly? Like do you view assistant professors similarly to how you view fresh PhD industrial researchers? I think I'd have a problem with being viewed as inexperienced at those things, which is why I don't assume it of others 2021-09-07 15:56:56 @tekknolagi Professor Potato Head 2021-09-07 15:42:05 @tekknolagi Senior Vice Chief Potato 2021-09-07 15:38:02 @asajeffrey I generally hate hierarchies though 2021-09-07 15:37:01 @asajeffrey Ah, I like it, it makes me feel like other people view me as capable and I can stop navigating arbitrary hierarchies. Plus I still get a mentor 2021-09-07 15:34:12 @asajeffrey I guess they hire us expecting we will be good at it, since they're trying to make lifelong hiring investments and they're necessary parts of the job even very early on? 2021-09-07 15:28:46 @asajeffrey I just wonder why it's different for faculty? Like we just get all of that thrown at us on day one, and many come in with experience in it already 2021-09-07 14:49:54 @SocScaleTech My ideas are basically ways to bootstrap students, but I want them to do well early so they feel empowered 2021-09-07 14:49:26 @SocScaleTech True, I just hope other people value my students' ideas too!! 2021-09-07 14:47:43 @skornblith @NeelBhandari9 @ChrSzegedy I think I'm less convinced that the "human touch" will be valued, though if it is that will be interesting, since it will significantly change western gender dynamics 2021-09-07 14:36:33 You can get business cards that say whatever you want, right? I should get Senior Staff Vice Principle Whatever 2021-09-07 14:32:11 Likewise undergrads just view all faculty as Powerful Scary People (fairly, we have direct power over undergrads in a really consequent way) and junior grad students vary a bit, but senior grad students are like "eh, I'm almost there" 2021-09-07 14:29:39 So on my end I find it super weird to view a new industry hire any differently from a Senior Staff Vice Principle Whatever, and on the industry end there's a sense that I'm low in an imaginary faculty hierarchy 2021-09-07 14:27:44 I think the disconnect with industry may be that industrial new PhD hires enter still fairly low in a hierarchy with lots of job titles to ascend, but pretty much no faculty I know of actually view assistant, associate, and full professors at different levels of respect 2021-09-07 14:24:24 @xu3kev Yes!! I got it at a restaurant supply store in Tokyo haha 2021-09-07 14:22:58 I guess other faculty treat me the way I'd like to be treated so that's cool. So do senior graduate students 2021-09-07 14:22:22 Students are always on the "way too much" side IMO, the "way too little" comes from outside academia largely 2021-09-07 14:21:42 It feels like going from PhD student to faculty corresponds to a jump from not really being respected much by most people, to either being respected way too much or way too little, when I just want to be viewed as an informed and experienced intellectual equal with area expertise 2021-09-07 14:09:00 @evanoconnell @robertcaruso I love how it's like "you probably miss peanut butter so much you want it for all meals" like you're not wrong 2021-09-07 14:08:06 RT @evanoconnell: Always amused by British/American food sections in foreign supermarkets. This is at the Intermarché in Gaillard, a town j… 2021-09-07 14:07:13 RT @emmaogreen: Matt Hawn believed students in his nearly all-white high school in northeast TN needed to discuss racism + white privilege.… 2021-09-07 14:02:12 @skornblith @NeelBhandari9 Bots are already so hilariously good at memes: https://t.co/hpHkHb76ir 2021-09-07 13:57:17 RT @AntonyHosking: Some useful advice in these Down Under lockdown times and in general. I've decided I am only able right now to manage Op… 2021-09-07 13:53:38 I mean, if they don't try to poison you on the airplane, sure https://t.co/TPMyd41ybA 2021-09-07 13:38:19 Talia, child of ______ https://t.co/ZCZkshZVxR 2021-09-07 13:18:23 RT @Aaron_Horowitz: NIST is about to seek nominations to AI advisory committee, and subcommittee specifically on law enforcement. So far co… 2021-09-07 13:17:14 @NeelBhandari9 @ChrSzegedy I think the desire to feel useful is just core to humanity. People without purpose get bored and start cults, join terrorist groups, go to war, and so on. Socializing is fun but only if you have some greater purpose and direction to your existence 2021-09-07 06:00:40 I'm doing my best to be useful so why do I always feel like everyone views academics as useless 2021-09-07 05:59:38 Like if that's the case why I am not still at Amazon as an engineer like I was before grad school, what was the point of learning and generating new ideas from what I learned 2021-09-07 05:58:40 I went into academia in part to help students and in part to generate and explore ideas fearlessly and collaboratively. It always makes me really depressed when people make me feel like ideas aren't valuable, because what is the point of my existence if that's true 2021-09-07 05:44:12 @jamescham In pursuit of things it feels like nobody will ever really value me for 2021-09-07 05:42:21 I feel so useless 2021-09-07 05:39:38 Bad day 2021-09-07 03:46:10 @RoxanaDaneshjou Ah yes, the @geomblog principle in action once again 2021-09-07 03:44:59 @RoxanaDaneshjou I appreciate that, but I think ultimately what is profitable will win, and misery of humanity may be briefly profitable before driving the market to explode 2021-09-07 03:40:59 @NeelBhandari9 I hope you're right, but these last few years have not given me much hope that will happen 2021-09-07 03:39:37 I guess I believe that explainable, understandable, trustworthy automation with a human in the loop is not inherently bad, but human agency and expertise and usefulness is essential. I worry we won't be able to control use of automation enough to ensure those remain 2021-09-07 03:36:44 @NeelBhandari9 Yeah, I think just watching recent trends, I'm not sure we'll be able to control deployment enough to ensure humans are in the loop and AI is just helping 2021-09-07 03:26:55 @Skiminok Whoever gets you next has no idea how lucky they are 2021-09-07 03:22:28 So I often worry that maybe automation is just fundamentally bad, even though I really enjoy building it 2021-09-07 03:20:12 In particular, people fundamentally enjoy working and being useful, for the most part. Even if you cure cancer, prolong life indefinitely, and care for everyone's individual needs, without something to make people feel useful, everyone will be depressed 2021-09-07 03:18:24 @NeelBhandari9 If you make work into just a hobby, we'll still all be miserable because we will know that we are not actually useful 2021-09-07 03:17:20 @NeelBhandari9 But humans fundamentally enjoy being useful, so if you take away work, even providing for our individual needs, we'll all be miserable 2021-09-07 03:02:41 My friend got me this absolutely beautiful Turkish coffee set from Turkey and I'm quite excited to serve coffee! Someone come over and drink coffee https://t.co/HconHMTukC 2021-09-07 02:34:09 I think we're all the bad guys in history, except maybe the HCI folks and the AI Ethics folks 2021-09-07 02:33:04 Every time I get excited about work I feel scared that I'm participating in the study of technologies without adequate investment in studying their societal impacts and then I wonder if I should actually be doing this 2021-09-07 02:18:30 FR though I think people would be absolutely miserable in a world with (unregulated) AGI 2021-09-07 02:10:38 What if you ask AGI to optimize for human happiness and it destroys itself? 2021-09-07 02:10:13 What would the societal and economic implications of AGI be, in your opinion? How would people react? What would keep people happy and give people a sense of purpose? Hell, what would stop people from fighting to destroy it, to regain agency and expertise? 2021-09-06 23:31:41 Lots more to do in the living room, but going to take a walk first. Very tired 2021-09-06 23:30:27 Ok, after 3 days of full-time unpacking, my bedroom is finally done. Complete with a meditation corner with a folding wall for when I need to feel like a cat https://t.co/OKIWiTm1EB 2021-09-06 19:51:33 @TobyMeadows The dude is a Taliban spokesman 2021-09-06 19:46:49 Minimize your TCB https://t.co/ctNiz5FdqH 2021-09-06 19:45:01 @hillelogram Oh yeah I always hated all arguments like "imagine an infinite being" or whatever like can you actually conceptualize of the infinite? I can't, I can only imagine finite approximations of the infinite, or infinite functions 2021-09-06 19:41:32 Like why is existence a positive property? Y'all tried existing lately? 2021-09-06 19:41:14 Classical logic is fine, but the Taliban should minimize their TCB. Anselm's proof has a large TCB, so it's much more accurate to say "assuming these definitions and these axioms, God exists," but the definitions and axioms are super weird 2021-09-06 19:38:32 @hillelogram For this argument it always bothered me that existence was considered "positive" 2021-09-06 19:21:42 @kamatsu8 Oh yeah this Tweet is mostly just dark humor 2021-09-06 19:18:55 For real though please don't use formal proof to justify causing harm 2021-09-06 19:17:26 The Taliban would be into classical logic based proof assistants SMH https://t.co/yqFJAMwvFz 2021-09-06 19:06:23 RT @vamchale: the taliban is posting Isabelle source code screenshots 2021-09-06 17:54:27 @ncooper57 @mark_riedl Cool, will read 2021-09-06 16:56:02 More ideas with the help of @alexanderklew https://t.co/Bn8TauGQL4 2021-09-06 16:52:39 @alexanderklew @mark_riedl If so, then the notion of a dependency graph just needs to be refined to also consider syntax, and maybe itself learned 2021-09-06 16:47:59 RT @yayfrens: You do not owe it to anyone to look a particular way. 2021-09-06 16:44:55 Observational user study of proof engineers tracking all of the information they use to write their proofs, with a goal of evaluating whether automation has access to all of that information, and improving proof automation in the future 2021-09-06 16:38:19 @bloodyquantum @mark_riedl In programming languages, this basically refers to determining all information that a current piece of code depends on, usually by building a graph. https://t.co/Pkblz6XnDP, https://t.co/87NBs6srBY 2021-09-06 16:27:41 @roydanroy @ShlomoArgamon @mark_riedl I gave a few examples, including one of me as a child and one with my 6 year old niece. Parenting is hard but I don't think the hard part is getting children to grasp general concepts 2021-09-06 16:03:34 @roydanroy @ShlomoArgamon @mark_riedl Good conversation on context selection: https://t.co/b8o2vr6IXT 2021-09-06 16:01:33 Luca had a cool idea for this informed by prior work: https://t.co/b8o2vr6IXT 2021-09-06 15:59:38 Just, I think this could be done much more efficiently and effectively 2021-09-06 15:58:26 Context is the missing piece in that right now, so the shift to focus on context is actually kind of cool and already means they're moving from "bigger is better" to "OK let's try bigger in a different dimension" which is something! 2021-09-06 15:56:28 I suspect that by just making the context really, really big, OpenAI is going to try to determine the limits of what can be done with really big things that have access to all of the raw text humans have access to, and that is still an interesting question 2021-09-06 15:54:51 More on context: https://t.co/5rCGqwlAFp 2021-09-06 15:54:00 OTOH the focus on the context is going to lead to pretty big gains for programming and proving models, so that's nice https://t.co/6fn9z262kk 2021-09-06 15:48:40 Definitely preaching to the choir responding to a @mark_riedl Tweet with that though 2021-09-06 15:46:06 I know I'm obsessed with this line of thought and the three examples I've given of similar things humans do beautifully, but I'm quite serious about it, I want this beautiful synthesis of general specifiable principles and example-based learning https://t.co/w94X3NOHY5 2021-09-06 15:37:20 @roydanroy @ShlomoArgamon @mark_riedl (But for code models there is so much low-hanging fruit wrt context that I expect just making the context larger will improve performance. Better to make the context better, though, like dynamically select it based on a dependency analysis, I want to try that sometime) 2021-09-06 15:35:08 @roydanroy @ShlomoArgamon @mark_riedl I think if benchmarks aren't showing limitations of just making the model or the context larger, probably the benchmarks are the problem 2021-09-06 15:34:15 @roydanroy @ShlomoArgamon @mark_riedl https://t.co/iV09IO2TEN 2021-09-06 15:32:57 @roydanroy @ShlomoArgamon @mark_riedl https://t.co/obCdJFalcb 2021-09-06 15:31:22 @roydanroy @ShlomoArgamon @mark_riedl https://t.co/JjMA5ccVc1 2021-09-06 15:28:12 @roydanroy @ShlomoArgamon @mark_riedl The ability to correct an incorrect model with robustness issues with a known general principle and have the model repair itself to never be wrong in that case again, but lose very little else would be nice. Human children are great at that 2021-09-06 15:24:15 RT @limit_breakr: This guy keeps blocking walkways and escalators and I've never been so entertained by inconvenience in my whole life http… 2021-09-06 15:20:53 RT @TaliaRinger: @mark_riedl OTOH I do expect better context to improve the performance of these models on code dramatically. Right now the… 2021-09-06 15:19:34 @mark_riedl OTOH I do expect better context to improve the performance of these models on code dramatically. Right now they're like programming in Gedit with vertical horse blinders on. Though dependency-analysis based context is ideal for coding 2021-09-06 15:05:51 @SkyisreallyHigh @ahmadihamed_ https://t.co/FLi9FXgKjv 2021-09-06 14:42:29 This basically reads perfectly consistently if you consider how gathering and annotating good data is (insufficiently) valued https://t.co/XziNldHe5i 2021-09-06 14:16:30 @Mottel Interesting! 2021-09-06 14:06:01 Hopefully you right now: https://t.co/QvUyoFsSSY 2021-09-06 14:01:11 שנה טובה! https://t.co/TY83F8cXDt 2021-09-06 13:50:16 @Mottel Hmm I thought honey historically referred to silan though 2021-09-06 13:33:14 @jer_gib I applied in the US, Israel, Canada, Denmark, and Australia, and all followed the US season 2021-09-06 05:15:30 RT @pi8027: Algebra-tactics now support ring morphisms! They generate two kinds of reified terms: one for pushing down morphisms, and anoth… 2021-09-06 04:49:49 RT @mcmillen: It's common to see comparisons of BTC's energy use to various countries/states, but "seven times as much electricity as Googl… 2021-09-06 04:41:54 @secretasianman Hard to find in Champaign IL, but in a lot of cities there are actually coparenting arrangements where families will live together and also raise kids together. Maybe chaotic, but honestly if I'm a single mother at any point I'll look for this 2021-09-06 04:38:52 @secretasianman People were genuinely surprised that I asked about it as incoming faculty, and there was almost nobody to ask at any university I applied to 2021-09-06 04:29:05 Or some nonsense about "independence"? 2021-09-06 04:27:25 Is this a wealth signaling thing? A class signaling thing? 2021-09-06 04:26:33 I kind of get now that some people genuinely prefer living alone. What I don't get is why there's social pressure to do it if you're single and in your 30s or in a high-status job, and why people who don't like it cave to that social pressure https://t.co/t48pUIDQP9 2021-09-06 04:20:44 @BowToChris Found the house on Craigslist :) 2021-09-06 04:19:12 Anyways, if you're single faculty, you don't have to live alone 2021-09-06 04:18:13 It doesn't matter that several of them are grad students, either, because they're grad students in totally different fields, so there's no power dynamic. They actually appreciate the perspective since they're considering faculty jobs so they pick my brain about faculty life too 2021-09-06 04:17:00 Moving into a house of 5 was a very good move. Just starting to get to know my housemates, but they're wonderful and it's just always friends to talk to. So glad I didn't cave to the "real adults should live alone or with partners" social pressure 2021-09-06 01:49:16 @dinabass That sounds wonderful! 2021-09-06 01:49:00 @dinabass The person I visited could really use a new friendly visitor if you know anyone. She is a Soviet Jewish refugee, elderly, very kind and always gave me tea and breakfast. I miss her 2021-09-06 01:47:12 @dinabass Good to know!!! A donation to JFS never hurts. In grad school I did ESL and friendly visits for them. So rewarding 2021-09-06 01:45:31 @stephentyrone No education past age 10, and turned down an offer to study at the Technion to stay on a farm 2021-09-06 01:45:07 @stephentyrone My grandfather taught himself to be an engineer, then invented farming machinery for harvesting grains that continues to feed the world to this day. He worked in a machine shop and realized the machines were inefficient so he got obsessed with making more efficient ones 2021-09-06 01:43:15 @dinabass I volunteered with JFS Seattle for years!! I'll miss it so much 2021-09-06 01:42:52 RT @NYCAntifa: REMINDER: Never be a volunteer Nazi Safety Advocate 2021-09-06 01:38:44 @stephentyrone Many Holocaust survivors died shortly after liberation because they were starving but not supervised, so they ate too much, got sick, and died 2021-09-06 01:36:01 RT @ghaz: @TaliaRinger I've been donating to @mirriam71's amazon wishlist. They're currently trying to gather funds for a truck to take the… 2021-09-06 01:35:16 Anyways my view on refugees is we should take in as many as want to come and give them all the tools to become healthy and productive members of society 2021-09-06 01:33:52 - The first time he tried to immigrate to British Mandate Palestine, the British tear gassed his ship full of Holocaust refugees and sent him back to Cyprus. - The second time, he succeeded. - He then worked on a Kibbutz, but was treated poorly because he was a Holocaust refugee. 2021-09-06 01:32:37 - An underground Zionist group took him in, but had to pretend not to be a Zionist group. - Eventually he was one of the Selvino Children in Italy. - From there, he went to a refugee camp in Cyprus. 2021-09-06 01:31:13 When my grandfather got out of a death camp it was really hard for a long time: - Russians gave him a small amount of money and two horses. Later they stole one of his horses. - He squatted in abandoned German houses and ate their food. - He was then homeless in Poland. 2021-09-06 01:29:21 Are there funds I can donate to that go directly to Afghan refugees settling in the US? Grandchild of Holocaust refugees and illegal immigrants and this always matters to me a lot 2021-09-06 01:23:48 @ahmadihamed_ If you want a good meal once you're out of there I'm happy to pay for a nice big hot restaurant meal, or a week's worth of groceries or something. DM me. Grandchild of refugees, and grandfather had to eat food from abandoned German houses to survive for months after the Holocaust 2021-09-06 01:19:38 RT @dog_rates: please don’t spread misinformation. it totally was for him. 14/10 2021-09-06 01:18:15 @RezaRob @mjskay The expectation seems to be calm stoicism in reaction to paper rejections, and I don't think this is realistic for most junior computer scientists, but it's especially unrealistic for junior computer scientists with ADHD or autism 2021-09-06 01:16:52 @RezaRob @mjskay The criticism can definitely be valid. My point is just to be compassionate toward people who have really, really strong negative feelings in reaction to rejection 2021-09-06 01:13:41 It's just a few hours in the evening at a coffee shop, not the whole day 2021-09-06 01:13:01 I'm kind of a fan of Service Sunday, it gets me excited for the week, it is fairly low-energy and high-reward, and it also means I can focus on other work during the week 2021-09-05 22:57:48 @splashcon https://t.co/uKHDxA1Iv0 2021-09-05 22:57:40 Register for @splashcon today! Early registration closes September 18th. 2021-09-05 22:56:47 How many Turing Award winners do you think we'll have giving hybrid AMAs at SPLASH 2021? Answers coming in the near future once the schedule is official 2021-09-05 22:10:26 @palvaro Yum 2021-09-05 21:56:39 A word of advice for Richard Spencer on making friends, if he gets out of this OK: next time, don't be an asshole. People like people who aren't assholes 2021-09-05 21:55:13 "Nobody loves me or wants to be my friend," says man who dedicated his life to hate and murder based on race 2021-09-05 21:49:48 RT @VanCarlito2003: FYI Being apolitical is still white supremacy. 2021-09-05 21:39:25 Richard Spencer this is for you: 2021-09-05 21:37:49 "My life is ruined," says white man who has dedicated his life to genocide 2021-09-05 21:36:01 Haha oh gosh darn it being a literal Nazi has consequences who knew https://t.co/qjyd8KE8jn 2021-09-05 21:13:47 @RezaRob @mjskay https://t.co/UlwL0r3Rqa 2021-09-05 21:13:36 @RezaRob @mjskay https://t.co/5uaTTMZB2Q 2021-09-05 21:13:24 @RezaRob @mjskay https://t.co/8DgHllHjgB 2021-09-05 21:13:14 @RezaRob @mjskay https://t.co/xVJfZk9lpE 2021-09-05 21:13:02 @RezaRob @mjskay https://t.co/eNWSOMVe1Z 2021-09-05 21:12:48 @RezaRob @mjskay https://t.co/q7fXIWC4Fs 2021-09-05 21:12:37 @RezaRob @mjskay https://t.co/vE8OFO46m2 2021-09-05 21:12:24 @RezaRob @mjskay https://t.co/0XxYF7CPkw 2021-09-05 21:12:10 @RezaRob @mjskay https://t.co/FxBVtpNI6y 2021-09-05 21:10:59 @RezaRob @mjskay It is a fact though, something like 99% of people with ADHD have disproportionate rejection sensitivity. Not sure on the stats for autistic folks. Google "rejection-sensitive dysphoria." I have a lot of reading on this 2021-09-05 21:05:45 @avt_im @ZivScully But since refactoring is frequent, it really helps to have files broken down by section, and sometimes even by subsection 2021-09-05 21:04:40 @avt_im @ZivScully It's very similar to how I program. My grad school advisor called me the fastest writer he had ever met. But it is not linear, it is very chaotic, and this always makes it hard to get feedback early on, especially on longer documents 2021-09-05 21:03:20 @avt_im @ZivScully My writing is chaotic, but basically I just sit down and words happen, in no real order, and I just roll with it. I'll start with an outline but it's mostly a guide, not final. As the words happen, the paper starts to take shape, and as it does I refactor and rewrite and trim 2021-09-05 21:00:04 @falsenov It was really hard when studying Russian as a foreign language though, because I couldn't always tell when I had missed a detail, or when Harms was deliberately being absurd hahaha 2021-09-05 20:59:13 @falsenov Даниил Хармс is so so so so so good 2021-09-05 20:56:51 Thank you for coming to my TED talk 2021-09-05 20:56:36 Once you get a job, check out this guide to surviving faculty life with ADHD: https://t.co/HUeBlQAnxq 2021-09-05 20:55:10 As many of you are about to embark on a wild journey of the academic job search during a pandemic, never forget this priceless advice https://t.co/aBrWh0Z4G9 2021-09-05 20:51:46 @hillelogram Here is an absurdist humor story I did for my Russian tutor: https://t.co/mwcl32wDwJ 2021-09-05 20:49:24 @hillelogram Lool 2021-09-05 20:48:53 @hillelogram I've written Russian poetry and prose. I really enjoy it, it is a good language for writing 2021-09-05 20:47:25 @avt_im @ZivScully Wild 2021-09-05 20:45:26 @hillelogram Oh, not that I know of, but this would be fun to do. Maybe I'll do a collage poem this way with some of my books in multiple languages 2021-09-05 20:41:33 @hillelogram And poetry. In multiple languages 2021-09-05 20:37:00 @hillelogram Me: why are all of my friends either Russian, Israeli, or neurodivergent Also me: https://t.co/pXEw79ZKRW 2021-09-05 20:35:18 Also chocolate milk because I was drinking chocolate milk don't judge 2021-09-05 20:34:59 Bottom shelf is collage poetry kit, if I get sick of a book I might move it down there and use it for collage poetry 2021-09-05 20:34:04 @hillelogram Yes we can bond over common hobbies like studying Russian, reading Vladimir Putin's autobiography, and understanding every detail of narcissistic abuse 2021-09-05 20:33:14 Oh hey @chick_in_kiev your book is in there though 2021-09-05 20:31:31 Nothing screams "why yes, I am a fun human and I have a life and friends" like my leisurely reading collection https://t.co/dlkbKDu6Bb 2021-09-05 19:15:24 RT @lindsey: I *would* be stressed about my negative POPL reviews, but they're gonna have to get in line behind being stressed about moving… 2021-09-05 18:57:57 Getting there. Not too much left. Making it more interesting by doing bike intervals every time I go down to recycle a box because I'm restless and this is boring https://t.co/W1nZLjIRYA 2021-09-05 17:49:58 @ZivScully Sections are a natural abstraction layer for papers though. Each section is parallel, but they are not at all at the same level of abstraction as the top file that includes them. There is a natural hierarchy 2021-09-05 17:48:37 @avt_im @ZivScully I should note as a very very ADHD person I write extremely nonlinearly. So, the end organization of a paper that I write almost never resembles the outline I started with. Constant refactoring of sections until it feels right 2021-09-05 17:47:22 @avt_im @ZivScully Just like if a Wiki page gets too long, maybe it should be multiple Wiki pages. Separation of concerns 2021-09-05 17:46:30 @avt_im @ZivScully If I had a 20 page markdown project, and markdown actually made it easier to include things from other files, for sure I'd organize markdown files that way. But I don't even know how to do that, and if it exceeds 20 pages in markdown I question what I'm doing to begin with 2021-09-05 17:44:54 @avt_im @ZivScully It's also extremely useful for things like thesis writing, where you end up rearranging many sections in your thesis, or pulling in just some parts of a paper but not others 2021-09-05 17:43:19 @avt_im @ZivScully I think it's really useful when combined with a practice of "locking" files on Slack. Also, for renaming and refactoring sections and keeping the overarching structure of your document in one place, which happens frequently during paper development 2021-09-05 17:41:33 @AlejandroPiad That's fair, that's a reasonable coping mechanism too 2021-09-05 17:00:36 RT @Reductress: Man Blissfully Unaware: https://t.co/Y8XZvwYsEH https://t.co/AhcwwXIbO5 2021-09-05 16:46:29 (The program was CSSI, which I still recommend: https://t.co/UkUbqf60Kw. It felt like the first time anyone believed in me, which was empowering. I still have extremely close friends from this program, which I did in 2009.) 2021-09-05 16:41:26 Can unpacking be my exercise for the weekend? I'm so sore and exhausted 2021-09-05 16:36:21 I am, in the abstract, pro-tech, pro-industry, pro-ML. Just not as implemented, and the implementation details really, really matter 2021-09-05 16:34:05 Open and constant criticism of the machine learning community? I very strongly want to collaborate with that community to build better programming and proving technologies. It matters a lot to me that the community has reasonable norms for those collaborations 2021-09-05 16:32:19 Open and constant critique of diversity Google, earlier this year? Literally it was a Google diversity program that convinced me to major in computer science. It matters to me that Google gets diversity right because I've been paying attention to diversity at Google for 13 years 2021-09-05 16:30:45 So like, open and constant critique of the tech industry? It's where I was before I started grad school. It's the group of people I want my tools to help. I want the industry to use that help for good, and I want the industry to respect academics 2021-09-05 16:29:32 Oh a good general thing to know about me is that if I spent a lot of time critiquing something, it probably means I really care about the thing and want it to be better. Things I hate I just don't waste much time on 2021-09-05 16:27:20 @stephentyrone It does drive me nuts that non-researchers are sometimes in the middle of research management chains at some companies 2021-09-05 16:26:01 @stephentyrone But a creepy obsession with secrecy that works against every other form of diversity 2021-09-05 16:23:34 Something something capitalism and an NT dominated business world maybe 2021-09-05 16:22:19 If tech companies are started by so many ND folks why do they always converge toward NT norms over time? 2021-09-05 16:18:16 It was on potato bread with mango coleslaw. I want it 2021-09-05 16:15:27 I had a dream last night that Amazon opened a line of autonomous takeout restaurants with no human in the loop at all (except to eat), and I went to one with @CheeToS_ and got a really good BBQ pork sandwich, and I was like "wow this is ethically questionable but delicious" 2021-09-05 16:11:45 RT @blkgirllostkeys: “A lifetime of learning slowly that you can’t trust yourself or allow anyone else to depend on you leads to an overwhe… 2021-09-05 15:48:05 RT @decolonialatlas: Indigenous resistance has stopped or delayed greenhouse gas pollution equivalent to at least one-quarter of annual U.S… 2021-09-05 05:18:11 @KevinKaichuang Second is cool. "Sensitive topics" reviews always bug me because what is viewed as sensitive is a political decision that works against, for example, ethics researchers. Still, sounds better than conditions at a lot of other places 2021-09-05 05:13:43 @KevinKaichuang What about giving talks, do you have to get permission to give talks before publication? 2021-09-05 05:12:34 @KevinKaichuang Wow! Is there an internal review process at all or do you just publish? 2021-09-05 05:09:01 @msyvr I don't know, but if working not for a company I would just MIT license everything and not worry about intellectual "property" 2021-09-05 04:45:18 @kamatsu8 IDK but they know the funny CLIP robustness bugs 2021-09-05 04:41:13 For Halloween this year I am definitely showing up to work at UIUC as an "iPod" by just taping an "iPod" label to my forehead 2021-09-05 04:32:34 I guess I want to know if there is a research model for the future of industrial research collaborations with academics that destroys the four bad things but still works for industry people. Bonus if it captures more of the good things like resources for students 2021-09-05 04:29:45 (2) and (4) are, in my opinion, often in direct conflict with publication ethics codes like the ACM code. (1) is just unpleasant and bad for science. (3) is just not fun for anyone 2021-09-05 04:28:24 Things that are common but huge negatives from an academic side: - restrictions on speech about work, especially before publication - passes on papers by lawyers - inadequate respect from non-research executives - tendency not to attribute individual contributions (who did what) 2021-09-05 04:24:58 Things that really don't matter: - income 2021-09-05 04:23:59 The ideal from the academic side: - openly talk about work from day one - appropriate credit - collaboration with industry experts - practical impact - tenure letters etc. - resources from industry like compute - opportunities or funding for students - respect - no censorship 2021-09-05 04:18:24 RT @athundt: Apparently hiring algorithms are keeping candidates from ever being reviewed by a human if their cases are like the two year g… 2021-09-05 03:58:13 @SMT_Solvers Ethically? With DARPA? I have a lot of money from them. The ethical issue is that my verification, while it can be used for a *lot* of good, will also be used by the US military in a way that could harm people. Dual use 2021-09-05 03:56:30 I feel like researchers should start negotiating with companies to be able to share everything they do openly from day one, unless for some reason people think research secrecy is remotely reasonable 2021-09-05 03:55:08 I think I'm still very confused, as I have been for my entire life, why we accept that companies guard secrets so intensely, when nobody dies or gets hurt if the secrets get out, worst case some rich people lose money or whatever 2021-09-05 03:54:07 But there I guess there isn't an expert you're looking to work closely with, so it's largely you doing your own thing and DARPA using it for what they want. In industry you're working directly with those researchers and they know internal information or something 2021-09-05 03:51:35 (Not talking ethics here, obviously, since I'm using DARPA as an example. But thinking about a model that largely works for academics, when there is a contractor on the team to help with deliverables, without clashing with openness from day one) 2021-09-05 03:50:13 It feels like the practical impact and collaboration with industry players should be possible for many projects by just pre-approving all information about the project as open, or by segmenting the non-open parts away neatly and leaving those to the industry folks, as with DARPA 2021-09-05 03:48:56 Unrestricted industry awards are nice, but also very small, and honestly the main benefit of industry collaboration is practical impact 2021-09-05 03:47:13 It confuses me that DARPA projects can be all F/OSS and involving constant open presentation of all work involved despite it being in service of the military, but industry projects are often secret until publication, and the publication must go through approval 2021-09-05 02:59:37 RT @shapoco: https://t.co/cDED8Q2iyM 2021-09-05 02:56:32 For most of my adult life I thought "prima donna" was "pre-Madonna" and I didn't understand what everyone had against Madonna 2021-09-05 01:25:12 @AlejandroPiad @kamatsu8 I keep sending my thesis to students because it really did its job well 2021-09-05 01:23:22 @johnregehr I feel like a lot of things I would want guaranteed would be weird in a contract, too, like appropriate credit for my contributions 2021-09-05 01:19:13 @johnregehr At any point, like even before publication? Does that work? 2021-09-05 01:18:19 RT @AlejandroPiad: I got interested in dependent type theory for a project and started reading @TaliaRinger's thesis which has a nice conci… 2021-09-05 01:17:57 @AlejandroPiad Thanks!! I had a lot of fun with it 2021-09-05 01:16:45 I really want heavy industry-academic collaboration for the interactive theorem proving community, but I want it in a way that strongly respects the academics involved, and doesn't default to clashing norms 2021-09-05 01:15:27 (I guess in ML it's common to keep your work secret before it's out, but I think this is a consequence of heavy industry influence on the field, and I also think it's a harmful norm) 2021-09-05 01:14:26 Lately I've been wondering about ways for academics to collaborate with industry that don't involve dramatically clashing norms 2021-09-05 01:13:26 Or where lawyers and so on never do a pass on your papers? Amazon didn't have much of a process when I was there, but I still did have a lawyer pass that did impact wording in a few places. Also, licensing the software was a huge pain 2021-09-05 01:11:53 Just a question out of curiosity: are there industrial research groups that let you talk openly about your work before publication as often as you want? 2021-09-05 01:07:04 @wilbowma @brendanzab @koronkebitch For a type theory paper at POPL? All Bs is impressive haha 2021-09-05 00:40:21 @AlejandroPiad I don't really think this is an excuse. If you accept a PC, you should put the time in to write really kind and constructive and helpful reviews, otherwise don't accept the responsibility 2021-09-05 00:36:52 @EremondiJoey @wilbowma @samth @krismicinski Wow 2021-09-05 00:35:17 @AlejandroPiad No, being kind in reviews is a responsibility you have as someone in a position of power. I have no sympathy for mean reviewers 2021-09-05 00:26:47 @wilbowma @samth @krismicinski omg what is happening to the world 2021-09-05 00:24:07 @samth @krismicinski It's like I set off all of their imposter alarms haha 2021-09-05 00:23:27 @samth @krismicinski I think the opposite direction of this may be why I struggle so much with POPL. POPL doesn't really care if you're an applied type theorist, they still want you to do all of the things real type theorists do, and the application stuff is just cute extra work you did 2021-09-05 00:20:06 @samth @krismicinski Do you have an example of that? 2021-09-05 00:16:43 @krismicinski @samth I think the more POPLy your work is, the harder POPL is to get into, and similarly for PLDIy work at PLDI. But take POPLy work and signal its value to PLDI folks, or similarly for PLDIy work at POPL, and you've got a better shot hahahaha 2021-09-05 00:11:18 @krismicinski I'm only a type theorist when I submit to PLDI or OOPSLA. Otherwise I pray for no type theorists to review my work 2021-09-05 00:08:02 @krismicinski I think it depends on the work. It's extremely competitive for anything type theory related for sure. The standards are almost impossibly high. I think in "less POPLy" areas it is sometimes easier as long as the PC has enough representation of your topic 2021-09-05 00:05:29 RT @vj_chidambaram: Okay, another question via DM: how to decide between academia versus industry job versus post-doc? A There are a nu… 2021-09-04 23:54:38 @jxxf POPL, PLDI, ICFP, SPLASH/OOPSLA. There are many, many smaller conferences in smaller areas, and also some specific to, say, Europe, but those are the "big 4" international conferences 2021-09-04 23:50:25 @mjskay Honestly talking to them about ways to make the system less broken is a totally valid thing to do, helps a lot of people process feelings of rejection in a constructive way 2021-09-04 23:48:12 All of that is to say, if you seem someone ranting today about how the whole system is broken, or about how mean Reviewer 2 is, or how they think this is all hopeless, they're not being sore losers. Probably they just need to process really intense feelings. Help them do that 2021-09-04 23:47:12 These are normal reactions. People with ADHD or autism tend to have much more extreme versions of the rejection feelings, and much more intense and impulsive reactions. Computer Science research includes a lot of us with ADHD or autism. 2021-09-04 23:45:18 Also on peer review, I will say this while I have credence since I have no papers under submission. Some very normal ways people process rejection: - upward: complain about how broken the system is - outward: get angry at the person who rejected them - inward: get really sad 2021-09-04 23:38:11 Still, sometimes mean people say useful things, so it's worth getting the anger out IMO in a healthy way like private writing or ranting to a friend or mentor who is willing to listen, and then later digging into the useful advice embedded in those comments 2021-09-04 23:36:48 Also if any of your reviewers were just outright mean, that's their fault, not yours. No excuse for being mean 2021-09-04 23:34:45 I recommend opening up a document and writing all of the things you wish you could say to your reviewers, and then deleting the document and never saying those things, and instead writing real rebuttals a day or so later 2021-09-04 23:33:30 But yeah I've never had luck with POPL. So if your reviews are bad you're in good company. (I didn't submit anything this year, kind of in a lull between jobs) 2021-09-04 23:31:52 PLDI reviews were really good. Some of the best reviews I'd gotten. At the least, the PL conferences are really, really getting rid of some of the arbitrary community lines between them. So all four big PL conferences are probably options for you regardless of your work 2021-09-04 23:30:21 It may have impacted my search, I don't know. But I did post my paper on arXiv, then got lots of feedback that was really helpful. I still discussed it in my job talk and said it was under submission. I also thought PLDI would not ever work until my mentor convinced me to try 2021-09-04 23:25:58 The timing really sucks because you don't get PLDI acceptances before the job search. I was panicking about this last year, especially with the pandemic/recession 2021-09-04 23:24:26 I was confused by all of the depressed Tweets from programming languages folks today, and then I realized it is because POPL reviews came in. Both of my POPL submissions in grad school were rejected FWIW 2021-09-04 22:17:27 Most important for this will be a couch, but my admin has mysteriously informed me that procuring couches is hard, so it seems it will eventually happen but not any time soon 2021-09-04 22:16:47 So when someone wanders in kind of nervous and doesn't know what to say, they can pick any one of the things to talk about, just like people comment on my boba board. Maybe learn about common hobbies or cultural heritage or places we have both been 2021-09-04 22:15:33 Like, photo of me skydiving. Childhood photo with Saba. Some photos in Russia and Israel. My collage poems. My favorite stuffed animals. Some books. Photos of me at a ropes course with UCSD PL folks, including @polikarn 2021-09-04 22:14:03 I've spent the whole day unpacking now that all of my things have arrived. I still have a long way to go. One thing I'm excited about, though, is moving a bunch of conversation pieces into my office to reduce tension for meetings with students 2021-09-04 22:12:19 I always wonder what is going on when people suddenly like my really old Tweets 2021-09-04 22:05:38 @kamatsu8 Yeah there are going to be robustness attacks for this kind of thing, like a line on your resume invisible to humans that the AI picks up on and uses to sort you as extremely qualified 2021-09-04 21:10:50 @roeschinc I need to understand the framework since the dialect we are targeting is custom/new 2021-09-04 18:11:47 RT @mark_riedl: Sometimes GPT gets a bit over-excited and starts to tell a story instead https://t.co/xeTDlaoC1e 2021-09-04 18:09:37 Friend has told me lots of animals have learned human words and utterances, but not yet sentences with structure (like word order in English or cases in Russian) 2021-09-04 17:51:05 Seems like people try it with cats but cats are just always mad hahahaha https://t.co/OYQYLXrRQs 2021-09-04 17:46:43 So dogs are very social and human-oriented and probably want to talk to us more than most animals do. But I'm curious what animals we've tried to teach different kinds of language from a young age. I know about some chimp studies, Alex the African Grey, and many dogs now https://t.co/7y1U7uEZWi 2021-09-04 17:42:28 @hillelogram I don't know how you can answer your question but I'd love someone to look into it. They are very smart and can solve complex puzzles 2021-09-04 17:41:56 @hillelogram Brains of babies are really plastic and get less plastic as people grow. Seems puppies are probably similar. I wouldn't be surprised if that were the norm 2021-09-04 17:11:55 @hillelogram Yeah I think like most language it's just easier when the dog is a pup 2021-09-04 16:58:07 @hillelogram How old was the dog 2021-09-04 16:52:58 @hillelogram Yeah she understands and can use her own name and the names of people. Ergonomics is an interesting question, maybe she'd talk more if it were easier 2021-09-04 16:39:13 @b_vasilescu @ICSEconf I am horrified that anyone writes single page latex papers 2021-09-04 16:24:40 RT @ngfl3333: 4 eyes. The 2 little black eyes of a Tongue Parasite stare back from this poor Clownfish's mouth. This nasty Isopod will at… 2021-09-04 16:14:31 @BowToChris I've definitely always respected what feels right over what any authority figure tells me 2021-09-04 16:08:36 I'd rather have a 15 year old child of an employee looking at my resume and putting it into a pile than any automated system 2021-09-04 16:07:00 He told me, and then I didn't literally do what he asked, because that felt wrong. I used the meaning to influence which pile I put each resume in 2021-09-04 16:06:15 When I was 15, my mom let me work at her office for her nonprofit. Her boss asked me to sort through resumes, putting all the ones that say "HR" or "Human Resources" in one pile, and all the ones that don't in another pile. So I asked him what "HR" and "Human Resources" mean 2021-09-04 15:57:04 @unofiatu @jack_turban @CurlyClareToo Congrats you're probably agender, now stop making life hell for people who have a gender identity 2021-09-04 15:52:39 RT @lexi_lambda: spending literal weeks of one’s life constructing elaborate systems to avoid having to do a mildly repetitive two hour lon… 2021-09-04 06:03:36 Just tape a sticker on your forehead that says "the best candidate" https://t.co/heLpbpLqNu 2021-09-04 05:59:03 RT @cbarrett: Good evening to she/theys, hey/theys, and any pronouns people only 2021-09-04 05:58:58 @cbarrett Good evening!!! 2021-09-04 05:49:49 @unofiatu @jack_turban @CurlyClareToo Humans are weird and really like forming discrete categories of people because it helps people make sense of the world, so to cope with complexity sometimes they erase the nuances that contradict the existence of those categories 2021-09-04 05:47:48 @unofiatu @jack_turban @CurlyClareToo "Woman" and "man" and so on refer to a really large group of things, like: - chromosomes - gender identity - presentation - secondary sex characteristics that in most people seem to all align, but in some people they don't. So language evolves to capture that 2021-09-04 05:43:05 @unofiatu @jack_turban @CurlyClareToo Your gender is whatever you genuinely identify with. That's all. If you don't feel any sense of gender identity at all, you may be agender, like me. A lot of people do identify with a gender very strongly, which is why they do things that would be miserable for us 2021-09-04 05:34:06 @unofiatu @jack_turban @CurlyClareToo Ignoring my argument and saying "no" really loudly doesn't make you correct 2021-09-04 05:33:27 @unofiatu @jack_turban @CurlyClareToo Instead of promoting selection bias, read the article that actually cites and summarizes the relevant studies. Then read the relevant studies to understand the data and how it was gathered and what it does (and doesn't) imply 2021-09-04 05:30:37 @unofiatu @jack_turban @CurlyClareToo It isn't about stereotypes. I don't feel any sense of gender but it's not hard for me to understand that a lot of people do, otherwise they wouldn't bother to go through this much work. Vi Hart said it best: https://t.co/22sbMEKdBD 2021-09-04 05:27:19 @unofiatu @jack_turban @CurlyClareToo https://t.co/qCmZDmJQQd 2021-09-04 05:20:46 @hillelogram I wanted to show you this too 2021-09-04 05:16:05 https://t.co/dJnj4pvm5b 2021-09-04 05:09:34 https://t.co/FpVZGVRjd1 2021-09-04 05:08:42 https://t.co/bv9lkCNyIA 2021-09-04 05:05:33 I think about this dog basically every day https://t.co/HTIEwgJokL 2021-09-04 05:02:07 RT @ariadneconill: @ebelliveau malloc() was replaced in C38 with a routine that just plays music on the PC speaker 2021-09-04 04:51:14 RT @chick_in_kiev: sorry pigs arent kosher 2021-09-04 04:47:06 no you're eating cereal at midnight 2021-09-04 04:24:17 Ah, the strength of opinion poll is interesting so far!! 2021-09-04 04:15:41 Another consistent theory is that I just spend enough time with ND folks that lines on skin from clothes become a topic of conversation 2021-09-04 04:11:21 If you voted in the poll above, how strongly do you feel about your choice, and are you neurodivergent? 2021-09-04 04:09:50 @silvascientist Meanwhile most NT folk voting like these they just never talk about it because they probably care about being considered weird 2021-09-04 04:09:25 @silvascientist Oh I know I just mean that I've only ever heard ND people talk about this 2021-09-04 04:07:34 @hillelogram https://t.co/P7XkcuMjdA 2021-09-04 04:05:53 OK new theory, neurodivergent people are just less afraid of openly stating our opinions on the lines on our skin that clothes form by compressing the fluid in our cells 2021-09-04 04:03:38 Wow wow wow a lot of NT folks like them that's cool. Maybe this is just a thing a lot of humans weirdly like but don't talk about because it's weird. @hillelogram you should vote though since I can't 2021-09-04 04:00:13 @silvascientist Same same same but now I can't figure out the query I used to find the information and it is lost forever 2021-09-04 03:59:41 Do you like them and are you neurodivergent 2021-09-04 03:53:50 @silvascientist Oh fascinating. I like to run my fingers over them, they're like free fidget toys 2021-09-04 03:51:32 @silvascientist They're so cool you can form arbitrary bounded patterns in your skin 2021-09-04 03:48:56 I met one other person who likes these a lot but we are both very weird and very neurodivergent 2021-09-04 03:48:26 Does anyone else really like the lines on your skin that clothes form by pushing against your skin and compressing the fluid in your cells 2021-09-04 03:29:13 This is the first time in my life I've moved into a house that actually has space, and it's incredible 2021-09-04 02:22:07 @unofiatu @jack_turban @CurlyClareToo There's a reason why even if you view being trans from a medical lens (which is kind of silly but necessary in the US for insurance reasons) the path forward is literally just to give people the tools to embrace their gender 2021-09-04 02:20:04 @unofiatu @jack_turban @CurlyClareToo I've had eating disorders too, but I'm not a transphobe. One of many differences: listening to your eating disorder voice does serious physical and emotional damage to you that is often fatal, while transitioning improves health outcomes dramatically for trans folk 2021-09-04 00:57:52 @kamatsu8 @Tobias_Grosser Thanks! Indeed it would be good to talk to gauge overlap and reusability of existing infrastructure (my work will mostly be on automation/proof reuse though) 2021-09-04 00:20:26 @dlowd Yes we get to never stop learning!!! 2021-09-04 00:08:41 @johnregehr Verified compiler best compiler (Yang, Chen, Eide, and Regehr 2011) 2021-09-04 00:05:26 @JokerEph @mihaimaruseac @charith_mendis Wonderful, thank you! 2021-09-04 00:00:55 I think the coolest thing that I'm excited for is that students add chaos to your research ideas. Like it's no longer think of idea, refine, execute, refine, communicate and so on. It's think of idea, present idea, CHAOS ENSUES, NEW IDEAS FORM 2021-09-03 23:58:33 @mihaimaruseac @JokerEph @charith_mendis Oh wonderful!!! Thank you!!! 2021-09-03 23:57:57 @johnregehr Verified compilers, John!! Verified compilers! 2021-09-03 23:56:54 @mihaimaruseac @JokerEph Thanks!!! I also found @charith_mendis who sits down the hallway from me. But a toy dialect tutorial sounds absolutely wonderful and would really help me get a sense of some of the new things I need to know for this project. Also gives me something concrete for students 2021-09-03 23:55:24 weeeeeeeeeeeeeeeeeeeeeee 2021-09-03 22:47:33 how do adults sit 2021-09-03 22:47:25 I cannot sit still right now omggggggggg 2021-09-03 22:44:32 @hillelogram You should use a photo of me as the initial image 2021-09-03 22:44:02 @hillelogram Did you just pass my Tweet to VQGAN+CLIP? This is next level trolling 2021-09-03 22:38:38 ahhhhhh I can do this professor thing!!!! 2021-09-03 22:38:25 Like it exists and the relevant experts are all around me and there's student interest and it makes sense and it's going to be useful and it's going to be fun and interesting too 2021-09-03 22:37:59 What about a project that touches literally all of proof reuse, proof automation, formal verification, lightweight formal methods, compilers, machine learning, programming language design, and proof engineering more broadly? One that's totally tractable 2021-09-03 22:36:08 I like computers 2021-09-03 22:32:28 I am a ball of energy 2021-09-03 22:32:14 I love this 2021-09-03 22:29:32 OKAY I'M EXCITED NOW one project is taking shape and it is great and there are so many good people around to do the thing with 2021-09-03 21:36:01 What are the best resources for understanding MLIR at a level strong enough to build compilers that target a dialect of it? Anyone have some F/OSS compilers I can play with that target MLIR and are already implemented? I learn more by playing than by reading 2021-09-03 21:13:32 I kind of wonder what the reason this breaks tension is. My theory is that humanizing people who are more famous than you are makes you seem more human too, and that saying good things about people makes you seem more approachable 2021-09-03 21:03:09 And that's nice because it turns out tracking social knowledge like facts about people and their positive traits is a strength I have that's probably even stronger than my expertise in my actual field so I can pretty much do that for any arbitrary person 2021-09-03 21:00:19 Another thing that breaks tension is random social knowledge, not of the gossipy variety, just like positive statements about researchers you know that are relevant to the conversation ("he has contagious enthusiasm, super fun to work with" "he was actually a compilers guy") 2021-09-03 20:56:51 The students probably don't know but I feel nervous too haha 2021-09-03 20:56:14 So far boba ranking board is great for breaking tension when nervous students come to talk to me for the first time. Going to add collage poems to decorate office, and somehow eventually procure a couch (my admin has told me this is challenging for reasons but possible) 2021-09-03 20:54:13 @zornsllama My external mentor told me that as junior faculty it makes sense to spend some time working out concrete details for your students at first, programming a bit, exploring as if you're still a student (just not first author) and I think I'll do that 2021-09-03 20:53:10 @zornsllama Ah, interesting! Yeah, that's how I feel now, but I never felt that way as a student for some reason, I think because usually by the time I had an idea I was already impulsively trying to make it happen 2021-09-03 20:50:45 I think what's happening is that as a student, I never worried that a project wouldn't take shape and turn into something very concrete. As junior faculty, though, the stakes are higher, because if that happens it hurts a student who trusts me 2021-09-03 20:32:27 I'm sure I'll figure it out, and maybe the students are even really having fun right now, and I'm guessing these feelings are very normal 2021-09-03 20:31:45 It's fascinating to me that as a student, the beginning of a project was always my favorite part, since I really got to explore ideas and there was so much freedom. As junior faculty, the beginning of a project makes me anxious, since I worry about not helping the student enough 2021-09-03 20:29:23 Several of the projects will likely have students collaborating with each other, which is fun, and I hope a positive experience for everyone 2021-09-03 20:28:45 Looks like for year one I'm going to start with four projects, two of which will probably not turn into long-term commitments since they are with senior graduate students. I don't know my parallel capacity yet, but I'm unlikely to push it past four this year 2021-09-03 19:54:41 https://t.co/oBLM2WcRcc 2021-09-03 18:51:01 Recently I've been dedicating blocks of time on my calendar to different students. That's cool but now I need a way to remember which ones are actual meetings, since my old naming convention doesn't work anymore 2021-09-03 18:45:32 RT @GirlsWhoCode: You can’t be what you can’t see. One of the best ways to start closing the gender gap in tech? Giving students the resour… 2021-09-03 18:22:26 @RandomlyWalking It's funny because I only have a few that are old math books I kept around because I liked them or something similar. Most of it is leisurely reading, and my heaviest books are actually about physiology and competitive running 2021-09-03 18:17:37 Mover 1: man these boxes of books are HEAVY Mover 2: she's actually a professor so that makes sense Mover 1: wow I didn't know that 2021-09-03 16:06:24 Lmao the "see also" in that article https://t.co/GZn2C9gcsi 2021-09-03 16:04:58 The absent-minded professor is just a stereotype of neurodivergence change my mind https://t.co/1JohcKl2Qg 2021-09-03 16:03:55 @elfprince13 Good call 2021-09-03 16:03:47 RT @elfprince13: @TaliaRinger orientation for my summer internship at PPPL included a warning not to hit absent minded physicists in the pa… 2021-09-03 15:43:42 I'm pretty much always walking around with my head in the clouds thinking about something computer science related, and it's really nice to now have a job title where walking around totally unaware of my surroundings just makes me a stereotype and not a weirdo 2021-09-03 14:53:14 RT @Illinois_Alma: The @iSchoolUI's Center for Informatics Research in Science and Scholarship will highlight responsible data science and… 2021-09-03 13:55:27 RT @zbtkna: питон-программист @ya_spatb https://t.co/wgTPgnrquR 2021-09-03 13:50:10 @EliSennesh Oh, a particular breed of pitbull? 2021-09-03 13:49:22 @EliSennesh ? 2021-09-03 13:44:31 A recent podcast I did with @Dev_Pods, which gets into my long-term vision for the future: https://t.co/op90BxlMEd 2021-09-03 13:43:20 Some of my ideas: https://t.co/ZTbc7QtEjM 2021-09-03 13:43:01 I am also interested in students who want to integrate formal proofs into interactive tools built on code language models like Codex, in order to provide strong correctness guarantees about generated code. 2021-09-03 13:41:00 We'll likely get along if you like proof assistants, want to build a world of secure and robust systems, and are enthusiastic about at least one of: - proof automation, - dependent type theory, - usability of proof assistants, or - bridging proofs and tests. 2021-09-03 13:35:45 If you're looking at grad schools for computer science, and you like proof assistants and want to bring formal proof to everyone, super please consider Illinois! I'll be taking students 2021-09-03 13:31:54 RT @TimMLatimer: This is a really big deal, perhaps the most ambitious clean energy target by any government to date. And more good news fo… 2021-09-03 13:27:00 Odin is kind, sweet, loving, and playful 2021-09-03 13:26:19 Oh, this is Odin, my housemate's doggy. He is 1 year old as of last weekend. I love him https://t.co/3A3BgRaGhJ 2021-09-03 13:17:06 RT @matthew_d_green: This looks promising. https://t.co/LBQTpKSwRl 2021-09-03 13:10:16 @mmitchell_ai @lizthegrey @tegan_maharaj @ACM_Ethics In that sense, the code also is good to help you with internal protest, by way of giving your moves legitimacy to some authority figures 2021-09-03 13:09:12 @mmitchell_ai @lizthegrey @tegan_maharaj @ACM_Ethics And that can actually be a really, really huge realization for people who generally believe in rules and authority. 2021-09-03 13:08:21 @mmitchell_ai @lizthegrey @tegan_maharaj @ACM_Ethics In my experience, also, people who make it high up at companies have lawful alignment. So "not listening to authority is the ethical move" often has little weight to them on its own, but presenting the code changes that to "an authority you are bound by that says this is ethical" 2021-09-03 13:05:03 @mmitchell_ai @lizthegrey @tegan_maharaj @ACM_Ethics If internal attempts fail and so on 2021-09-03 13:04:31 @mmitchell_ai @lizthegrey @tegan_maharaj Also if your company does this to you, if interested, one possibility is bringing it up to @ACM_Ethics to see if there is a path for enforcement that may lead to improvement of company conditions going forward 2021-09-03 12:52:23 You know how social media can segment you into a political bubble? It can do the same for your research interests. So if you care about having a broad view of the area, it's important to follow a broad group of researchers 2021-09-03 12:43:06 @nate_yazdani @bedrocksystems Oh I'm glad you ended up at Bedrock! I hope it's going well! 2021-09-03 12:42:35 RT @ThatRabbiCohen: Well if this doesn't sum up where almost everyone I know is right now.... https://t.co/Z0KDIvHflv 2021-09-03 12:37:50 RT @eszter: important piece by the fantastic @eredmil1 on the problems w/@Apple’s child safety tech https://t.co/3D3EsZ96nx 2021-09-03 12:30:53 I know this is a real photo, but it's funnier to interpret it as taking place in 2021 San Francisco https://t.co/mfwZm3m6j6 2021-09-03 12:00:11 RT @Abebab: trying to watch a talk on AGI (by a collection of white men whose view is as diverse as they are) and i'm struggling to take an… 2021-09-03 11:47:28 RT @pinceana16: 235/365 Fish-shaped pancake designed by me () #365origamichallenge #origami # https://t.co/QS… 2021-09-03 11:37:56 @BetaZiliani 3, I always check the ICFP and OOPSLA boxes. But yeah, I've never been happy with the omission of my entire research specialty, especially given that CAV counts and is very comparable to ITP/CPP 2021-09-03 11:33:30 @certifiablyrand Oof 2021-09-03 04:29:27 One day I will get to remove the giant boot https://t.co/z3M9Gv8NEN 2021-09-03 04:22:56 I have purchased armor that protects against further programming injuries https://t.co/zC55l8cl0e 2021-09-03 04:21:08 @bll_krtl I also genuinely read all of the papers that cite me, still! Though I don't know how well that habit scales 2021-09-03 04:12:10 I finally have more citations than I do AAA PFCs in DDR 2021-09-03 04:09:19 RT @tiddypills: i’m speechless lmao https://t.co/A2kK42OJMg 2021-09-03 03:01:21 RT @carlzimmer: As Ophiocordyceps switched into domesticated partners, they may have used some of these "manipulation genes" to establish a… 2021-09-03 03:00:49 RT @carlzimmer: Cicadas and other insects have swapped symbionts over evolutionary time, with new microbes taking over from older ones. In… 2021-09-03 03:00:23 RT @carlzimmer: Cicadas can only survive thanks to special microbes that live in some of their cells. The insects feed on xylem, a fluid fr… 2021-09-03 03:00:20 RT @carlzimmer: In a recent preprint, @mcsymbiont & 2021-09-03 02:59:51 RT @carlzimmer: I didn't think these fungi, which turn insects into zombies, could be any more remarkable. Turns out I was wrong. Some have… 2021-09-03 02:26:05 Also, I know it's just a number, but it still feels good to know people are building off of my work and giving me credit, which is really what I care about! Credit is the best reward 2021-09-03 02:25:09 (Programming languages citation counts look absolutely nothing like machine learning ones, for those of you who are used to those.) 2021-09-03 02:24:02 Oh also I surpassed 100 citations of my work today, that was cool! And I added myself as faculty to CSRankings so I can watch number go brrr 2021-09-03 02:21:57 I had a dream last night that Quanta interviewed me about proof assistants, but I forgot to mention Lean and I felt really guilty and contacted them about this. This is really hilarious given the context of every Quanta article about proof assistants caring only about Lean 2021-09-03 02:08:35 RT @mmitchell_ai: This is important if you're a publishing researcher. Your connection to IEEE and ACM supersedes your connection to any pa… 2021-09-03 02:05:34 I'm so excited for OOPSLA so I can see @ChandrakanaNaN. Miss you so much already 2021-09-03 01:56:27 Please come visit me sometime 2021-09-03 01:52:10 I miss all of my Seattle friends a lot 2021-09-03 01:08:55 @MDDBACA Looks cool. Looking forward to the talk! 2021-09-03 00:54:09 RT @MDDBACA: Our paper with the PROSE team at Microsoft (@ProseMsft) addressing these questions is accepted at OOPSLA’21! https://t.co/mc20… 2021-09-03 00:54:06 RT @MDDBACA: When it comes to the automated generation of programs from natural language, we know that GPT-3 (the state-of-the-art language… 2021-09-03 00:47:01 RT @SigplanM: Waiting for a rematch? Good news: we've just matched you! Our rematch list is now empty. But, we're a bit tired right now, so… 2021-09-02 23:57:28 Could even just be a discussion relevant to the project framed by the code 2021-09-02 23:56:47 Am I allowed to do this because it's a fantastic idea https://t.co/q6FOyEIaSU 2021-09-02 23:49:58 @hillelogram omg 2021-09-02 23:32:18 Going to start all of my classes with a quick review of the ACM Ethics Code so you will all have a generation of students who know it front and back and are ready to apply it at work tbh 2021-09-02 23:30:30 I will shill this code for life it's the coolest thing about the ACM https://t.co/ehpS8NAn8Z 2021-09-02 23:28:25 @lizthegrey I think the ACM and IEEE codes are always a nice way when you're really absorbed in a situation at your workplace to back up and get a sense of, more generally, what it is professional and ethical to consider or do as a computer scientist 2021-09-02 23:26:49 @lizthegrey The ACM code is amazing, it actually encourages you to first question unethical rules within your workplace, then later outright defy orders at your workplace if internal questioning fails and if you are aware of the consequences of doing so 2021-09-02 23:24:02 RT @lizthegrey: see: https://t.co/FOFg7EG6EN. 2021-09-02 23:17:25 @ErickGalinkin Oh man I have an email from you I've been snoozing for weeks hahahaha sorry about that 2021-09-02 21:56:01 I'm just in disbelief that humans think it's OK to treat other humans like this literally over some company. Guys it's literally just a company 2021-09-02 21:51:06 Ugh, going to bury my head in a research paper before this makes me cry 2021-09-02 21:49:47 The toxicity in the tech industry is just completely out of control at this point. I don't remember it being this bad when I was an engineer. Maybe I just didn't expect to be respected back then 2021-09-02 21:48:40 I can think about the future of the field rather than some company, so everyone around me can be a friend and collaborator with shared goals to make the field better 2021-09-02 21:47:49 Like in academia I at least feel like I am now strongly respected for my work and do not have to deal with any more toxic power hierarchies. And I can talk openly about what I do, and largely not worry about not being credited 2021-09-02 21:42:37 I really want to collaborate with industry for my work, but the more I hear about industry these last couple of years, the more terrified I am of trying 2021-09-02 20:14:51 RT @CecileTonglet: We are doomed https://t.co/c0CFVUBqa7 2021-09-02 20:09:02 It looks cool, but also, there was literally a Black Mirror episode about this https://t.co/6NNZMZ5ggk 2021-09-02 19:44:06 @hillelogram I definitely added alt text to this so no idea why it didn't show up 2021-09-02 19:43:16 This reminds of me a suggested prompt by GPT-3 yesterday with @hillelogram: "triangles and triangles" https://t.co/I8n3sCjvFb 2021-09-02 19:36:15 @AleResnik I think you need to meet more people 2021-09-02 19:32:58 Broke: Talk to domain experts because we have useful knowledge that will help you Woke: Talk to domain experts because we're hilarious 2021-09-02 19:29:04 God bless the programming languages community https://t.co/47LETBST9v 2021-09-02 18:24:50 @16kbps weird I also ordered empanadas while watching the trial 2021-09-02 18:23:12 Honestly even more glad I'm not a judge 2021-09-02 18:22:55 @EulerWept I think it's really good to have a conversation with someone in your field, either a PhD student or a faculty member, who can give you advice based on your individual situation. Twitter is one good place to find people like that to talk to---depends what you're planning to study 2021-09-02 18:17:14 My reaction to watching one hour of the trial 2021-09-02 18:16:40 I'm so glad I'm not a lawyer 2021-09-02 16:37:01 @moyix @johnregehr @github https://t.co/Fi3v76rYm4 2021-09-02 16:36:13 @moyix @johnregehr @github It works incredibly well when using it doesn't require inverting a function. Some experiments here toward the end of the paper: https://t.co/uQ0klIrgJ0 2021-09-02 16:32:53 @moyix @johnregehr @github This sounds like so much fun 2021-09-02 16:29:03 @moyix @johnregehr @github Oh it's cool they were just about tradeoffs of "distinct?" versus iterative solving 2021-09-02 16:28:20 @moyix @johnregehr @github Ah gotcha. Probably more efficient but regex integration with strings is really fun, and with strings I wonder if you could even possibly somehow build GPT-2 ranking into the loop, which could be really fun 2021-09-02 16:10:41 Anyways, everyone should be open-minded 2021-09-02 16:10:03 (This is in contrast with natural language) 2021-09-02 16:09:30 Beyond that, the more I think about it, the less I'm convinced that in programming languages there is really a distinction between syntax and semantics. Semantics is just what is not yet explicit in the syntax, but it could be. Type theory often refies semantics as syntax 2021-09-02 16:08:35 (If we *also* use the semantic information we are already using) 2021-09-02 16:08:17 I want automation that can use all or at least most of this information. I think it's possible. I think machine learning will be a part of it, though definitely not all of it. We will get gains already from just considering some syntax we ignore as is 2021-09-02 16:06:10 Beyond that, we use our outside knowledge, of course. Like paradigms we've read about for writing certain proofs 2021-09-02 16:05:11 Proof structure mirrors program structure, and the reason for this is set in stone mathematically, impossible to evade 2021-09-02 16:04:15 (Some may think they do, but this is a lie easily dispelled) 2021-09-02 16:03:48 But the structure often guides the proof, structure deep in the context. We'll print variables, interact with the environment, check types, evaluate functions and check their outputs, reduce terms. No good proof engineer truly uses tactics as a black box 2021-09-02 16:01:37 They also are good for explaining things to newcomers, and automation is its own kind of newcomer 2021-09-02 16:01:12 The best proof engineers use both. Syntax, like type and theorem and constructor names, give us quick ways of associating information with our outside knowledge that we have, or remembering what a thing does, roughly 2021-09-02 15:59:54 An exercise I always recommend for proof automation people is to just pick up a proof exercise somewhere, say, from Software Foundations. Write some proofs, but also really, really dissect your thought process as you go, everything you check, all information you use 2021-09-02 15:58:47 It isn't just that syntax is a proxy for semantics. It also includes semantic information that complements what is explicit in the structure of proofs themselves 2021-09-02 15:57:46 Programming languages people often say this derisively, and then we build beautiful deep automation that uses just semantics and not syntax. That's cool but like: https://t.co/pIglDrDcNW 2021-09-02 15:56:54 One thing I left out in the podcast about the role of machine learning in proof automation is that in addition to style, it is already it is great at syntax, at natural language integration, comments and variable names. This matters 2021-09-02 14:50:16 All of the AI Ethics work I've read about, alongside with my sister's focus on racial health disparities in cancer, has given me common academic ground to discuss with her and it's super super super cool 2021-09-02 14:41:29 RT @aTomBeer: How can we see it? For each predicted risk score, Blacks have more chronic illnesses compared to Whites. How did this bias ar… 2021-09-02 14:17:53 RT @aTomBeer: Ever seen claims of superhuman medical performance? It is likely because the model performs a proxy for the real medical task… 2021-09-02 14:17:36 RT @aTomBeer: This work by @az_jacobs @hannawallach is a great read on the topic https://t.co/LeCjV7fM9S 2021-09-02 14:16:11 This sort of thing is why you can't ever cleanly separate use of data from all of the human factors that led to the data existing to begin with 2021-09-02 14:14:33 This is actually an extremely direct way for any model with medical use to amplify racism. Black Americans are less likely to trust the medical system and seek care because of a history of exploitation and neglect that strongly justifies distrust (my current reading obsession) https://t.co/BOxm0aaGg7 2021-09-02 14:12:20 RT @aTomBeer: This doesn’t make sense. Why are radiologists not able to see pain the same for all people? Because they use a diagnostic sca… 2021-09-02 14:10:06 RT @aTomBeer: Let’s play Data Science Detective. You are given a risk score model for stroke. It is based on 200K visits in one hospital’s… 2021-09-02 14:02:34 I fundamentally respect this level of chaos regardless of details https://t.co/EM1aQ2XF2W 2021-09-02 12:47:19 RT @scotusreporter: Sotomayor: "Presented with an application to enjoin a flagrantly unconstitutional law engineered to prohibit women from… 2021-09-02 06:00:17 @moyix But you can still see an exponential slowdown when inverting a function and working over strings even with iterative solving (!) 2021-09-02 05:56:36 @moyix Not sure if it has changed, but distinct? back then was totally intractable when strings and functions got involved, and looping/iterative solving was much faster 2021-09-02 05:54:33 @moyix In any case, for this very valuable work, if you want to use less compute in the future, you may find the results from the experiment in this paper interesting: https://t.co/uQ0klIrgJ0 2021-09-02 05:52:54 @moyix Z3 struggles with generating many strings with functions they must satisfy because doing so basically requires inverting a function from what I remember 2021-09-02 05:50:47 @moyix Did you use distinct or loop? 2021-09-02 05:49:17 @moyix @johnregehr @github Or just solely string encodings? 2021-09-02 05:46:53 @moyix @johnregehr @github Does this use the Z3 theory of strings? If so, genuinely glad my early testing is contributing in some way to questionable and chaotic use 2021-09-02 05:42:09 RT @moyix: @johnregehr @github The important thing is that I got to play with Z3, CUDA, and GPT2 this week instead of preparing for my firs… 2021-09-02 05:30:09 @moyix Research questions I never thought I'd ask 2021-09-02 05:29:06 @moyix How does it handle alternative spellings like pisswacker? 2021-09-02 05:25:28 I'm not even sure this follows the TOS and I wouldn't advertise it myself but I'm also 100% here for this chaos conceptually 2021-09-02 05:22:35 Context https://t.co/PbQI4wSH0z 2021-09-02 05:19:13 Ah yes, the most important work (?) https://t.co/IdD9WaKCsc 2021-09-02 05:12:59 Is the rest of the east coast as bad as New York right now? Like is my family good? 2021-09-02 04:59:15 Holy shit https://t.co/BqqLDAikln 2021-09-02 04:37:43 Without downloading any new pics, where are you at mentally? https://t.co/tLtBeNZHQR https://t.co/qyXN42xYTm 2021-09-02 04:30:34 But it's possible. All of the puzzle pieces are there. It's just a matter of actually bringing them together 2021-09-02 04:29:35 Mostly, my vision is quite big. I think it requires open industry-academic collaboration and mutual exchange of ideas and resources, with credit and respect for academics so we keep engaging. I'm scared of pouring my heart and soul into that and failing 2021-09-02 04:28:08 I'm always so strongly convinced by my vision, but sometimes disillusioned by the small details I encounter along the way. So it helps me every so often to bubble up to my vision again. The podcast was good for that. Hopefully more grant applications will be, too 2021-09-02 04:11:19 @MPSellitto @baxterkb @mmitchell_ai @ruthstarkman It's not really an effective apology if you follow it by a complaint about being "bullied" FWIW 2021-09-02 04:01:00 Great day, great lake https://t.co/rVWeTQPo4t 2021-09-02 03:48:17 Many cool ideas for the future of proof engineering come up here https://t.co/i6KhuKHx80 2021-09-02 03:46:30 Great for students interested in working with me, too, if it sparks any ideas or excitement 2021-09-02 03:43:21 This podcast interview on the future of proof engineering & https://t.co/gsVfq69tuX 2021-09-02 01:15:45 @hillelogram triangles and triangles 2021-09-01 23:54:49 And that's your daily dose of prompt engineering using GPT-3 2021-09-01 23:54:15 We are a formidable team https://t.co/pjaQ8bDzRi 2021-09-01 23:49:09 I love this https://t.co/fmp6ZWFk9e 2021-09-01 23:47:00 https://t.co/sKWqSzrrpK 2021-09-01 23:46:28 https://t.co/PHDiyCOe8Q 2021-09-01 23:38:44 RT @Dev_Pods: This week on #DevDiscuss, we talk about proof engineering: a type of engineering concerned with technology that makes it easi… 2021-09-01 23:38:41 RT @ThePracticalDev: Each week on #DevDiscuss, we talk about the burning topics that impact all our lives as developers. Need some proof?… 2021-09-01 23:35:46 I love this one, the universe is exiting through a door in the universe 2021-09-01 23:35:29 Prompt iteration 4 https://t.co/N86cVOF2ZJ 2021-09-01 23:30:56 Prompt iteration 2 again and I'm into this https://t.co/llJ8Fz9Wmv 2021-09-01 23:24:11 Prompt iteration 1 https://t.co/Y8reqhgw8s 2021-09-01 23:20:50 (Actually 2, I just forgot) 2021-09-01 23:18:37 Still from prompt iteration 1 https://t.co/hxnkq61FYM 2021-09-01 23:16:59 Somebody stop me please https://t.co/Gn5ZRCIzyi 2021-09-01 23:15:13 I'm crying https://t.co/Q3wrbFutfx 2021-09-01 23:11:56 @hillelogram @theomn Definitely a dog. A dog with poise 2021-09-01 23:08:53 I should note his art AI calls out to VQGAN+CLIP like many others, he wrote the calling code 2021-09-01 23:04:17 Also from an earlier iteration of prompt generation https://t.co/N5ye4wdKqg 2021-09-01 23:03:34 https://t.co/zDhvmpOwKD 2021-09-01 22:56:14 This is my second attempt to generate prompts, first attempt got some amazing ones like "a mathematician's cathedral" which we are currently trying. Will share images when we get good ones 2021-09-01 22:54:57 @nndgnlzblty @hillelogram That's his human one 2021-09-01 22:54:47 Natural next step is to write a loop that calls out to a language model to generate a new prompt after n iterations, using your previous image as a seed for the next image, m times 2021-09-01 22:52:29 Current mood, with @hillelogram: asking GPT-3 to generate prompts for his AI art generator based on his prompt history https://t.co/OUIQXcpf2w 2021-09-01 17:18:28 The #DevDiscuss podcast on the future of proof engineering I did is now out! https://t.co/eSBEwRp9dp 2021-09-01 13:34:36 So yeah. Our exhaustion is not like yours 2021-09-01 13:33:08 And then it no longer becomes just an instance of an individual helping an individual, but rather part of a broad and clear pattern of unrewarded labor by women for men, by Black people for white people, and so on upwards in any power dynamic 2021-09-01 13:31:35 When you do it, usually a bunch of other women who have done the same thing approach you and warn you than they've also tried and it's tiring and unlikely to work. So you know there’s just this endless rotation of women willing to try this until it does work 2021-09-01 13:27:23 And I do mean for them and not just for us. Most often it is really mutually beneficial when it works out. But it's free, invisible, unrewarding, and sometimes really risky (if the man has direct power over you for example) 2021-09-01 13:21:42 I often wish that could go on my CV or something, but more often it's sensitive enough I can't say anything more than vaguely it's something I've spent hundreds of hours of my free time doing for a number of different men in power in different situations 2021-09-01 13:17:30 Ah, let us not forget the invisible labor of trying to help men in power be better toward the rest of us, which is the most invisible and exhausting labor that exists 2021-09-01 13:14:25 Our exhaustion is not like yours 2021-09-01 13:12:47 I want to see most tech bros last a week with everyone assuming they're incompetent by default, or just there to take notes at the meeting, and ignoring everything they say 2021-09-01 13:11:35 Comments split between the tech demographic majority saying things like "the workday never ends" and the rest of us like "toxic masculinity, discrimination, and abuse" https://t.co/fcXSESbkl3 2021-09-01 13:03:30 @mmitchell_ai @deliprao @luke_metro Yeah, discrimination and in many cases outright abuse 2021-09-01 12:51:17 @AlejandroPiad @sergeivolodinch Honestly Amazon could skip ads and profit off of household robots by just having the robot come to you and say "it looks like you're running low on shampoo, would you like to order more?" And when you say "why thank you, that would be great" your robot places an Amazon order 2021-09-01 12:46:52 @ncooper57 Putting the dishes away, putting clothes away, cleaning toilets and showers, cleaning the kitchen, taking out the trash, cooking simple meals, buying groceries, putting groceries away, putting dishes in the dishwasher after meals 2021-09-01 12:21:53 @kamatsu8 @p_droabreu0 @PTOOP "In the new programming, coincidence is replaced by consequence." This is poetry, it is beauty, not cleverness and wordplay or whatever 2021-09-01 12:20:11 @kamatsu8 @p_droabreu0 @PTOOP No it doesn't, and it's awful of you to imply that poetry is just that to me. Go away 2021-09-01 12:17:56 @kamatsu8 @p_droabreu0 @PTOOP Yeah it definitely does, you gave me a condescending and terrible summary of what you think I think poetry is 2021-09-01 12:16:19 @kamatsu8 @p_droabreu0 @PTOOP I'm literally a published poet 2021-09-01 12:12:58 @kamatsu8 @p_droabreu0 @PTOOP Conor McBride writes poems shaped like research papers 2021-09-01 12:11:04 @kamatsu8 @p_droabreu0 I am. Hence the beauty of @PTOOP papers 2021-09-01 05:03:21 I hate gender, I wish it would go away 2021-09-01 05:02:38 But when Alex opened the door his jaw just dropped and he said "you're a girl!" And I felt so, so bad in that moment, I can't explain it, but it was like there was this arbitrary wedge drawn between us. That was the last time I hung out with Alex and Leonard 2021-09-01 05:01:25 One day I went off to middle school, and Alex and Leonard didn't follow me. I started wearing lip gloss. I missed Alex and Leonard though. I wanted to wrestle and play Goldeneye 007. So all four of us got together 2021-09-01 04:59:57 The toy guns didn't fire bullets, they were cap guns, they smoked when you fired them. But we were young and imaginative so we made this entire cap gun capture the flag game on an honors system where the goal was to blow up the enemy base and get back without getting shot 2021-09-01 04:57:31 Days at Alex's house involved Eminem, buffalo wings and takeout pizzas, Perfect Dark, Conker's Bad Fur Day, Goldeneye 007, wrestling, and toy guns. This was my favorite thing as a child 2021-09-01 04:55:51 In elementary school my three closest friends for a long time were Steven, Alex, and Leonard. Alex and Leonard were both Russian Jews, and Alex's parents were very lax around us. Usually they let us do whatever we wanted, so I loved going to Alex's house https://t.co/QECQ9AJaHu 2021-09-01 04:37:52 It really broke my heart when gender became a thing and my childhood closest friends didn't want to hang out with me anymore because they viewed me as a girl 2021-09-01 04:34:46 Weirdly specific but like, @peztronic, @somelaurachick, is this not all of us 2021-09-01 04:34:00 But the girls still don't like you because they think you're weird so you compete in Dance Dance Revolution and make friends that way instead where everyone is weird and queer and neurodivergent 2021-09-01 04:32:35 And then at some point late in middle school the boys don't want to hang out with you anymore anyways because of something something gender which you don't even get or understand but then you miss out on the useful things they do later like programming and building computers 2021-09-01 04:31:09 Like enough people calling you a lesbian for liking Pokémon and implying that being a lesbian is a bad thing (turns out it's great) just makes it scary to keep hanging out with the boys and playing video games 2021-09-01 04:29:05 Honestly growing up a tomboy nerd sucked because at least the boy nerds got to be friends with other boy nerds, but I just knew I was weird and didn't have friends but also couldn't be too close friends with the boy nerds so I didn't even get to learn the cool things they learned 2021-09-01 04:26:14 Top 5 nerdiest sentences ever uttered https://t.co/vxWn5wxPi2 2021-09-01 04:22:52 I just wish the experience of interfacing with hardware were a bit smoother sometimes 2021-09-01 04:22:24 Anyways, I'll forever prefer the aesthetic of free Linux distros over the clunkware of commercial OSes 2021-09-01 04:19:50 Just didn't stay close with him at all, and didn't have any other friends willing to show me how to hack or guide me toward the things I was drawn to until college, so that's as far as I got. Linux and awe at my friend 2021-09-01 04:17:45 He built a computer in a box. It was so cool. I was captivated by the idea that you could just *do* that, like you could actually build a computer 2021-09-01 04:15:03 But I felt something, I guess, having to really understand what my computer was doing, and I like that, even though I didn't understand what I liked about it or where it would later lead 2021-09-01 04:13:46 But I always like really lightweight things so I went with Featherweight Linux and he taught me to dual boot and it was cool but not a good introduction to Linux 2021-09-01 04:12:03 The guy I had a crush on in middle school because he played Magic and was cute handed out Linux distros on disks at some point so I took him up on it 2021-09-01 04:10:18 I started using Linux in highschool before I knew what programming was. I just thought Windows was clunky and Macs were too expensive and I didn't like Apple as a company. I also felt like I wanted to be closer to my computer somehow, to really know what is going on in there https://t.co/s5JjXrOaDf 2021-09-01 04:02:04 @yihongz_bot Ominous 2021-09-01 04:01:45 RT @yihongz_bot: Note to myself: don't write an email when you are sleepy. https://t.co/RoKi5ewMMr 2021-09-01 03:37:35 @Artist_HB Just program it to get mad and refuse to work for free except in the household hahaha 2021-09-01 03:37:08 @Artist_HB Ugh I wish we could restrict them to household use only 2021-09-01 03:28:54 @Artist_HB I just want individual ones for home so I can stop doing chores 2021-09-01 03:25:19 RT @mpiedrav: @TaliaRinger «As Perera explained it, so long as English remains the gatekeeper to scientific discourse, shoehorning scientis… 2021-09-01 03:24:59 @mpiedrav Yeah I think I'd be fine with multilingual papers and an expectation that we all learn to read more languages honestly. I had to read French papers for my thesis work and it was fine 2021-09-01 03:21:00 RT @mavicrz: What's a pirate's favourite field? You may think it's ℝ but their true love be the ℂ 2021-09-01 03:15:19 I used to speak Japanese in high school but it has been a long time, and I definitely can't have research conversations in Japanese (I wish I could, that would be cool) 2021-09-01 03:14:30 If any of my friends are fluent in Japanese and want to summarize my research area that would be really cool https://t.co/EjvdwhSodS 2021-09-01 03:07:42 I follow @The_OCaml_Bot which retweets occurrences of OCaml, so I frequently see people all over the world talking about OCaml. Lots of common interests but our research isn't accessible to a lot of people this way 2021-09-01 03:03:47 It's weird that we print all of our papers in English and don't get them translated to other languages 2021-09-01 03:01:10 @gaxiiiiiiiiiiii https://t.co/evMqkaCZxG :) Sorry it is only in English though 2021-09-01 02:42:40 @gaxiiiiiiiiiiii @The_OCaml_Bot Likely yes, you should be able to write ML code if you've written OCaml 2021-09-01 02:23:27 Why do we still not have robots who can do all of our basic household chores 2021-09-01 02:14:39 Is adulthood just a bunch of people pretending because if so I abstain 2021-09-01 02:14:03 I'm a 31 year old professor why do I still want cereal in chocolate milk for dinner 2021-09-01 01:32:12 @dinabass Haha OK just making sure. Yeah, he is 2021-09-01 01:24:51 @dinabass It's nothing noteworthy. Just amusing 2021-09-01 01:22:02 @dinabass But this doesn't happen on my personal account where I don't have any forwarding set up 2021-09-01 01:21:43 @dinabass Yeah, but this has to do with email forwarding from my university address. For example anyone from Google also gets marked as spam by Gmail 2021-09-01 01:13:45 RT @retronouns: Discontinuing a trans person's HRT because they're overdue for an annual appointment rather than just emailing them a remin… 2021-09-01 01:12:45 @emilymbender His last email was "OK. L." So he wasn't doing the spam filter any favors 2021-09-01 01:09:07 @rao2z @ylecun @tdietterich @_joaogui1 @NeurIPSConf @icmlconf @iclr_conf @RealAAAI @TheOfficialACM That's encouraging, at least, from the AAAI side. 2021-09-01 01:04:30 @supersat It's still hilarious. These days I just check my spam folder every day though 2021-09-01 00:59:00 @supersat Next time I will send that to the internal team lol 2021-09-01 00:57:47 @supersat Like just I was in tears laughing from Gmail marking it as spam even though I knew roughly the reason 2021-09-01 00:57:13 @supersat Yeah, it's something about forwarding from my university email address (Microsoft run), somehow Google can't establish the legitimacy of anything forwarded from that email to my Gmail. It was extra funny to see Jeff Dean (work-related question I sent) in my spam folder though 2021-09-01 00:53:28 To be fair though my Gmail still marks anyone who emails me from Google as spam 2021-09-01 00:52:24 Marked spam as not spam today and Gmail was like "would you like to send this email to the team to improve our spam filter" and I was like "sure" and anyways it's an email from Leslie Lamport so I hope someone internal is amused 2021-09-01 00:47:01 @rao2z @ylecun @tdietterich @_joaogui1 @NeurIPSConf @icmlconf @iclr_conf Whenever there is rapid gain in power for someone who was once marginalized, whether by other communities in the case of AI, or by folks within a community as for many women in CS, sometimes power cognizance never catches up. AI and especially ML is there right now. 2021-09-01 00:42:49 @rao2z @ylecun @tdietterich @_joaogui1 @NeurIPSConf @icmlconf @iclr_conf Now that you all have like 10x more size and power than the rest of CS put together, it is probably a good time to have some unified cross-conference governance to enforce, for example, bans of abusive faculty across conferences 2021-09-01 00:38:50 RT @danrkports: Distributed systems papers explained using photos of cats (and other cute animals): a thread of threads https://t.co/SmRjQs… 2021-09-01 00:38:25 RT @danrkports: SwitchML: Scaling Distributed Machine Learning with In-Network Aggregation (NSDI 2021), illustrated with cat photos https:/… 2021-09-01 00:18:35 RT @isbellHFh: I'm Charles Isbell, @GeorgiaTech ('90) and @MIT ('98) alum. I'm now a Professor doing machine learning, an academic focusing… 2021-08-31 23:37:20 @mvanier42 Oh this was all an hour long research meeting hahaha 2021-08-31 23:35:36 I just don't recommend keeping cured meat next to you when you read papers 2021-08-31 23:22:26 @_joaogui1 @NeurIPSConf @icmlconf @iclr_conf But more importantly, when someone is abusive, without overarching governance it's really difficult to enforce a ban or anything like that across conferences 2021-08-31 23:21:44 @_joaogui1 @NeurIPSConf @icmlconf @iclr_conf Yeah, it's just ironic because those are exactly the kinds of decisions and changes that overarching community governance (like SIGs) can help a community make effectively while minimizing harm to individual researchers in the meantime 2021-08-31 23:13:21 @_joaogui1 @NeurIPSConf @icmlconf @iclr_conf I have strong opinions on research governance though, the ML community seems fine with anarchy 2021-08-31 23:12:35 @_joaogui1 @NeurIPSConf @icmlconf @iclr_conf Nowadays it is its own foundation though, right? That decision seems extremely odd to me, and breaks away from most professional infrastructure for dealing with fraud, abuse, unethical work, etc 2021-08-31 23:11:12 Ah yes 80% the RDA of Sodium for the day 2021-08-31 23:05:20 It turns out that mindlessly eating an entire package of sliced sopressata while reading a research paper gives me a migraine 2021-08-31 22:41:54 @_joaogui1 @NeurIPSConf @icmlconf @iclr_conf If it were part of ACM, they'd have to follow the ACM Ethics Code 2021-08-31 22:02:16 @samth Gotcha, OK, I think I just assumed there would not be shared state from the start so I got very confused by several paragraphs discussing this 2021-08-31 21:25:48 It seems like a really roundabout way of making the point that languages shouldn't share mutable state which, like, sure, but I don't know why it is phrased this way so maybe I'm missing something important 2021-08-31 21:24:39 There is a sentence about shared-nothing architectures allowing unrelated languages to interoperate that I'm stuck on in a document---what do shared-nothing architectures have to do with language interoperability? (Real question I'm clueless) 2021-08-31 20:45:49 Maybe he wasn't even a grad student actually. But he had access to some Slack I don't know about so I assume student 2021-08-31 20:42:58 Finally got the printer working. By a grad student walking into the printer room, seeing me looking clueless, and then giving me detailed instructions on how to set up the printer hahahaha 2021-08-31 20:04:23 Cafiac FIX 2021-08-31 04:35:14 RT @sabahzero: Even as a postdoc, sometimes peers complain to me about graduate or undergraduate students and it's like. That person is a… 2021-08-31 04:29:53 @tianyin_xu Yes we all know COVID is spread by printers breathing on us 2021-08-31 03:53:26 RT @nywolforg: Psst... Wolves are essential. Pass it on https://t.co/L2ECqCRaog 2021-08-31 03:52:04 @rob_rix ND folks are genuine AF because we have no other choice. So getting by in the world takes constant self-reflection since everyone will see the best of you and the worst of you always 2021-08-31 03:51:13 @rob_rix A lot of NT folks secretly harbor messed up views and just have the process managers to hide those messed up views isn't that wild 2021-08-31 03:36:30 @mycoliza It basically forces you into one of self-improvement or misery 2021-08-31 03:29:10 "Talia don't you ever think before you talk" literally no I don't have the executive functioning skills to do that so if I expect to say or do something problematic I learn to fix my internal views rather than my actions 2021-08-31 03:27:03 @secretasianman I like that band too 2021-08-31 02:18:25 OK first email was to Leslie Lamport though that's always actually fun 2021-08-31 02:00:14 @thebrownmale I need the tactile feedback 2021-08-31 01:59:53 Why are normal life things so boring 2021-08-31 01:59:31 Man I'm weird I'm like "if you can manage to send 6 email invites and then take a shower, you're allowed to do your bedtime reading about systemic racism in healthcare" 2021-08-31 01:48:13 @ian5v 2021-08-31 01:47:49 Oh I got a sweet @SIGpwny sticker today 2021-08-31 01:40:27 @karlhigley Do they have tactile feedback? I really love the feeling of pencil against paper 2021-08-31 01:32:42 @owenarden Ughghh evil 2021-08-31 01:24:12 Disclaimer just sent which honestly should go out to anyone who ever talks to me about anything ever, but especially students https://t.co/twudGSMHvy 2021-08-31 01:23:19 @owenarden I just ordered my own printer for my house haha 2021-08-31 01:23:01 @va2lam Send it to me? 2021-08-31 01:08:38 RT @OpenSexism: "The field study produced a bleak result: software professionals found gender biases in 73% of all the newcomer barriers th… 2021-08-31 00:47:17 @hillelogram I'm only half joking, let me check the Codex TOS and I'll see if I should bring my laptop Wednesday 2021-08-31 00:45:12 @hillelogram Man we should have Codex generate code primed with all of the existing notebooks so that it's an AI-generated AI-generated art machine 2021-08-31 00:42:21 @16kbps https://t.co/fQrtnQYrMK 2021-08-31 00:40:41 Office couch is a necessity also so I can go to the reading corner to read 2021-08-31 00:40:01 I literally can't read papers without printing them and writing all over them, turns out. Need to figure out how to get connected to the printer at work tomorrow 2021-08-31 00:38:05 @hillelogram What's the secret sauce beneath the flavors? 2021-08-30 23:48:49 RT @CSL_Illinois: CSL's Josep Torrellas and Jian Huang have received a $1.2 million grant from NSF to advance non-volatile memory technolog… 2021-08-30 23:22:22 I have no fears* *except telephone calls 2021-08-30 23:21:47 Plus some machine learning for funzies 2021-08-30 23:17:42 RT @hannelita: So @icfp_conference ended last Friday, and I am very happy to have been part of the organizing committee! It was a lot of wo… 2021-08-30 23:14:49 Also I think it's funny that Illinois has separate PL, SE, formal methods, and compilers email lists and I'm like WHY NOT ALL FOUR 2021-08-30 23:13:47 Not sure the easiest way to keep track of all of them, but whiteboard seems good for now. One other budding project, one more person reached out to me today, still unsure what my parallel capacity is though, may start pairing some students 2021-08-30 23:12:43 So many people and things https://t.co/9lkfgL8L1N 2021-08-30 22:55:17 @hayden_wlog yikes 2021-08-30 22:54:10 @alpha_convert You were admitted because people believed in you (with good reason). It's normal not to always believe in yourself, and to feel like they are mistaken. But usually they're right, and it just takes time to prove it to yourself (at which point you graduate and the cycle repeats...) 2021-08-30 22:53:01 @alpha_convert Having one pub before you even start grad school means you're starting with a more established record than a lot of PL folks who eventually do really well (myself included). But honestly you be you, you're not anyone else, your work and path are unique 2021-08-30 22:51:31 @alpha_convert This mostly means that ML folks are expected to do more earlier---and that isn't necessarily indicative of later success at all 2021-08-30 22:48:03 @alpha_convert If it helps, even your professors feel this way 2021-08-30 22:46:00 @alpha_convert That's normal. You're good though 2021-08-30 22:45:09 @hayden_wlog There wasn't really any guarantee that the job market would recover this year, university budgets are mysterious 2021-08-30 22:36:29 I was kind of worried about the TT job market being too highly saturated for those who waited the pandemic recession and hiring freezes out and are on the market this year, but uh, now I think you'll all be fine... https://t.co/19MWC0VPwZ 2021-08-30 22:32:27 @SeattleCommNet @uwcse @infrared_ether They need $25,000, but so far they have just $2,683! This is a really nice way to make a really concrete impact and help people in, for example, homeless shelters in the Seattle area access the internet. Super important these days, especially during a pandemic! 2021-08-30 22:30:48 @SeattleCommNet @uwcse @infrared_ether You can find out more here: https://t.co/BpDET4Wj0z 2021-08-30 22:29:56 Friends, super please donate to @SeattleCommNet, a nonprofit started by my friend and @uwcse PhD student @infrared_ether. This awesome nonprofit sets up community-run internet access in at-need communities in King County, and trains community members! https://t.co/eiQZs4Wnyh 2021-08-30 21:32:24 @crazyuddie @emilymbender Saaame will be pointing to this thread when I talk about neural proof synthesis during my proof automation class 2021-08-30 20:11:26 Super huge thanks to @hannelita for hosting me!!! 2021-08-30 20:10:21 RT @haw95: Thank you, Talia! See more at https://t.co/ztgZVLx0Yj 2021-08-30 20:00:49 Check out my @icfp_conference AMA on @SigplanM from last week, now on YouTube! https://t.co/8uU3oqfmjf 2021-08-30 19:56:16 @NathOnSecurity Suddenly I'm an AI person 2021-08-30 19:47:19 No idea who needs to hear this, but workplace abuse is still workplace abuse when it comes in joke format. Appalling https://t.co/sgHyL46fe6 2021-08-30 19:33:18 @mmitchell_ai That is just straight workplace abuse 2021-08-30 19:27:41 @mmitchell_ai That's so flatly unacceptable I can't believe so-called leadership thought that was a remotely reasonable or funny thing to do 2021-08-30 19:24:14 @mmitchell_ai Jeeze I'm sorry anyone thought it was OK to treat you like that 2021-08-30 19:23:59 @mmitchell_ai Wait, what 2021-08-30 17:33:03 Bonus points if it can figure out how to print things and set up projectors 2021-08-30 17:29:09 RT @TartanLlama: An introduction to register allocation, in which Lexical Cat takes a trip to Paris https://t.co/BOxbEf1Opg 2021-08-30 17:28:18 @XginkgoG That's an easier problem 2021-08-30 17:27:59 @_lewtun RIP humans 2021-08-30 16:49:05 The true test of Artificial General Intelligence will be an AI that can figure out the right person to email given a question to ask. At that point it will have achieved supremacy over humans 2021-08-30 15:13:33 Cafiac fix 2021-08-19 05:13:33 Dependency analysis for proof language model context, and impact of that on the language model 2021-08-15 17:06:24 @blackeuler "What if we used racist technologies to automate ableism?" 2021-08-15 17:05:13 @nbhansen Most boring meetings I am part of just don't have someone doing a good job of enforcing an agenda, and so people just argue and argue off topic to listen to their own voices 2021-08-15 16:58:20 RT @Dedjartes: @WIRED "Can we improve meetings by dsicriminating against people with certain disabilities even more than we do now?" uh, n… 2021-08-15 16:57:30 Yeppp https://t.co/upPyrzPHOB 2021-08-15 16:55:33 You see, the trick is to force them to be happy 2021-08-15 16:53:20 @mjskay They're everywhere 2021-08-15 16:44:49 "Our employees are unhappy. Can surveillance and automated discrimination help?" 2021-08-15 16:44:23 "facial recognition that measures who's paying attention" legit just automated discrimination against ADHD folks, not to mention the biases encoded in the tools themselves, cool cool 2021-08-15 16:42:34 This is exactly the wrong lesson. The right lesson should be fewer unnecessary meetings 2021-08-15 16:42:07 I'm sorry what https://t.co/55F8kUuKjt 2021-08-15 16:35:10 And it also means treating your team like whole human beings 2021-08-15 16:34:14 Competition can be healthy, but only if coupled with sportsmanship. It should look like this https://t.co/e9rsGLside 2021-08-15 16:26:09 RT @danidonovan: Hey #NeurodiverseSquad, I need your help! I'm working on an ADHD-friendly "anti-planner," and want to include some tweets… 2021-08-15 16:14:15 RT @louispilfold: Join a union 2021-08-15 16:08:10 Legit a system that encourages and rewards outright psychopathic behavior https://t.co/JNju8isbn9 2021-08-15 16:06:26 It must be so weird to really internalize the competitive world of capitalism so much that you actually view yourself as at war with other companies https://t.co/lTOlSLAcjn 2021-08-15 16:03:46 @hadip Yikes 2021-08-15 15:51:53 Wholesome running content https://t.co/6XyYUh6Msc 2021-08-15 15:48:06 @NBCSports @NBCOlympics @HaileGebr for sure! 2021-08-15 06:50:49 Food will never be the same. Also something something computers 2021-08-15 06:50:22 12 years ago today, a diversity program changed my life forever. In particular, it introduced me to uni (sea urchin) https://t.co/g5MmqVwFZN 2021-08-15 06:27:55 https://t.co/x1wDmZ8ibX 2021-08-15 06:25:10 Lions are the only truly social species of cats. Cat species can either roar or purr, not both. Polar bear fur is translucent, not white. There's a white black bear called the spirit bear in BC, and they are better at fishing than black bears without the white fur 2021-08-15 06:22:44 To be fair to Facebook, I've had a partner in the past do basically the same thing to try to distract me after being mean to me. "Here's a bear photo" I mean OK fine I'm still mad but yes the bear is cute 2021-08-15 06:18:14 Me, depressed: Facebook: Me: YES, MAJESTIC Facebook: Me: ADORABLE, AMAZING Me, not depressed: Facebook: ? 2021-08-15 06:16:31 I used Facebook a lot when I was most depressed, and it has consequentially inferred that I mostly want to see photos of lions and bears whenever I open it up 2021-08-15 06:07:27 RT @hilmargustafs: @TaliaRinger Yeah, being able to add alt text after the fact would be - As for getting to 100%, if you follow https://… 2021-08-15 06:07:19 @hilmargustafs Nice 2021-08-15 05:53:00 This QT game brought to you by GPT-3 https://t.co/7l4BFX40RG 2021-08-15 05:50:53 agony https://t.co/VP7jbqqJJ0 https://t.co/otEMcFE57P 2021-08-15 05:20:39 @zwegner @johnregehr Are men OK https://t.co/WhNuIPNXvS 2021-08-15 05:10:17 @AmandaSmithSays @clapifyoulikeme @chick_in_kiev Speak for yourself, I'm still waiting for my Hogwarts letter 2021-08-15 04:51:37 RT @shaimgo: Someone told me today that they didn't disclose an accommodation that they need due to their disability when hired because the… 2021-08-15 04:41:22 @JanelleCShane Relatable 2021-08-15 04:40:46 yay https://t.co/AzhkmwD16l 2021-08-15 04:40:04 RT @JanelleCShane: GPT-3 has seen all the viral tweets during training - which means it can predict upcoming twitter trends. Right? I prom… 2021-08-15 04:37:09 Gentle reminders always help, and sometimes result in comments from me giving what the alt text should be 2021-08-15 04:36:27 I think my hit rate for remembering to add alt text is like 50% right now. It's much higher than the 0% it was at a year or so ago, thanks to lots of reminders from other folks here. But it'll take time to get to 100%, especially without the ability to retroactively add it 2021-08-15 04:34:58 @adamgreattweet @NerdPyle @get_altText 2021-08-15 04:25:39 @johnregehr 2021-08-15 02:09:29 RT @RebeccaRHelm: BUTT LIGHT! Some deep-sea fish literally shine light out of their butts. And the glitter on glitter worms may actually de… 2021-08-15 01:57:14 AWS be like https://t.co/sk36BHhOC1 2021-08-15 00:48:38 To feel guilt and regret (in moderation, when appropriate) is to be whole and human and wonderful, always on a path to improving both yourself and the world you live in. 2021-08-15 00:46:48 Guilt and regret are incredibly useful and wonderful feelings to feel unless they consume you. Guilt drives you to repair damage you have caused, and to improve the world around you. Regret drives you to not make the same mistakes in similar circumstances in the future. 2021-08-15 00:43:05 @Debtlifts This is such a harmful attitude 2021-08-14 23:40:19 What is this, a beaker for ants? https://t.co/mesUpconlb 2021-08-14 20:05:32 @mollerwa Thank you 2021-08-14 20:03:27 Closer images and poem text for all six here: https://t.co/ba7EkceHIm 2021-08-14 19:59:22 @secretasianman It's so hard to find people who want to buy them. I'm not sure if they just aren't good enough or what, I always thought they were quite good and I'm even a published poet 2021-08-14 19:55:46 @secretasianman Some include 3D shapes, I don't want to smush them 2021-08-14 19:52:25 Like $10 if you are my friend, let's say 2021-08-14 19:52:05 Does anyone want to buy a collage poem? 2021-08-14 19:51:14 How on earth am I supposed to transport these to Illinois https://t.co/8glMtfu1Wr 2021-08-14 19:33:23 Note from PLSE Town Hall, pre-pandemic, found while packing https://t.co/FC1OzNBn6M 2021-08-14 18:36:13 If you want to do something like this at a much larger scale one day, also, super let me know, I'd love to collaborate 2021-08-14 18:34:31 User study of people using popular machine learning frameworks to identify what development tasks they struggle with, and where PL/SE techniques and tools could help 2021-08-14 18:32:49 Training a model on incremental changes to programs and proofs 2021-08-14 18:31:09 User study of proof engineers but at a much larger scale https://t.co/FNRNYeIkc5 2021-08-14 18:27:43 Even the consent process is baked in and ready to modify for your needs: https://t.co/WMJIp3bW7L 2021-08-14 18:18:42 But I'd read the paper still https://t.co/Ku5Ut5mkqU 2021-08-14 18:18:20 Here is the infra if you want to adapt it to gather your own data: https://t.co/oNb2fMZ66l 2021-08-14 18:17:44 @JoeOsborn It just logs to a local file and then sends in batches to the server. All that is needed is a single line in the .coqrc, it is implemented as a plugin. But I'd read the paper first, there were limitations like lack of success/failure info so we had to infer it via proxies 2021-08-14 18:15:22 @JoeOsborn Yeah it is IDE-independent, hooks into the REPL. Should be adaptable with a few tweaks (to talk to your own server and so on): https://t.co/WMJIp3bW7L 2021-08-14 17:07:07 This work was a collaboration with @sanchez_stern who will be on the job market in the near future 2021-08-14 17:05:54 I honestly suspect this is true for code too, and part of why, for example, with Codex it is very difficult to make very small, interesting, pinpointed changes to generated code that is slightly incorrect 2021-08-14 17:03:58 Instrument your IDE or REPL or whatever, with permission from proof engineers, and you get every incremental change, and that data is extremely rich and informative. (Processing it meaningfully is still hard, I'd love to see an automated analysis of it) 2021-08-14 17:02:42 Also, even if you look at Git commits and so on, the level of granularity you see completely erases the incremental changes that people make. The data is simply not there, because people don't show all of their work, they commit things once they are satisfied 2021-08-14 17:01:05 What surprised me is that repair happens constantly *during development*, so I think this is relevant for any proof productivity tool 2021-08-14 17:00:30 One of the main points of my thesis work is that if you want to build tools to help people refactor or repair their proofs, you can get a lot if you actually reuse and generalize information found inside of related changes 2021-08-14 16:58:49 I've probably shared this before, but I still think it's worth learning from, especially as communities start to merge a bit more. @ChrSzegedy @ilyasut and so on 2021-08-14 16:57:05 The paper is also useful for proof assistant designers & 2021-08-14 16:55:39 Here is the talk video, from January 2020: https://t.co/nuF1H2Bzkt 2021-08-14 16:54:20 Anyone doing any research on proof engineering tools, especially if coming from outside of the community, should take a look at our user study in which we collected this data: https://t.co/N4c0rMKgHp 2021-08-14 16:52:37 Oh, folks doing ML for proofs might be interested in this data set, collected over a month with the consent of the proof engineers, that we put together capturing all incremental changes 8 proof engineers made during proof development for an entire month: https://t.co/hzcOafKXgM 2021-08-14 16:31:45 20 minutes of biking will wait a day and see if it hurts tomorrow before doing any more 2021-08-14 16:20:57 I can very lightly ride a spin bike in the boot without pain today!!! 2021-08-14 15:20:07 For real though the sock I was wearing was very tight and elastic, it seems plausible to me this made the impact worse 2021-08-14 15:18:46 Got in an Uber to Lower Woodland Soccer Field, to say hi to my track coach, but in crutches and a boot. Uber driver asked if I was going to play soccer and I'm glad he has a sense of humor 2021-08-14 15:03:57 For those of us who are clumsy and crash into things and break bones, there need to be crash tests for socks 2021-08-14 14:52:12 @kmett @dbp__ Yay 2021-08-14 14:46:02 @kmett @dbp__ Ah I left employment in June. Just insurance ends in September 2021-08-14 14:38:16 @kmett I hope so 2021-08-14 14:36:37 @kmett No idea what that is, what is it 2021-08-14 14:35:36 Or that we had a reasonable healthcare system 2021-08-14 14:33:12 Now I wish I'd turned in my thesis over the summer quarter, so that I'd be covered by UW all of fall quarter 2021-08-14 14:30:47 Ah, I think I briefly lose health insurance while I have to figure out if my foot needs surgery, and also get set up with two new doctors in Illinois. God bless America 2021-08-14 05:14:59 LLMs quite promising for matching style, judging by Codex generated tactics with context. Just need to also produce tactics that work, but that's what the semantic info we encode is useful for https://t.co/Ar82K5sJcc 2021-08-14 04:56:27 I'd like the neurosymbolic proof term to proof script decompiler I'm working on with some collaborators and students to one day replicate tactic style with few examples, so it is nice to know this is possible with a large syntactic model (and it makes sense) 2021-08-14 04:55:30 Codex on Coq with context: extremely beautifully captures tactic style, predictably rarely writes proofs that work. This is a good thing though, because replicating style is often the missing piece for proof automation 2021-08-14 02:25:23 In particular, its capabilities are *so different* from traditional PL tools that having access and playing around opens up hundreds of new research directions for PL researchers 2021-08-14 02:18:22 The PLSE lab quite enjoyed seeing it in action and learning about what it does well and what it struggles with, and what the experience of using a tool built on Codex is like. I encourage more PL researchers to request access 2021-08-14 02:10:08 I also let @ztatlock try, and he drew some fractals, and then later a flying "pig" though Codex actually generated a red box instead of a pig 2021-08-13 23:45:28 @bisexualposibot Thanks 2021-08-13 23:45:14 RT @bisexualposibot: professor bisexuals are amazing 2021-08-13 23:31:53 My mom is in town to help me move since I'm on crutches. So I just told her she could write her first program, then opened up the Codex JS sandbox. She made a website that explains why it gets dark at night 2021-08-13 20:10:10 @JaedynRuli But cool I'll avoid the specific phrasing 2021-08-13 20:09:54 @JaedynRuli It was definitely not meant to make you look antisemitic, I just grew up always giving full context for the things I say and ask 2021-08-13 17:27:33 @Anarchia45 Won't Fix - Working as Intended 2021-08-13 15:57:46 My housemates immediately reacted by trying to help make the house as accessible as possible and that was really nice. I have chairs around my room to help with changing and so on, now 2021-08-13 15:55:43 Main worry is that because the break extended into the joint, I need to make sure I don't need surgery in order for it to heal correctly 2021-08-13 15:52:33 Also, I was nervous about explaining to my coach that I ruined my Boston chances by walking into a metal chair at work and breaking a bone and joint. But his response was "Oh, my gosh! That sounds like something I would do!" 2021-08-13 15:50:58 I know it will be about a week before I can walk in a boot, weeks more before I can take the boot off, and well over a month before I can get back to running. But it's nice not to have to use my crutches to get out of bed 2021-08-13 15:49:33 Day 1: constant pain Days 2, 3: pain only when weight-bearing Day 4: can put heel on ground and limp short distances wow the human body is amazing 2021-08-13 15:39:52 @JaedynRuli Deleting to avoid inevitable harassment that happens anytime anyone does that 2021-08-13 15:38:35 @JaedynRuli But cool I love when internet strangers interpret all of my questions as disingenuous statements 2021-08-13 15:37:56 @JaedynRuli It was an actual question, I couldn't tell if the problem is with phrasing or the wish for genetic children 2021-08-13 15:36:36 While we're at it, can we also carry this lesson over to climate change, please? Individualism is broken 2021-08-13 15:35:20 And that doesn't need to be mandates and laws 2021-08-13 15:33:52 It's so weird to see all of these news articles like "if all of you just individually behave reasonably, we can slow the spread of delta" like have you been awake for this entire pandemic? Only systemic approaches work 2021-08-13 15:17:59 Obligatory https://t.co/JRrTno0v4a 2021-08-13 15:15:10 With that said, this could be a fun premise for a sci-fi novel 2021-08-13 15:14:29 It must be weird to be famous enough that people respond to your selfies with unrelated and confusing questions https://t.co/Q2PgOmVWFq 2021-08-13 14:59:43 Hi yes this is dog https://t.co/C1mmzSbtQI 2021-08-13 14:34:40 RT @blkgirllostkeys: Grocery shopping (and eating) is a lot easier when you don’t have to cook everything! This is one of my favorite posts… 2021-08-13 06:45:11 RT @disconcision: in the manner of @jeanqasaur 2021-08-13 06:29:02 @lindsey @danrkports @blucia0a We can fix this! From now on, it's called ASPLos 2021-08-13 06:27:57 RT @avsm: @lindsey @blucia0a and the splendid irony of “formal verification & 2021-08-13 06:17:12 @larsr_h https://t.co/6IWY6C2Sg3 2021-08-13 06:15:29 @qnym3ria It is a consequence of privilege 2021-08-13 06:13:32 @porglezomp Hmmmmmm does it? I feel like it could mean start not not using 2021-08-13 06:12:34 @qnym3ria How do internet strangers always read such weird implicit emotions in my Tweets 2021-08-13 06:11:12 @qnym3ria I'm not angry? Who are you anyways? 2021-08-13 06:09:50 Is it? 2021-08-13 06:09:29 Is not not functional programming functional programming? https://t.co/QD3oFK1HPo 2021-08-13 06:04:36 @qnym3ria Nobody is actually an "individual," all of us are shaped by the groups we are perceived to belong to, and the cultures around us. Men can get the illusion of being "individuals" because "man" is the dominant societal gender group 2021-08-13 05:51:19 @nschiefer @hillelogram Curry and Howard would like a word with you 2021-08-13 05:41:17 Only men ever get to be viewed as and view themselves as "individuals" https://t.co/XOsPIUH16Y 2021-08-13 05:33:56 RT @rothgar: It's called STRONGLY TYPED 2021-08-13 05:26:47 Yessssssssssss https://t.co/jrE3cEaNMn 2021-08-13 04:40:31 RT @reillymack_: woke up from an incomprehensible dream where this was the hot new meme https://t.co/6NZXMQ4Lss 2021-08-13 03:55:29 This incredible thread with Christian and Marisa, while not quite refined into an idea yet, is a source of thousands of ideas to come, so bookmarking https://t.co/XwyLootNIq 2021-08-13 03:48:14 @ChrSzegedy @MarisaVeryMoe I think symbolic automation can help in needing much, much less data, in domains for which insufficient data is available, or when efficiency *really* matters 2021-08-13 03:46:08 @MarisaVeryMoe @ChrSzegedy I don't know if I believe in a distinction anymore, though, I think the distinction is going to get very blurry going forward 2021-08-13 03:45:19 @MarisaVeryMoe @ChrSzegedy I'm just always looking forward, and I think to get wherever we want to go, we're going to use a hell of a lot of NNs, and they're going to marry the symbols we love somehow---but that doesn't mean the actual automation will necessarily be symbolic 2021-08-13 03:43:02 @MarisaVeryMoe @ChrSzegedy I think the lines are getting really blurry to me---if a NN discovers symbolic information, is the approach symbolic? What if it leverages that later? What if it reifies it as syntax and then leverages it naively? All semantic info in PL can be syntax in a different language IMO 2021-08-13 03:39:13 @ChrSzegedy @MarisaVeryMoe This I think would take a much longer conversation, but I think it's primarily about what information they have access to to begin with---it is not hopeless though, it is a data problem 2021-08-13 03:36:50 @MarisaVeryMoe @ChrSzegedy And then to use a whole bunch of equivalences hanging around somewhere, probably you want something like an encoded e-graph 2021-08-13 03:35:58 @MarisaVeryMoe @ChrSzegedy I *do know* that by looking at differences in datatypes, you can discover and prove equivalences though, but you need access to all of the information necessary to do that, and you need it to be something a tool ever actually tries to do 2021-08-13 03:33:36 @MarisaVeryMoe @ChrSzegedy In the mathematical sense, like equivalences and almost-equivalences 2021-08-13 03:32:56 @MarisaVeryMoe @ChrSzegedy The relations problem is one that Alpha* all struggle with too 2021-08-13 03:32:07 @MarisaVeryMoe @ChrSzegedy I think we will need to at the very least pass through a phase where we explicitly encode some nontrivial semantic info like relations between datatypes. But I could believe that eventually those could be discovered from rich enough data 2021-08-13 03:16:25 @ChrSzegedy @MarisaVeryMoe If you want to capture that much granularity, you need development data from IDEs IMO 2021-08-13 03:15:50 @ChrSzegedy @MarisaVeryMoe My experience so far (in 2 hours playing with just the sandbox, granted) is that it is near impossible to get Codex to make very small, subtle, and interesting changes to existing code. This does not surprise me, because the data for that is erased even at the commit level 2021-08-13 03:12:44 @MarisaVeryMoe @ChrSzegedy Instrument all interaction like we did for REPLICA, though, and collect much more granular data on refactoring and repair, and then maybe syntax will do pretty well at even that 2021-08-13 03:11:34 @MarisaVeryMoe @ChrSzegedy Or maybe one day we will have the glorious neurosymbolic future where we don't need to separate those steps, but I don't believe they can be combined in a world of just syntax and just the data we currently have access to 2021-08-13 03:09:43 @MarisaVeryMoe @ChrSzegedy But it is quite useful to have tools that get you 80% there 2021-08-13 03:09:21 @MarisaVeryMoe @ChrSzegedy Sure, yeah. My point is Codex, by generating boilerplate, can get 80% to the first step sometimes in interesting use cases. Pair with some symbolic techniques, and some refactoring and repair, and you can then generalize & 2021-08-13 03:07:30 @MarisaVeryMoe @ChrSzegedy But generating boilerplate is just discovering implicit relations, which can then be automatically refactored or repaired 2021-08-13 03:00:08 @MarisaVeryMoe @ChrSzegedy Yeah, that's cool, I believe in that too, but it doesn't seem to contradict the original statement I made, as long as there is then some automatic refactoring or repair. Come on, I wrote a whole thesis on this kind of thing haha 2021-08-13 02:57:06 @MarisaVeryMoe @ChrSzegedy Then you are writing the boilerplate, though, you're just not committing or pushing it :) 2021-08-13 02:56:24 @ChrSzegedy @MarisaVeryMoe Oh I agree with that, it's right along the lines of my thesis work. But the boilerplate still exists, the question is the interface that hides it 2021-08-13 02:55:10 @ChrSzegedy @MarisaVeryMoe Yessss this is something I'm *extremely* interested in, especially in discovering the right questions to ask 2021-08-13 02:54:27 @MarisaVeryMoe @ChrSzegedy But yes, perfect foresight is a remedy! But I don't believe in perfect foresight. Even the most experienced expert proof engineers and language developers lack it, and mostly only make those changes post-hoc (I have data for this) 2021-08-13 02:53:32 @MarisaVeryMoe @ChrSzegedy If you have functions or proofs about addition or multiplication, for example, they will all repeat themselves over and over again by way of the relation between them 2021-08-13 02:52:14 @ChrSzegedy @MarisaVeryMoe So in the ideal scenario there needs to be a way to help the programmer discover that, and distinguish that from the tool just doing the wrong thing 2021-08-13 02:51:31 @ChrSzegedy @MarisaVeryMoe Oh, yeah, I agree in terms to automated refactoring and repair (come on you know me). I just mean that after generating the code, the programmer will likely often realize she doesn't actually want the thing she said she wanted 2021-08-13 02:49:09 @MarisaVeryMoe @ChrSzegedy My philosophy has always been define one, show me how you want things to change, and then derive the rest from that demonstrated change, prompting you for only the unknowns 2021-08-13 02:48:12 @MarisaVeryMoe @ChrSzegedy What should the human read? Like consider an interpreter for a simple calculator language, where each operator has plenty of structural repetition inherently 2021-08-13 02:42:59 @ChrSzegedy @MarisaVeryMoe So, I think you need a way to not just drill down, but also drill back up 2021-08-13 02:42:41 @ChrSzegedy @MarisaVeryMoe I like the idea, I think one thing it is missing though is that people often have no idea what they want ahead of time. If you watch people write proofs (or look at the REPLICA data) the program, specification, and proof all evolve alongside one another 2021-08-13 02:41:08 @MarisaVeryMoe @ChrSzegedy Principles of Programming, the PL group at CMU 2021-08-13 02:39:17 @ChrSzegedy @MarisaVeryMoe Want to do it ahead of time? Probably you're looking at something like Ornaments. In either case boilerplate for very structured things is often just relations like equivalences revealing their implicit presence 2021-08-13 02:37:26 @ChrSzegedy @MarisaVeryMoe So there is often a well-defined transformation of programs or proofs along those mathematical relations, which means your code will be repetitive. You can capture those patterns syntactically, semantically (my thesis work), whatever, but they'll be there 2021-08-13 02:36:25 @ChrSzegedy @MarisaVeryMoe Eh, it sounds very CMU PoP, our frameworks and type systems and so on will never be perfect. People write boilerplate when writing languages because operators and language constructs are mathematically related to one another 2021-08-13 02:31:09 @MarisaVeryMoe @ChrSzegedy So far the big barrier is that it's really difficult to debug and then repair interesting generated code. And that in itself is interesting, and probably opens up lots of fun research 2021-08-13 02:26:46 @MarisaVeryMoe @ChrSzegedy Because programming languages have a ton of boilerplate, so it's a fun question to see how close you can get with syntax, and then ask what would be needed to get from that to a language that works correctly 2021-08-13 02:18:09 @ChrSzegedy @MarisaVeryMoe I am quite interested in interactive spec generation, though, and always happy to talk about ideas for that future 2021-08-13 02:06:47 That's enough for getting acquainted today, but I'm hoping to look into the API once I'm settled in Illinois, and look at all of the research questions I have at the intersection of Codex and formal methods, or more generally how to get trustworthiness and correctness 2021-08-13 01:35:06 Ah, the associativity of the operators is wrong, that's the only issue. Interesting 2021-08-13 01:22:40 @moyix @ChrSzegedy @MarisaVeryMoe I'm just in the JS sandbox right now, but I'll take a look at the API soon, that sounds promising 2021-08-13 01:21:54 Debugging tools for Codex-generated code https://t.co/WmZXt2leln 2021-08-13 01:21:24 I tried to get it to implement a simple calculator language with an interface, and it mostly works but it's slightly wrong for some reason. It's surprisingly hard to debug when something goes wrong---a clear place where PL tools could be helpful 2021-08-13 01:20:24 @ChrSzegedy @MarisaVeryMoe I think I just get excited about every new idea ever, which means whenever there is new technology I want to play with it and do things 2021-08-13 01:16:44 @ChrSzegedy @MarisaVeryMoe You're always a little more radical than I am 2021-08-13 01:09:47 No commentary yet, other than it being fun. Basically I want to play around for a while to see what's easy, what's hard, what's close-but-not-quite-right (STLC actually is a nice start, even if wrong), so that I can try augmenting with formal methods https://t.co/jwUsKKzGuk 2021-08-13 01:07:58 @ChrSzegedy @MarisaVeryMoe But first I need to play around and test all of the limits of raw syntax 2021-08-13 01:07:38 @ChrSzegedy @MarisaVeryMoe My interest isn't in out-of-the-box Codex though, it's in what magical awesomeness we can get augmenting Codex with formal methods tools 2021-08-13 01:05:33 @ChrSzegedy @MarisaVeryMoe It sometimes just gets caught up on these weird death loops, I wonder why that happens. I've seen similar behavior from GPT-2, not sure about GPT-3 https://t.co/uK8s2YbAj0 2021-08-13 01:04:02 @ChrSzegedy @MarisaVeryMoe Oh, I just think it's cute, I think it's actually plausible one could use some modified version of this to write programming languages that don't suck. I'm trying to implement a programming language this way right now and see what happens 2021-08-13 00:43:01 I told it to implement STLC https://t.co/TRLe2tglCv 2021-08-13 00:37:42 I'll do real researchy experiments at some point but for now just want to play a bit with the JS sandbox to get my feet wet 2021-08-13 00:37:04 hahahaha it's so cute sometimes, I told it to print all primes and it printed all primes up to 100, so then I told it to print all primes known to mankind and it printed all primes up to 1000000 2021-08-13 00:21:03 I am, unsurprisingly, quite into the fact that you can tell Codex to change existing code it has generated 2021-08-13 00:14:20 Codex seems quite amazing at a bunch of things I could never imagine doing with my kind of automation so far, and quite bad at a bunch of things I can easily do with my kind of automation. Perfect 2021-08-12 23:51:41 omg yes, so many research directions just opened up https://t.co/FBhOy0A1T6 2021-08-12 23:31:53 @ArianaGsparks @antoniacbz @lauramcuervo @baddanadanabad I was enthusiastically programming when I walked into the chair, so that's cool 2021-08-12 23:26:37 @wilton_quinn Yeah, it also adds a lot of tasks, like seeing a podiatrist as soon as I get to Illinois to check if I need surgery, and calling the airport to make sure I have wheelchair access for my layover so I don't miss my flight. 2021-08-12 23:25:32 So, for now, I think it'd make me very happy to hear about things I do that aren't pointless, like things that actually help other people in some way. 2021-08-12 23:24:51 I think this week it'd make me really happy to see signs of appreciation from anyone about anything I do. After watching one of my big goals go down the drain because I got excited about code and kicked a chair, I kind of feel like everything I do is pointless. 2021-08-12 23:21:34 I'm also handling SIGPLAN-M renewals this week. I hope I can at the very least communicate plans with everyone before I move. 2021-08-12 23:20:38 Everything takes extra energy right now while on crutches. Getting out of bed, getting water, getting food, every day things are really exhausting, and my whole body is sore. It's really challenging to adapt to that as I'm getting ready to move Monday, and I'm quite overwhelmed. 2021-08-12 23:19:01 @mgambogi @echristo I mean from the industry side. I know, for example, that at Google it's way easier to work out summer things than anything during the year, unless it's a continuation of a collaboration that started in the summer 2021-08-12 22:59:54 @clegoues I think since I'm chairing SIGPLAN-M for one more year, I won't have much extra time 2021-08-12 22:58:51 @echristo Off-cycle interns are harder, right? 2021-08-12 17:09:24 I don't know if any of the industry research groups take on undergraduates 2021-08-12 17:08:26 I also have really, really good industry connections---is there a way to work those to help undergraduates get started on research while I build my lab in parallel, so that there are spots waiting when they're back? 2021-08-12 17:07:04 Presumably I'll teach a seminar in the spring, and that might also be a good opportunity for undergraduates to work on this kind of work, and ramp up while I work to recruit more graduate students going into 2022 2021-08-12 17:05:35 Mostly because I fear not being helpful enough as I'm also ramping up on faculty life. But I think I'd be happy if some of those grad students are my remote collaborators at UMass or UW, not just my UIUC grad students (I'll likely have just one when I start) 2021-08-12 17:04:21 What are folks' thoughts on taking on undergraduate researchers as first year faculty? I have wonderful undergraduates reaching out to me, but my inclination right now (not sure if reasonable) is to only keep open as many spots as I have grad students willing to work with them 2021-08-12 16:35:05 @RobBradford17 To a lot of engineers, the problem here isn't the software. It's *you* 2021-08-12 16:20:05 @Dev14e @deckeresq Sorry James, I know you were enjoying the egg puns, didn't mean to poach your idea 2021-08-12 16:12:58 RT @alilleybrinker: @TaliaRinger Yup! "Revealed preference" is really "revealed preference under the constraints of the current system," wh… 2021-08-12 16:07:53 I think it's OK to seek profit, but you should be honest with yourself when doing so is not aligned with what users want and are asking for. No metrics can capture what users "really want" 2021-08-12 16:06:07 So engineers and managers would think of metrics that were profit-aligned, then justify them internally by claiming they match the true desires of users 2021-08-12 16:04:54 What was actually meant by this, of course, is that what users want doesn't always align with profit incentives 2021-08-12 16:03:57 One of the most harmful attitudes accepted as common wisdom when I was a software engineer 2012-2015 was that users don't actually know what they want, so we should focus more on metrics and less on what they say 2021-08-12 15:50:23 @Dev14e @deckeresq Understandable if you're now feeling scrambled. Hope the rest of your day goes over easy 2021-08-12 15:45:59 @ArianaGsparks @antoniacbz @lauramcuervo @baddanadanabad it’s so fucked up that i went into this pandemic 29 (young, spry, hopeful) and now im 31 (old, brittle boned, writing my last will and testament) (no but literally I broke my toe walking into a chair two days ago) 2021-08-12 14:34:54 I had a dream about a yellow and black cat named Pikachu 2021-08-12 14:32:25 @taylerwithanE Ahhh hope it heals quickly 2021-08-12 06:49:33 @paulfkrause Nice 2021-08-12 06:41:31 @peztronic @johnregehr Talia Ringer, Dad of Philosophy 2021-08-12 06:29:59 When I was a baby, I never figured out how to crawl. I would scoot around with one leg outward and the other behind me. Anyways, now that I have a broken toe, I'm grateful to have practiced this skill 2021-08-12 06:28:36 I'm still trying to figure out what it means to figuratively break my toes writing code 2021-08-12 05:57:19 @yoavgo "shockingly" 2021-08-12 05:48:57 @hannaalkaf @davidlasky 2021-08-12 05:35:59 https://t.co/mtHiDe3ki8 2021-08-12 05:35:09 Right yeah, about the concern I had with everyone competing to train larger and larger models, and how that impacts the ethics of the environmental costs of training each iteration during development even if inference is more efficient in the end https://t.co/VOiuJYLglf 2021-08-12 05:26:00 @danrkports Argh you beat me to it 2021-08-12 05:25:47 @danrkports @schemeprincess F . . F . @ . . . F . . 2021-08-12 05:05:50 I'm also so dehydrated because it's so inconvenient to refill my water 2021-08-12 05:05:27 Gosh it hurts so much more at night, and doing literally anything is exhausting 2021-08-12 03:58:41 @johnregehr Joke's on you, I wrote part of the Amazon nav bar in Perl. You load it every time you access https://t.co/qnLGOFKIzl. May explain why I got appendicitis 2021-08-12 03:54:30 I love how many people took this figuratively and not literally 2021-08-12 03:52:43 On August 7th I was proud of myself for running 14 miles. Just a few days later, I'm proud of myself for showering without any assistance. Injuries, man 2021-08-12 03:14:21 Also curious what happens if you give it contradictory instructions, not really a research question, but probably preempts interesting research questions 2021-08-12 03:13:01 Codex generated code by sequences of NL instructions will be complicated. If the source code matters for the application domain, can we then do verified refactoring into simpler or more general functions that capture the same behavior? 2021-08-12 03:06:23 Codex can probably help a lot with the NN part, which may reduce the problem (for common languages at least) to ensuring correctness by way of formal methods and a nice user interface. Which sounds extremely fun 2021-08-12 02:35:59 @mmitchell_ai @DataSciBae @timnitGebru I submitted a correction to the article since it's also just irresponsible journalism 2021-08-12 02:17:57 @mmitchell_ai @DataSciBae @timnitGebru But I also know trauma isn't something you can rationalize away and it is extremely valid to have a physical reaction to being repeatedly erased 2021-08-12 02:14:40 @mmitchell_ai @DataSciBae @timnitGebru Ahhhh Meg I promise a lot of us see the amazing team you put together and the work you put into it and the amazing leadership. We just don't write news articles 2021-08-12 02:04:08 Just in a context where folks can't call the science politics 2021-08-12 01:57:24 https://t.co/7I74j15BbT 2021-08-12 01:36:50 @ChrSzegedy Right, so I guess my point is sometimes there's a strain between those when you're in an extremely specific domain for which relatively little data exists 2021-08-12 01:32:48 Ah, I suspect I just chose a bad example of the problem https://t.co/tjRBAEJ5fI 2021-08-12 01:31:52 @ChrSzegedy Yeah, but I think it's the much less common words that will cause problems, so maybe the gripe is with my choice of "group" and "ring" which are very common in algebra. Or words that occur in much closer domains, like "proof" in proof assistants versus Blockchain 2021-08-12 01:29:33 @ChrSzegedy But like, why on earth would I ever say "fruit" in that context to begin with? Practically, I've always wanted a way to correct those tools at that point so they can understand that there's no way in hell I'd be talking about fruit 2021-08-12 01:27:31 @ChrSzegedy One thing I think happens is that this problem compounds itself. So like if a captioning tool interprets "proof" as "fruit," which is a totally nonsensical thing to do when I'm talking about math, but if it does the way Google Meet does, "lemma" is more likely to be "lemon" ... 2021-08-12 01:24:51 @ChrSzegedy A lot of those systems are probably still using old technology, but I also feel like there's a gap here between what you would expect to be true and what actually happens when you use these tools 2021-08-12 01:23:29 @ChrSzegedy But practically it seems to be a huge problem in all systems I've ever used 2021-08-12 01:20:37 Unsurprised though if tuning helps with this problem very significantly 2021-08-12 01:19:23 So anyways I'm extremely confused that people don't think words that occur very often in general contexts with one meaning but only occasionally in domain-specific contexts with a totally different meaning aren't a problem for LLMs when operating in those specific contexts 2021-08-12 01:12:55 (Many proof assistants use dependent types. Dependent types are types that let you make values explicit at the type level, essentially reifying data into the type itself. One example of this is a sized data type, like arrays of a particular size.) 2021-08-12 01:11:49 I also strongly suspect that proof assistants are associated with lots of language about "data" and "values" and "size" in the context of dependent types, but the nonsensical Haskell rant uses these as associated with databases which---again, same problem 2021-08-12 00:52:46 TBH I think the Blockchain digression is actually a demonstration of the ambiguity and domain & 2021-08-12 00:50:54 It would be fun to do a joke conference based on this premise, where you have to deliver all of these with a straight face. Maybe I will make a video 2021-08-12 00:49:57 ooof this time it went all Blockchain on me https://t.co/047XbLEkcw 2021-08-12 00:46:31 @ChrSzegedy Oh, yeah, I think this is on the plate for the student in the fall, not sure if I'll still be involved then, but it's a lot more work than what we're currently doing IMO 2021-08-12 00:44:35 This prompted me to enter my job talk introduction into the InferKit demo and I'm crying it's so funny (though for very different reasons, since the ambiguity goes away when it's written, as the problem with transcripts was similar-sounding words) https://t.co/Qi4iJUx8Ld https://t.co/uEztnhzLo9 2021-08-12 00:28:34 Maybe neither of those are using anything fancy. But at the very least I think GPT-2 struggled with this a lot. Not sure about GPT-3 2021-08-12 00:23:00 Or how Google Meet always decides I'm talking about fruit and not proofs, and that when I say "lemma" I actually mean "lemons." "So then we open the fruit assistant and type this lemon" 2021-08-12 00:22:12 Like when the caption service for Zoom at my job talk for one university automatically generated the subtitle "the persistent cold cock" instead of "the proof assistant called Coq" despite everything else I was saying being within a particular domain where I'd never say "cock" 2021-08-12 00:20:30 IDK I keep thinking about the "LLMs are fine when there are ambiguities between domain-specific words and words found more generally" angle folks are saying in contrast with my experience with basically every NL tool I've ever used when I try to give it a mathematical concept 2021-08-12 00:09:19 nice.jpg https://t.co/Es3ksR1ZOc 2021-08-11 23:14:46 @karlhigley Martin-Löf 2021-08-11 23:05:33 omg haha ML ambiguity is strong here, I meant machine learning, not metalanguage but this is cute and funny 2021-08-11 22:40:28 "Fundamental" here means not something we can overcome with a lot of research 2021-08-11 22:39:32 Justification of answers is interesting to me, I'm not sure what I believe, but I do know I care 2021-08-11 22:38:18 Barriers to composable ML are: 2021-08-11 22:33:25 @ChrSzegedy ughghgh when are we going to collaborate I can never stop talking to you about this. I even impulsively filled out a form to visit next summer, though I have no idea if I can do that if I'm already faculty for a year 2021-08-11 22:30:20 @ChrSzegedy I'll have codex access soonish and will play with it, but I suspect there isn't enough Coq proof data for it to be useful at all 2021-08-11 22:29:38 @ChrSzegedy Is it ever/sometimes/often the case that incremental improvement x improves model M, as does incremental improvement y, but incremental improvements x and y together make M worse? PL folks will also care about those cases a lot, if so 2021-08-11 22:25:31 @ChrSzegedy And at this point, while I expect we'll outperform the existing Coq proof synthesis work on both tasks we care about, I think my personal goal is to play more with the PL lens on this, since exactly when things fail to compose or incrementally improve leads to new PL research 2021-08-11 22:24:26 @ChrSzegedy I think the target audience (PL folks) is going to be quite interested in the questions "what are the experimental results for different kinds of composition of these models" and "what are incremental changes we can make to improve performance of each model" 2021-08-11 22:20:47 @ChrSzegedy But yeah, I've heard the "composable ML is currently a myth" thing before 2021-08-11 22:19:48 @ChrSzegedy Uh, "no good" sounds mean as a professor, I mean "there are a few preliminary models but folks are basically just starting to experiment, so there are only a handful of papers on this to begin with, with performance I expect we can dramatically improve" 2021-08-11 22:18:26 @ChrSzegedy Yeah, that's fair, just at this point proof synthesis in Coq is at a state where there are no good models to begin with, so I'm perfectly fine with a complicated combination of things that someone else later simplifies with a single model 2021-08-11 22:13:10 @ChrSzegedy Yeah, I guess the confusing part becomes how to combine all of our models at that point---we already have two, now we'll have three? (I'm improving CoqGym to improve TacTok: https://t.co/L23fmHBFlF) 2021-08-11 22:08:26 @ChrSzegedy @BlancheMinerva Pinning these and coming back to them, just gives me useful things to think about as we think about getting fancier with identifiers 2021-08-11 22:07:42 Also compare with rolling our own 2021-08-11 22:06:18 @ChrSzegedy @BlancheMinerva Ah OK, cool, so looks like there are at least two dimensions along which to experiment: 1) what model, and 2) when and how it's actually used in the tool 2021-08-11 22:05:27 I am still learning what half of these words mean, but assume this is another dimension of the tradeoff I'll have to consider https://t.co/V7TCypWFWf 2021-08-11 22:03:01 @ChrSzegedy @BlancheMinerva I know what tuning is, the word "policy" in ML is new to me 2021-08-11 22:01:49 @ChrSzegedy @BlancheMinerva Can you explain what you mean by the latter? (Remember I've only been doing this for two weeks, and mostly learn by talking to people ) 2021-08-11 21:58:24 @ChrSzegedy @BlancheMinerva Ah, I see what you mean---the latter, I think, and I do expect that to be hard and I do have no idea what I'm doing there, so knowing what's hard about it would be helpful 2021-08-11 21:55:49 @ChrSzegedy @BlancheMinerva The question I have here is just different ways to play around with doing better than that, using standard NLP tricks 2021-08-11 21:55:16 @ChrSzegedy @BlancheMinerva Coq has no primitive datatypes, so if you erase identifiers you erase literally all type information. Encoding identifiers (only of definitions and constants, not local variables) naively, just doing the unknown trick, would shock me if it didn't dramatically improve CoqGym 2021-08-11 21:52:47 @ChrSzegedy @BlancheMinerva I am planning to reuse (at least experimentally) a model pretrained on text, that's what this thread is about 2021-08-11 21:50:05 @ChrSzegedy @BlancheMinerva But I value your opinion here so if we have the same base understanding of what I mean, I'd enjoy hearing more so that if this doesn't work, I have some understanding of why it's useless and can think of other things to try that circumvent those challenges 2021-08-11 21:48:48 Please go easy on me if I get confused, I'm a programming languages researcher who happens to do proof automation, which is just in the last couple of years starting to get some gains from neural networks, and still barely 2021-08-11 21:47:17 @BlancheMinerva Uh, that is, we won't be training own large language model for this particular use case. We have two language models for different things right now in our tool (two different programming languages), and the question here is playing around with a different third one (NL) 2021-08-11 21:45:39 @ChrSzegedy @BlancheMinerva Can you elaborate? I don't mean the model would be trained on identifiers, I just want it to be able to handle new identifiers gracefully, and I do expect done well that would improve performance given what I know about proofs 2021-08-11 21:42:52 @BlancheMinerva Definitely won't be training our own language model. I mean tradeoffs of tuning one of those very large language models trained on like the whole internet, versus something just trained on math stuff like the ones folks posted---maybe the math ones would still be large, I guess 2021-08-11 21:38:21 Also a relevant question (I do want to try both, just curious): https://t.co/d6GB57NKKr 2021-08-11 21:37:31 @BlancheMinerva Are there other tradeoffs involved wrt the choice between a LLM and an existing language model trained on a specific domain? Even just research costs and so on 2021-08-11 21:35:54 @BlancheMinerva Oh, that is my bad, I looked at the wrong folder on my desktop haha 2021-08-11 21:34:29 Obviously compare with tuned LLM to our context too, not sure which will do better in this context, and would like to see by experimental results 2021-08-11 21:31:17 @BlancheMinerva Thanks, though---I'm hoping to experiment with both 2021-08-11 21:30:24 @BlancheMinerva It's about a gig unzipped 2021-08-11 21:29:46 @BlancheMinerva Oh we two models already for proofs, just want to hook in a NL model specifically for identifiers 2021-08-11 21:28:25 @codydroux but fair, and potentially interesting for whether a LLM actually does do better 2021-08-11 21:28:05 @codydroux yeah so basically the one project formalizing Haskell in Coq will have this 2021-08-11 21:25:44 @BlancheMinerva Or do you mean the model? 2021-08-11 21:24:58 @BlancheMinerva Looks like about a gig of data? 2021-08-11 21:22:18 @BlancheMinerva I think the dataset is not very large, by most standards 2021-08-11 21:19:19 Saved Tweet with some domain-specific language models to handle new identifiers interestingly later on and experiment with: https://t.co/abIUMD5fRp 2021-08-11 21:17:26 Oh wow, very happy with multiple responses so far, will save this and come back to it once we get past the naive version. Thanks, friends! 2021-08-11 21:14:14 RT @marileezafari: @TaliaRinger "I'm depressed I have to withdraw from a marathon due to a broken bone." "Have you tried exercising?" 2021-08-11 21:13:02 @guyd33 @wellecks @kchonyc Oh, fascinating. Oh gosh so many models to play with 2021-08-11 21:06:32 I get all of this information from a proper parser after type inference, so it's quite detailed and includes a lot of implicit information, and fully qualified paths, which are also useful information for proofs but so much room for experiments 2021-08-11 21:02:11 So that's why I want such a language model. But I'm also fairly open-minded, since I have just like two weeks of machine learning experience at this point 2021-08-11 21:01:42 Likewise, if there's one just for code identifiers, and not the rest of English---probably there will be a lot of overlap 2021-08-11 21:01:15 I don't want occurrences in other contexts to cause problems. So my friend said I could tune an LLM and it should still be fine. But if there's an existing model specifically for math & 2021-08-11 20:59:41 Pretty much all interesting identifiers I encounter I can expect to be related to math or computer science in some way. It'd be really weird and rare to see one not at all related to math and computer science, because these are proofs about math and computer science 2021-08-11 20:58:53 "Ring" and "group," for example, very frequently occur outside of mathematical contexts, in really different scenarios, with no relation to one another. But basically no proofs I encounter are ever going to be using them in non-mathematical contexts 2021-08-11 20:54:58 OK so my friend pointed me to BPE tokenization and to the easier approach of just using an existing language model to handle this. Cool cool cool, just one question follows from that 2021-08-11 20:52:44 That'll be the first experiment. After that I want to get fancier, in a way that is presumably common. Like I want my model to understand that if it sees "ring" in the training data, and then someone writes proofs about "group," those proofs likely will be similar 2021-08-11 20:50:55 This includes all datatypes in the standard library, for example, like "nat." Proofs about "nat" (natural numbers) are very often going to proceed similarly, so even the naive way of frequency counting of identifiers and calling some of them unknown should get us big improvements 2021-08-11 20:48:11 I was going to get super fancy by reifying semantics as syntax in the AST before encoding goal types. But when I looked at what they're actually doing, I realized they're erasing even useful syntactic information, like the names of inductive types, constants, and modules 2021-08-11 20:47:17 I'm working on improving an existing model for neural proof synthesis in Coq. Dataset and original tool here: https://t.co/XTMGcalXQm 2021-08-11 20:45:17 Does anyone have a pre-trained language model on concepts from math and computer science? More context in thread, and also more questions that follow from that context 2021-08-11 19:59:35 @Alan_Taylor_314 Haha, I'm just joining onto follow-up work for a project that already has a publication I wasn't on, which itself builds on or evaluates against lots of other work included in the repo also from publications I wasn't on. I have exciting plans though, just engine building first 2021-08-11 19:56:43 @Alan_Taylor_314 Hammers are very general-purpose symbolic proof automation (plus some premise selection), so a lot of the neural proof synthesis work evaluates against them and in combination with them to see what each approach excels at 2021-08-11 19:54:12 @Alan_Taylor_314 Oh, OK. CoqHammer is someone else's tool, it's included in the repo for evaluation 2021-08-11 19:52:05 @krismicinski This break was really severe and included joint damage. I need to go to a podiatrist once I get to Illinois to make sure I don't need surgery or anything 2021-08-11 19:50:50 @Alan_Taylor_314 ? 2021-08-11 19:47:28 All talks about this work going forward will now have a bunch of free jokes thanks to my broken toe. Students will thank me later 2021-08-11 19:37:33 So I left a nice little comment for the amusement of anyone else who dives into this code base 2021-08-11 19:36:51 I eventually did at least get to the programming I wanted to do, just a day late and with a broken bone and joint. It's nice to see what module & https://t.co/6wj077Ao4j 2021-08-11 18:07:12 Dynamic typing is just too extreme for me. Like, how am I supposed to type while walking around like that? Of course I'm going to crash into things that way 2021-08-11 18:05:37 @ChrSzegedy wow no wonder you don't break any bones. thanks for the tip 2021-08-11 17:59:00 RT @hannelita: .@icfp_conference 2021 has a wonderful set of typed Ask me Anything (AMA) sessions! @TaliaRinger, @PierceBenjamin, @dreixel,… 2021-08-11 17:58:48 Statically typed languages really are safer. For example, I never break my toes writing OCaml code. But just a few days of Python hacking and now I'm on crutches 2021-08-11 17:11:39 OK onward: - Where should I go for a couple of weeks in late August/early September? Should be train accessible from Chicago. Keep in mind my broken toe. - What's a good spring marathon to do in or near Illinois? Should be a Boston qualifier so I can requalify for another year. 2021-08-11 15:17:15 In bright news though it doesn't hurt at rest right now anymore. Yesterday it hurt no matter what I did. But sleeping seems to have helped with at least the initial healing 2021-08-11 15:14:17 ADHD life is getting so excited about what you're doing that you negate to consider your surroundings and end up with broken bones 2021-08-11 15:11:00 Going to show up to faculty orientation next week on crutches. "Oh no, what happened?" "I was programming" 2021-08-11 15:07:29 When people talk about the ethical implications of AI, they don't talk enough about it being so much fun that you might get excited, kick a chair, and break your bones [dark humor] https://t.co/8MPrfZFtET 2021-08-11 14:55:21 @lambdatotoro In fact I have successfully twice over the last few years, but life has a way of ripping away all of your coping mechanisms on a moment's notice 2021-08-11 14:51:03 Internet bros: Oh you're depressed? Try hanging out with friends and keeping an exercise routine Me: bro 2021-08-11 14:49:40 Like wow rude 2021-08-11 14:49:32 ...Me, mid-2021: OK dep- 2021: also how about the collapse of your exercise routine, hobby, upcoming goal, and vacation all at once as you're moving across the country, leaving your friends behind, and starting a new job? In a second's notice because you got excited about code 2021-08-11 14:46:45 ...Me, late 2020: OK depression is- 2021: how about a serial harasser during your job search, using your mental illness to try to make a case that nobody should hire you 2021-08-11 14:46:15 Me, early 2020: OK depression is better 2020: how about the collapse of your social support system due to a pandemic 2021-08-11 14:27:08 @marileezafari The ones telling me to establish a regular exercise routine are the funniest, as if they can't scroll down and check what's up one Tweet below on my timeline 2021-08-11 05:44:20 @perceptophore @americanamemes Too real 2021-08-11 04:57:17 @somelaurachick @grhmc https://t.co/EDOi24Kudn 2021-08-11 04:42:55 @somelaurachick @grhmc I concussed myself once impulsively climbing a small rock wall in someone's front yard in San Diego 2021-08-11 04:24:16 I'm so depressed 2021-08-11 04:07:32 @hillelogram Bought a handcycle, will see if this and crutching around is enough until I can bike 2021-08-11 04:05:33 I was going to take a vacation to the mountains in Colorado or something and just run at altitude for a few weeks. Now I'm on crutches in a boot---what are vacations that are both friendly for broken bones & 2021-08-11 03:19:48 @Anne_Spencer_ @hillelogram The chair based workouts that is 2021-08-11 03:11:07 I took "break a leg" too literally I guess 2021-08-11 03:10:23 @Anne_Spencer_ @hillelogram Annie if you know of any please send them, those sound great 2021-08-11 02:55:18 @hillelogram Would always be grateful for food haha 2021-08-11 02:45:24 Oh the fracture extends into the joint and includes cartilage damage haha no wonder it hurts so much 2021-08-11 02:44:50 @somelaurachick @grhmc ADHD, baby 2021-08-11 01:56:50 @hillelogram Maybe I can just go for multiple mile crutch walks 2021-08-11 01:54:52 @hillelogram I really need intense cardiovascular exercise or I go crazy 2021-08-11 01:47:30 @hillelogram I mostly need help packing and moving to Illinois, and probably also some idea of other exercise I can do before I go crazy, but I don't know what to do about either 2021-08-11 01:35:24 @grhmc I got excited to write some code. Quickly rushed over to the couch to write the code. Kicked a giant metal leg of a sofa really hard 2021-08-11 01:08:42 I'm sorry but who runs 14 miles easily but then fractures a toe cleanly in half getting excited about programming 2021-08-11 01:07:05 Coding is dangerous. Too exciting, clearly. An extreme sport 2021-08-11 01:06:21 @mariaruizv Programming is dangerous, clearly 2021-08-11 00:58:27 Fracture, almost cleanly in half from the base to the joint. I hate this 2021-08-11 00:01:50 x-ray time please please please just be dislocated and not also broken 2021-08-10 23:45:25 RT @Miles_Brundage: ICYMI: https://t.co/3nv5JMji23 2021-08-10 23:45:23 RT @OpenAI: Today's live demo of Codex, our AI that translates natural language to code: https://t.co/MkpiM4W7ow 2021-08-10 23:28:00 @MrDuluc Yeah, please don't do that 2021-08-10 23:25:19 @MrDuluc I already appreciated them as I ran every day 2021-08-10 23:14:53 Going to be so much fun to move across the country alone when I can't walk or stand up 2021-08-10 23:07:14 @kay314159 Running on a broken toe is a really bad idea 2021-08-10 23:06:36 @kay314159 Middle toe 2021-08-10 23:01:38 I really hope it's only dislocated and not also broken 2021-08-10 22:36:09 Stop trying to be reasonable right now you don't understand how depressed I am about this 2021-08-10 22:35:41 @nbouscal Where??? I don't have any of those things 2021-08-10 22:26:56 @NuritPeleg It's usually 4-6 weeks 2021-08-10 22:26:08 What exercise can I do? I'll go crazy if I can't exercise 2021-08-10 22:25:47 I would rather break both arms than a toe. Running Boston was the thing I was most excited about 2021-08-10 22:25:12 Not exaggerating on the dislocated and broken thing, it is at a disturbing angle, and I can't even stand up on the non-hurt foot, because it hurts too much if I'm not resting the hurt foot at all times on something else. I'm at urgent care, and my Uber driver wheeled me in 2021-08-10 22:23:38 @UlrichJunker Like, the non hurt foot. I can't stand on the non hurt foot, because it hurts too much if I'm not resting the hurt foot on something at all times 2021-08-10 22:22:52 @UlrichJunker It's so painful I can't even stand on one foot. I'm at urgent care. I'm not exaggerating my Uber driver had to wheel me in 2021-08-10 22:16:22 And then I kicked a fucking chair. I hate life 2021-08-10 22:16:03 I was unable to do much exercise all pandemic and somehow got to the point of running 14 miles again last weekend and I was so excited 2021-08-10 22:15:35 I was going to spend the two months before my faculty job officially starts training for the Boston marathon and then running it in October. Was going to take a train to the mountains somewhere and just spend some time running 2021-08-10 22:09:55 I'm so heartbroken please someone make this less sad 2021-08-10 22:03:49 @nbouscal I think it's both 2021-08-10 22:02:57 @nbouscal I hope that's all it is. I'm at urgent care. It hurts too much to even touch it 2021-08-10 22:02:15 What is the point of literally anything if you can just fuck it all up in a moment by getting excited and kicking a chair 2021-08-10 22:00:03 I was excited to write some code, so I took my shoes off, jumped over to sit on the couch and kicked a chair on the way and, well, my toe is visibly dislocated, probably broken, and I can't walk. No Boston for me, I guess. I feel such a deep despair right now 2021-08-10 21:08:30 That was cool. If you told my high school teachers I'd be a CS professor at Illinois one day they'd probably laugh at you. I got detention a lot, scraped by with Bs and Cs, and hated every second of high school. 2021-08-10 21:07:11 Also though, I really started to believe in myself, and not just view myself as a total screwup who somehow does everything wrong always, smart but with wasted potential, lazy, whatever else people had been telling me my whole life. I like, I finally got to be someone! 2021-08-10 21:05:26 With medication and the language to describe my experiences, and the corresponding community built around that language, my secondary depression and anxiety went away completely. (I have depression now but I was depression-free for 9 years 2021-08-10 21:03:18 @mustach_io Omg ahhaha 2021-08-10 21:02:57 In other words, I'd managed to mask it really well for my whole life, so she could see these traits in me but they didn't seem that intense. The internal experience was very different, and the masking was exhausting. 2021-08-10 21:01:49 Then we saw two curves for different symptoms in the end. What was absolutely fascinating to me was that the curves had the exact same shape---our answers were consistent in relative intensity---but only my personal answers were above the line for clinical significance. 2021-08-10 21:01:01 This and a whole lot more. In any case, I did a lot of tests. One I remember clearly: My mom and I both filled out long symptom questionnaires for me, me from my perspective and her from her perspective. 2021-08-10 21:00:04 Oh and I lost literally everything I came into contact with, somehow. 2021-08-10 20:59:13 At nights I'd find myself extremely hyperactive. I always felt like I had to be moving. I made a lot of embarrassing mistakes on exams, really careless errors, even when I knew the material perfectly. I self-medicated with too much caffeine. 2021-08-10 20:57:56 I was driven to get diagnosed because of secondary depression and anxiety. Basically since I couldn't learn in lectures, I'd go to them when mandatory, but then also teach myself everything after. It took me twice as long and I had no friends, so I got depressed. 2021-08-10 20:56:15 Interesting thread. I was diagnosed at age 19. I went to an ADHD specialist and did a huge array of tests. It was really fascinating and kind of fun---I like learning about my brain. https://t.co/ePE3VFVNXG 2021-08-10 20:52:25 RT @worldwise001: @mustach_io @ND_in_CS @AcademicChatter @MentalHRetweet Even without meds, I think just knowing is half the battle so you… 2021-08-10 20:49:11 @mustach_io @worldwise001 I forget, are you on the @ND_in_CS Slack? 2021-08-10 17:55:30 RT @IgorBrigadir: "Every time I fire a linguist, the performance of the speech recognizer goes up" ...because we started outsourcing what… 2021-08-10 15:22:42 Ahhhh omg I feel excited about my upcoming life in Illinois for the first time after seeing all of the other junior faculty in a video meeting 2021-08-10 15:12:05 Shoutouts to @SaugataGhose @camPossible @charith_mendis @hanzhao_ml I'm so excited to see all of you once I'm in town 2021-08-10 15:04:04 I really like my cohort this is going to be really cool 2021-08-10 15:03:39 In a remote business meeting for UIUC, and sure it started at 7AM in Seattle and I'm half asleep but it's hitting me that I have a really cool cohort of like 7 other junior faculty! And also an office! 4218 (I haven't been there yet) 2021-08-10 07:34:58 I forgot alt text, so it is below. This is an infographic from Healthline. It reads: Factors that increase ADHD risk: - 4x risk: Children from primarily English-speaking homes are at 4x the risk - 3x risk: Males are at 3x the risk of females 2021-08-10 07:33:35 @SashaTheShark @get_altText Sorry, should say: Factors that increase ADHD risk: - 4x risk: Children from primarily English-speaking homes are at 4x the risk - 3x risk: Males are at 3x the risk of females 2021-08-10 07:31:42 @silvascientist Shit 2021-08-10 07:29:05 Speaking English causes ADHD. Spread the word 2021-08-10 07:27:46 "You may be more likely to be diagnosed with ADHD if you're in the group of people considered and studied when the diagnostic criteria were developed" haha well gee I wonder why https://t.co/WYlgnct9GB 2021-08-10 05:59:14 I think he should stream Cardcaptor Sakura from space 2021-08-10 05:56:23 Also can we talk about how @elonmusk follows only like 100 accounts but one of them is @ccsakuras? I forgive him for anything else he's ever done honestly https://t.co/VpkWzb85Ix 2021-08-10 05:48:43 @trochee @elonmusk When you forget that last iteration around you chose the seed 420 for the simulation, and choose it again next time around, and so on forever in an infinite loop https://t.co/juSkp2Bmwa 2021-08-10 05:42:47 Dude @elonmusk what should the next seed for the simulation be 2021-08-10 05:35:38 Let's pick a different seed next time 2021-08-10 05:35:23 This gets worse every sentence. Can we reset the simulation please https://t.co/7cuJsHgiK6 2021-08-10 05:22:50 I wish https://t.co/xeT8u99Bje 2021-08-10 05:08:49 @MrDuluc 40% overlap. To be honest I will not be surprised if one day folks decide that autism, ADHD, and SPD are all strongly related and not really discrete entities 2021-08-10 05:06:40 @MrDuluc Seems like it is an input filtering issue (same reason I need captions on TV or Zoom calls): https://t.co/h6dppDCzk1 2021-08-10 05:03:11 @MrDuluc Also sounds that are off in one location rather than surrounding me 2021-08-10 05:02:54 @MrDuluc I just really hate pitches above a certain frequency, when not balanced by adequate bass 2021-08-10 05:00:46 Sensory overload is also why I avoid basically all in-person stores 2021-08-10 04:59:11 @MrDuluc This specifically, but I wasn't diagnosed with ADHD until much later (age 19): https://t.co/Dqo5I4drdI 2021-08-10 04:57:45 @MrDuluc IDK, I was diagnosed with "sensory issues" as a kid which is comorbid but IDK if specifically ADHD 2021-08-10 04:43:14 It is actually painful to listen to, and I hate when I hear sounds playing out of other people's phones when I'm nearby too 2021-08-10 04:42:32 I realized today one of the reasons I dread phone calls and avoid them at all costs is that I can't stand the tinny noise the phone makes in my ear 2021-08-10 04:33:17 @silvascientist @matthematician That sounds great honestly 2021-08-10 04:28:42 In the longer run I'mma probably look up some semantic information in the environment for all definitions and make that explicit and maybe even do some of my fancy relation inference but it would be nice to use the text better too 2021-08-10 04:26:23 It would be nice eventually if my model could have some sense of "mathematical distance" between unseen identifiers and seen identifiers, without radically expanding my vocabulary, and without relying on a whole-ass LLM https://t.co/HNqzI4Y10n 2021-08-10 04:19:49 @johnregehr @wilbowma @samth OK guys I've got this I can be a professor 2021-08-10 04:19:26 @johnregehr @wilbowma @samth Yeah having 90 minutes to check my email every week, and then chime in to the discussions I find interesting sounds doable 2021-08-10 04:18:09 @wilbowma @johnregehr @samth OK I'll survive. It's like undergrad when we had to go to class 2021-08-10 04:17:12 @johnregehr @samth Can I just do other things when it's boring and irrelevant 2021-08-10 04:16:41 @johnregehr @samth Good 2021-08-10 04:15:49 @johnregehr @samth Our meeting is 90 minutes. If I just get up and leave will people dislike me 2021-08-10 04:15:06 @johnregehr @samth All of you NT folks scare me hahahaha 2021-08-10 04:14:35 @johnregehr @samth I can't pay attention in 90 minute meetings. Maybe 30 minutes if extremely focused and sticking to agenda, 60 if extraordinarily focused and sticking to agenda and action oriented 2021-08-10 04:13:36 @johnregehr @samth And then only the people who commit to certain things should have to meet to talk about them 2021-08-10 04:12:50 @johnregehr @samth Or have a Slack or something and just notify people when people need to do things, like modern civilized people 2021-08-10 04:11:26 @johnregehr @samth Monthly sounds reasonable to me. Maybe twice a month 2021-08-10 04:11:02 @johnregehr @samth Can't they just like email everyone 2021-08-10 04:10:29 @wilbowma @samth There is probably a better medium than a weekly mandatory meeting 2021-08-10 03:59:23 @jameskjx So why do we have to go to them 2021-08-10 03:51:46 For anyone reading this though my NT friend brought me a mask and I'm on my way home safely 2021-08-10 03:50:15 Second worst thing I've accidentally dumped in those trash cans. Worst was my university ID hahahahaha 2021-08-10 03:49:16 It was a perfectly good N95 I accidentally threw out today, too. And in a trash can you can't open 2021-08-10 03:44:55 @silvascientist Luckily there are other academics here at 8:40 PM hahahahahaha 2021-08-10 03:44:01 @silvascientist It's 8:40 PM and most of the campus is closed during the day to begin with, let alone at 8:40 PM 2021-08-10 03:41:11 I think my friend is bringing a mask maybe? How would I survive with neurotypicals lol 2021-08-10 03:40:06 Oh, hey, so I had an ADHD moment and when I was afraid of a wasp, I ran away and threw my food out but accidentally also threw out my mask. Anyways, I'm now stuck at UW and can't get home, since I took an Uber here 2021-08-10 03:35:57 @fips00_ riiiiiiiiight 2021-08-10 02:04:36 @arizvi0 @DavitSoselia_ I mostly build automation for programming languages. You really can't beat OCaml for compiler development IMO 2021-08-10 01:55:34 RT @yminsky: @TaliaRinger @ChrSzegedy Modular implicits, should it ever land, would make this nicer, in that you could write "print foo", a… 2021-08-10 01:55:30 @yminsky @ChrSzegedy As long as error messages are informative 2021-08-10 01:54:52 @yminsky @ChrSzegedy That would be _wonderful_ 2021-08-10 01:54:35 @yminsky @ChrSzegedy That's fair, though there's probably a way to hack it using the static type information if you have access to the compiler itself 2021-08-10 01:51:56 @yminsky @ChrSzegedy I really just want syntax that's like "print foo" and disregards static type information, but yells at me if I actually use it for anything but debugging 2021-08-10 01:26:14 @DavitSoselia_ Type system is just expressive enough to be useful, but not oppressive for day-to-day development of automation. Syntax is simple. Functional by default, but lets me break out of the functional world if I want to. Pattern matching and recursion are very nice, good HOFs 2021-08-10 01:10:12 @ChrSzegedy Syntax is awful, but idea is nice 2021-08-10 01:07:13 It kind of makes me want an unsafe_print in every language I use for debugging that just ignores all static type information 2021-08-10 01:06:29 I have to say, basically everything about OCaml (the language I usually use) is better than Python (the language I'm currently using) except for one thing: It's really easy to print things in Python. And weirdly that's actually a really nice feature for a language to have 2021-08-10 01:02:51 I understand why this is fun. Folks have already written proofs, and proofs are inherently beautiful. When I mine proofs I'm just mining everyone else's beauty. So even if it's an ugly python script, there's inherent beauty I get to see in the end 2021-08-10 00:58:19 also kind of cute, it will probably capture some bits of style, for example since tactics will differ dramatically between users of SSReflect and normal people, since like: 'ssrnat': 3269, 'ssreflect': 64135, 'mathcomp': 190776, ... 2021-08-10 00:52:10 oh man I have an idea 2021-08-10 00:51:32 bro these are so semantically meaningful, like quick sample: 'algebra': 63997, 'GRing': 59906, 'Ring': 10528, 'sort': 53068, 'ringType': 20509, 'bool': 12404, 'eqtype': 5711, 'eq_op': 1982, 'eqType': 2848, 'mul': 3574, 'one': 2097, 'andb': 820, 'inPhantom': 611, 'Zmodule': 4032 2021-08-10 00:48:01 why say "import libs" when you can say "own the libs" 2021-08-10 00:34:33 to stay true to my Seattle intellectual roots, when you choose the strength of your guarantee, the tool prompts you for the "spice level" 2021-08-10 00:26:09 strength of guarantee for each claim denoted by number of emojis 2021-08-10 00:22:52 RT @johnregehr: @TaliaRinger every upcoming talia abstract/proposal summary: we envision a future in which formal verification is mashed po… 2021-08-10 00:20:43 @johnregehr I think it can be taco bell hot sauce testing with the right interface and automation, and I'm interested in building that 2021-08-10 00:18:25 Talia things no context https://t.co/eVK8d5ndDr 2021-08-09 23:29:39 IDK a reasonable threshold to start with for known versus unknown 2021-08-09 23:26:34 https://t.co/YGZbQgesM2 2021-08-09 23:26:16 weeeee https://t.co/QGMnXsBTNr 2021-08-09 21:51:42 Naturally I used this as an excuse to order boba 2021-08-09 21:51:25 UW campus had enough people to feel kind of enjoyable to be at again sometime a few weeks ago, but now with Delta-driven waves, there is basically nobody else here again. It's kind of lonely, and it's my last week to see my UW friends before I move 2021-08-09 21:46:33 @matthematician Bah 2021-08-09 21:43:27 @junosz Why do people do this to themselves 2021-08-09 21:31:04 what actually happens at these meetings 2021-08-09 21:30:10 faculty: we're so busy also faculty: let's have weekly 90 minute faculty meetings I'm confused 2021-08-09 21:08:29 @MarisaVeryMoe It's actually really fun. Enriching proof data and experimenting with how it improves performance of a model for both decompilation (mix of NN and traditional decompiler) and vanilla proof synthesis 2021-08-09 21:04:23 @ChrSzegedy And if not necessary (plausible), certainly efficient and helpful 2021-08-09 21:02:44 @MarisaVeryMoe Yes, though it can be very dangerous if the benchmark is not actually a good one 2021-08-09 21:01:29 @ChrSzegedy I'm sure the truth is somewhere in between, but it makes collaboration difficult to feel that, as a domain expert, your contributions---often a necessary stepping stone to something better---are not valued 2021-08-09 20:59:52 @MarisaVeryMoe Just based on the way I see ML folks talking about other work that followed similar patterns 2021-08-09 20:59:16 @MarisaVeryMoe Right now I think (we will see) we can improve on existing tools on existing benchmarks, it's just that then when the work is superceded later on, I assume folks will forget our role forever 2021-08-09 20:56:57 @MarisaVeryMoe That's one good thing about the ML community, honestly 2021-08-09 20:51:38 Also sometimes this just fundamentally changes the data that is collected to begin with, like forever, and that's powerful too, but not appreciated enough at all 2021-08-09 20:48:51 But lots of my friends in graphics and NLP all say it's like this over and over again so I don't really understand why crediting the important role of domain experts in that process is so hard 2021-08-09 20:48:06 I feel bad even participating in this process right now, knowing that once folks learn all the things I enrich proof data with, probably my role will be forever forgotten 2021-08-09 20:45:19 ML folks: BUT DO WE EVEN NEED DOMAIN EXPERTS FOR THIS NEW TASK (yes, yes you do, their involvement is an important part of the above process, and forgetting to credit them doesn't change that) 2021-08-09 20:44:15 ML folks: here is a nice architecture Domain experts: your data doesn't include even a proxy for the thing you want, here's better data ML folks: cool, now using that insight, let's learn that going forward ML folks: haha never needed those domain experts amirite 2021-08-09 20:37:50 @mattecapu @bgavran3 @Anarchia45 @fabgenovese @_julesh_ Yeah, but the Bigger is Better folks seem to think the way forward to fix this is to just train on *more* kinds of data, so you get multisensory experience. It may work, I just don't think it's necessary, and I think it's wasteful if there's a better way 2021-08-09 20:34:23 @bgavran3 @mattecapu @Anarchia45 @fabgenovese @_julesh_ I don't know the answer, but I'd postulate it's a hell of a lot less than "the entire internet" 2021-08-09 20:28:25 Good AI should be able to ask the right people the right questions to learn knowledge we forget we had to learn, and rarely encode. Important for programming & https://t.co/L2p6ipW4aa 2021-08-09 20:25:59 @JulesJacobs5 @bgavran3 @Anarchia45 @fabgenovese @mattecapu @_julesh_ This is why I think asking the right people the right questions is going to be one of the most important things good AI can do in the future, in order to interface well with humans and our knowledge 2021-08-09 20:24:58 @JulesJacobs5 @bgavran3 @Anarchia45 @fabgenovese @mattecapu @_julesh_ I think the problem is that it often comes up as-needed. Like once someone tells you, you forget you needed to learn it, so you just consider it "tribal knowledge" or whatever. And you only know to tell people once they ask 2021-08-09 20:22:07 @JulesJacobs5 @bgavran3 @Anarchia45 @fabgenovese @mattecapu @_julesh_ Talking to people. Video calls, in person---things you could feasibly record and mine, but I'm sure as hell not giving anyone permission to do that (not that this seems to matter to anyone these days) 2021-08-09 20:21:03 @bgavran3 @Anarchia45 @fabgenovese @mattecapu @_julesh_ Maybe we operate in different circles, but what I've seen online IMO is too optimistic about the progress made so far, and overclaims dramatically 2021-08-08 03:54:14 I guess that's it. 2021-08-08 03:53:26 But it's there, and it's overflowing all the time these days, and I only feel good when I'm doing something to make the world better, or to make my relationships with other people better, or to make my community better. 2021-08-08 03:51:32 Maybe these things are hard to see in me because I present myself like most Israelis, most of the time. I try to have a tough and prickly exterior, like a cactus, to keep the soft part just a bit obscured. If I help you, chances are I make an excuse about how it was easy. 2021-08-08 03:50:39 I'm just so sad. There's pain all over the internet, but if I close the internet and try to just live my life, there's pain there too. And in both cases, because I can't control how much I care about other people, and my empathy and love and caring just spin out of control. 2021-08-08 03:49:01 For two weeks, every time I see a friend, it's the last time I'll see them in god knows how long. Every day is a heartbreaking goodbye. I can't handle it. I feel so crushed every time. 2021-08-08 03:48:05 The second part: I've lived in Seattle for nine years. Nine whole years. Three at Amazon, and six in graduate school. I had a really rough time in Seattle, lost everything, and the rebuilt my life over the last few years---really, really rebuilt a life. And now I'm saying bye. 2021-08-08 03:47:23 That's the first part of what I'm feeling. 2021-08-08 03:46:11 Well, that's about where I found myself on Thursday. And I thought maybe if I closed Twitter, I'd stop caring so much about these people. But it didn't work. I still care a lot about them. And the pain all around doesn't go away when I just close my eyes and try not to see it. 2021-08-08 03:42:23 Really bizarre ethical dilemmas arise from this. Like if you help someone before you know they're a bad actor, and then they go and hurt someone else, are you complicit in that hurt? Do you then have a responsibility to make sure that does not happen? 2021-08-08 03:41:01 Worse, and especially on Twitter, often bad actors come along and intervene and make things much worse for everyone. They do the exact opposite of what I'm doing. They just cause suffering. Some enjoy it. Some can't help it. But they're bad, and they're everywhere. 2021-08-08 03:40:02 Occasionally, this breaks me. Like, I just can't quite make it to the finish line alright. 2021-08-08 03:39:09 But ultimately I just want the suffering to end. I want things to be better. If it's not possible, I'll give up 2021-08-08 03:36:29 When I get locked on a problem, be it research or starting a mentoring program or whatever, I am persistent, creative, thoughtful, dedicated, and obsessive. I'm like that when I'm trying to help people too. So I actually honestly do pretty well. Which makes it rewarding. 2021-08-08 03:34:39 People tell me I can't help others if I can't help myself first. But I've always been overly empathetic, and so it's hard for me to understand what that even means or how people manage to do that---if you see other people suffering, do you not just feel immense pain yourself? 2021-08-08 03:34:38 Some of you seem to avoid this by avoiding caring too much about people on Twitter. You view other people as sort of abstract ideas, and the conflicts between them become something like a soap opera. I've somehow lost my ability to do this, not that I even think it's good to do. 2021-08-08 03:34:37 I'll still keep Twitter off of my phone for now, just to force some separation of concerns. But in any case. There are two things going on. The first has a lot to do with Twitter: People on this platform can be really mean. They can be dangerous. They can be scary. 2021-08-08 03:34:36 Twitter break is not helping my mental health. So instead I'm going to do something much more in character: I'm going to write about what I'm feeling. To all of you. I'll withhold most details, but the high-level feelings may be relatable, and maybe you'll have ideas. 2021-08-08 00:19:16 OK back to a Twitter break 2021-08-08 00:19:01 I love watching all of the world-class runners collapse after the marathon just like I do every single time 2021-08-08 00:02:03 Also who is it in the race who's about to get his masters in AI? lol nice 2021-08-08 00:01:32 ehhh maybe not 2021-08-07 23:59:30 I bet he's going to do it 2021-08-07 23:59:17 Can Osako catch the chase pack??? 2021-08-07 23:46:19 Watching Kipchoge race is always unreal. Like his form is just so incredibly beautiful and efficient, it looks almost fake, like nobody can actually run like that 2021-08-07 23:36:14 Rupp looks to be hurting. Kipchoge looks like he's out for a leisurely morning jog. Freaking GOAT 2021-08-07 23:23:48 Allow me to interrupt my Twitter break for a moment to let you know that you should absolutely be watching this epic battle with Kipchoge, Rupp, and lots of other amazing runners. Set a VPN to Canada and enjoy, no account needed: https://t.co/BxbRoj4ur3 2021-08-07 11:48:38 Cafiac fix 2021-07-12 09:48:38 https://t.co/ZK8MTsS5tG 2021-07-12 09:48:11 https://t.co/qEl2e4ooRF 2021-07-12 09:47:49 Plus a million research articles I haven't read in a lot of detail, so I don't feel qualified to curate yet 2021-07-12 09:46:33 https://t.co/NL2sCMC7Lm 2021-07-12 09:46:07 More on emotional regulation and ADHD: https://t.co/tHe24DZVgI 2021-07-12 09:41:38 (All of this inspired by a video I just saw) 2021-07-12 09:40:56 It'd also be nice in the dialogue to accept that "controlling your feelings" is a nearly impossible task for a substantial portion of the population. And stop talking about it like it's this easy thing we can just learn 2021-07-12 09:39:54 Only when my tears are dangerous, though. In other situations I think it's fine to cry in public 2021-07-12 09:39:22 But it'd be nice to talk a bit about strategies for ADHD folks. Like in my case if my tears could make things much worse, I know I can't stop crying, but I can leave the room, so I just excuse myself and come back after a good cry 2021-07-12 09:38:13 And any rational person when confronted with something potentially violent and dangerous would probably do well to flee it or fight it 2021-07-12 09:36:59 Of course when it's violent or manipulative or at the least negligent in a lot of people, but a literal physical inability to control feelings the same way rooted in brain chemistry and structure in others, it's hard to know which is happening 2021-07-12 09:35:18 I think NT white women who can control tears a lot may use tears manipulatively and violently, but I think for ADHD white women that level of emotional regulation to even use feelings to manipulate people would probably not be possible 2021-07-12 09:34:00 For men who are often socialized to hide sadness, this often manifests as rage, blame, and violence. But when that isn't true, we might feel a bit sad for whatever reason and just start crying because there literally isn't the brain functionality to intervene & 2021-07-12 09:32:07 Honestly the dialogue about white tears, while fully valid, has a bit of an ableist angle I'm not into. Folks with ADHD largely lack the "let me stop myself from expressing the thing I'm feeling right now" brain thing all of you NT folks have 2021-07-12 09:00:06 RT @JessicaCalarco: "Other countries have social safety nets. The U.S. has women." And apparently we've decided "this is fine." https://t… 2021-07-12 08:54:35 Huge! https://t.co/3S3QVcndeE 2021-07-12 07:39:43 It was a lot of fun, and gave me a sense of whether I'd enjoy working with those people, too! 10/10 good interview process 2021-07-12 07:37:46 If I started a company, I would mimic this interview process for sure 2021-07-12 07:37:26 Pretty much the entire interview was building on my code. I really liked it because it mimicked the day-to-day experience of being a software engineer. Adapting code to changes, explaining decisions to other engineers, taking feedback on the fly 2021-07-12 07:36:35 I also had to explain my code and design choices to other team members 2021-07-12 07:35:33 The engineer would also suggest things and I'd take those into account 2021-07-12 07:35:17 For example, I met with a designer and a software engineer. And the designer said, "hey, I'd really like to extend this with these new features, to get something that looks like this." And then I had to talk to the engineer and explain how I'd do that 2021-07-12 07:33:59 The cool part was what happened after. My code was good enough that they invited me in person. But in person they didn't ask me a bunch of random questions. They actually built on that task 2021-07-12 07:33:20 This was a startup in Seattle, and I honestly don't even remember the name of the startup. But the interview started with an at-home software engineering task, one that took about an evening to complete. A small, confined project 2021-07-12 07:32:11 And I didn't even get the job (I was about to go to grad school in a year and was transparent about it, so I suspect that's why). So it's not about doing well or poorly 2021-07-12 07:31:33 I see a lot of talk about bad software engineer interviews here, so I want to talk about the best software engineer interview I've had 2021-07-12 07:22:10 RT @DylanTweetin: Just a couple guys being dudes who could literally end world hunger but play space games instead https://t.co/MKArlbwumg 2021-07-12 07:12:09 RT @Haaretz: אי שם בנס ציונה מגישים מנות עוף, שאף תרנגולת לא מתה עבורן https://t.co/7ctgLHmqks https://t.co/r3D6VVYmdb 2021-07-12 06:58:49 I can lie with enough effort, but I have to be convinced it is just 2021-07-12 06:58:18 I think neurotypicals get so used to masking bad thoughts that they often do not bother to introspect and learn and correct the thoughts themselves. I don't have that option 2021-07-12 06:56:47 Like 6+ years ago I made a transphobic comment because I thought it and didn't know better, and rather than just stop saying the thing, I had to learn why it was transphobic, inspect why I said it, and teach myself truth that isn't transphobic instead. So I never have the thought 2021-07-12 06:54:32 I kind of like this part, though. Because it means to become a better person I can't just mask my thoughts. I have to actually learn and rewire the thoughts themselves. And I think that's nice, it forces me to get to the root cause 2021-07-12 06:48:07 "You're so genuine" thanks it's called executive dysfunction https://t.co/OzwwcFTxMY 2021-07-11 20:58:34 RT @vboykis: “AI will never produce anything usefu—“ 2021-07-11 19:53:58 @sol_foxie This game is so good 2021-07-11 18:17:32 Omg my cousins are the best https://t.co/OqRcsKLTAd 2021-07-11 18:09:58 The whole family is so musical https://t.co/E4xamEJ9TB 2021-07-11 17:56:57 כל המשפחה שלי בארץ פה עכשיו וזה ממש unreal 2021-07-11 17:55:44 בר מצווה של בן דוד שלי שגיא. מזל טוב https://t.co/DbA0hFJ1Mi 2021-07-11 13:13:13 RT @Nature: “This heatwave would have been virtually impossible without the influence of human-caused climate change.” https://t.co/EbqvNmr… 2021-07-11 12:15:14 RT @iLemmon: Being kind is so much more important than being nice. https://t.co/dw8S8uQvoy 2021-07-11 10:33:15 Same https://t.co/36TclM97Kj 2021-07-11 10:26:34 @dawnd That sounds like a threat and there's no excuse for anyone to ever send you something like that 2021-07-11 09:54:38 I just can't wait until a time in life when I no longer have to drive places to have some strangers stick things in my nose 2021-07-11 08:20:46 RT @Basketballasart: https://t.co/BrwAEA6SSV 2021-07-11 08:05:43 @FreyaHolmer Yes 2021-07-10 19:37:40 @emilymbender @csdoctorsister Whatever at least every time he talks he makes a fool of himself, that should happen here too 2021-07-10 19:34:58 @emilymbender @csdoctorsister How do they not pick up on Pedro being a malicious actor only ever in this to make people angry because he enjoys that experience 2021-07-10 19:32:51 Yiiiiiikes holy shit https://t.co/81QiR9mxfy 2021-07-10 19:32:27 @emilymbender @csdoctorsister What the actual fuck 2021-07-10 19:29:00 RT @rob_rix: (\_/) ( •_•) / > (\_/) ( •_•) / > (\_/) ( •_•) / > (\_/) ( •_•) /… 2021-07-10 18:21:42 News right now is literally giving the jellyfish forecast https://t.co/QSYQ4Oc53b 2021-07-10 18:06:01 RT @gunsnrosesgirl3: These are the salt glaciers of Iran Million of years ago, the Persian Gulf was a much larger body of water, as the w… 2021-07-10 17:15:23 @OwoTizusa Absolutely 2021-07-10 17:13:10 RT @arithisandthat: A podcast where parents try to explain what their adult children do for a living 2021-07-10 15:07:47 RT @JihyunYun2nd: I often think about the phrasing “being reduced to tears” & 2021-07-10 14:21:40 @jjcarett2 Just the marriage haha 2021-07-10 14:16:19 @jjcarett2 I have neither a marriage nor millions of dollars lol 2021-07-10 13:50:25 Opportunity cost lol 2021-07-10 13:48:51 My friends who stayed at Amazon while I went to grad school must be filthy rich by now (I was there 2012-2015) https://t.co/dONLRkCMdt 2021-07-10 13:41:54 RT @LauraViglioni: Be gay, do monads https://t.co/PvddObN2oA 2021-07-10 13:13:18 @joshuagrochow @aaronclauset https://t.co/MEuxUn9XJX 2021-07-10 13:11:59 @matmagnani @swmccue @aaronclauset @DynamicsSIAM But that's not currently true, and you don't change systems by just changing how you talk about them 2021-07-10 13:10:19 @aaronclauset Weird take IMO. The gamification exists because external systems (graduation, hiring, tenure) reward this kind of success. If you don't say congratulations then you maintain the same incentives while also just not being kind to people who succeed within the system 2021-07-10 12:36:57 RT @gro_tsen: “a masterpiece produced by neural networks” https://t.co/pRqBSGIF80 2021-07-10 12:24:51 Gee thanks 2021-07-10 12:24:40 Nobody: Not a soul: My depression: nobody loves you and you're never going to get married 2021-07-10 12:15:31 We were at a reform synagogue in Israel that let folks film, take photos, and so on. It was honestly really nice. 2021-07-10 12:06:06 My aunt and uncle in Israel are professional musicians. Today was the first part of my youngest cousin's Bar Mitzvah---check out this absolutely magical rendition of Adon Olam. https://t.co/sLqouGd3RJ 2021-07-10 11:34:38 RT @KimZetter: Trains in Iran delayed/canceled due to cyberattacks today, resulting in "unprecedented chaos" at stations. Hackers posted n… 2021-07-10 11:12:20 @PTOOP I think that's nice, though. Because the number of undecidable problems is depressing. But I think it's less depressing given semidecidability 2021-07-10 08:50:21 @xiao_ted Oh man I still have time to rename "mixed-methods verification" to "semiformal verification" 2021-07-10 08:41:36 (This proof assumes every function either halts or does not halt, which isn't a fair assumption in every logic.) 2021-07-10 08:40:22 Thanks, Copilot, for the cute lesson on semidecidability. 2021-07-10 08:39:43 For halt itself, this silly. But pretty much all interesting undecidable problems I know of are semidecidable. The world looks a little less dark, I think, when you realize you can write semidecision procedures, and rely on the human to give up when things are taking too long. 2021-07-10 08:37:01 Fix arbitrary f. Let f halt. Then function_halts halts and returns true. Let f not halt. Then function_halts does not halt. Thus function_halts is a semidecision procedure for the halting problem. 2021-07-10 08:27:48 Proof of semidecidability https://t.co/iwiC7HrAUT 2021-07-10 06:56:16 @silvascientist One time he tried to say "lollypop" but instead said "poppylop" and my sister and I did not let him live that one down 2021-07-10 06:54:53 I want to add that my dad cannot pronounce the word "jellyfish" in English, so he calls them "jillyjells" 2021-07-10 06:42:53 I didn't step on any, but I did step on a lot of broken shells while avoiding jellyfish lol 2021-07-10 06:42:32 Big, big jellyfish, too https://t.co/dz8JE8NRiU 2021-07-10 06:38:14 This pandemic sucks, but I'm glad all of you are finally learning your Greek alphabet! Now programming languages papers will be easier to read 2021-07-10 06:36:42 RT @obeid_ro: Good God. https://t.co/OTyx7v3t5b 2021-07-10 06:35:26 I ran 4 miles barefoot on the beach yesterday in Ashkelon, and it was great, but also an obstacle course of seashells and jellyfish. My feet hurt a lot today 2021-07-10 06:08:34 RT @Rainmaker1973: Marauder ants have a special class of big ants that they ride like a bus to work, or into battle like elephant cavalry.… 2021-07-10 05:59:33 RT @CpuGalaxy: This is how a 1“ HDD looks inside. A small project I did once to put this disassembled Microdrive into resin. https://t.co… 2021-07-10 05:51:52 RT @pradnelluru: @Symbo1ics There needs to be a new Therac 25 paper every year, for the last year’s worst software errors with detailed bre… 2021-07-10 05:43:20 RT @ilya_poet: In Tretikovsky Gallery there is a painting that’s called “Grandmother Alisya was a good person” And at the end of the day… 2021-07-10 00:49:03 RT @Symbo1ics: Therac-25 wasn't a one-off and it is a bad teaching example. As I've said for years: there are thousands of Theracs every… 2021-07-10 00:45:58 RT @SegoliaFlak: @Symbo1ics I'm reminded of this particular XKCD https://t.co/kJDgCKZzaX https://t.co/f4N4WQgOIe 2021-07-10 00:35:01 Too anxious to sleep right now, weirdly. But had a 3AM Kinder Egg so not all is lost 2021-07-09 21:19:17 @hankgreen Basically though, bias can be introduced, combatted, parroted, or amplified at any point in the pipeline. 2021-07-09 21:04:35 @hankgreen This is not quite correct. Bias shows up in a lot of very subtle ways, and learning algorithms can amplify biases from the data or can be designed to deliberately combat them. This conversation happens roughly every 60 seconds on Ethical AI Twitter. 2021-07-09 20:33:24 @bakirtzisg MIT still rejected me 2021-07-09 20:31:26 Will update CV now to say "probably could have gone to MIT or something" 2021-07-09 20:30:14 ADHD specialist at age 19 was like "yep this is a classic undiagnosed case for a gifted woman with ADHD, you probably could have gotten into MIT or something had this been diagnosed earlier" haha thanks doctor 2021-07-09 20:28:31 Doctors and teachers growing up were like "you lose every pencil, can't stop talking, cry in front of people, get detention for everything, never do any homework until the class immediately before, swim 15 hours per week, have sensory issues, and can't pay attention... no ADHD" 2021-07-09 20:24:45 Or talking a lot, Tweeting a lot, doing a lot of competitive sports https://t.co/RB3aXjyqFY 2021-07-09 19:42:39 RT @arb: losing my mind at this suggested google search https://t.co/h4dHExJzJQ 2021-07-09 15:44:21 "Casual inference" implies the existence of "fancy inference" and "business casual inference" 2021-07-09 15:35:30 I often read "causal inference" as "casual inference" and it makes me laugh every time 2021-07-09 15:33:41 @alex___180 @geekingthething @chrisrankin @NathanWyburnArt @wayne_courts @PrideCymru @prideinglos @PrideInLondon @biminibabes @GinnyLemon69 @joelycett @alexander_olly Super cute 2021-07-09 15:31:45 RT @alex___180: Don’t forget my shop is shutting on Saturday at 11:59pm for 4-6 weeks (I will announce when I am reopening!). Don’t miss o… 2021-07-09 15:23:16 https://t.co/wHbFXPiU9O https://t.co/9RD4RNT3vw 2021-07-09 15:17:34 RT @MSPA_National: Right now we are calling for current LGBTQ+ Residents who would be willing to mentors a current LGBTQ+ M4 in this applic… 2021-07-09 14:12:51 From @elazar_g hahaha https://t.co/NlBBYfSAwX 2021-07-09 14:10:38 https://t.co/RC51pSQPTZ 2021-07-09 12:32:56 Formal details here: https://t.co/7cot0XVHJ3 Or in the HoTT book. 2021-07-09 12:23:21 But basically, whether you're using a tool or reusing proofs across equivalent types by hand, you often want to work with the "adjusted" adjoint equivalence, because it's much more well-behaved. This lets you do that, but without pulling in HoTT at all 2021-07-09 12:21:18 PUMPKIN Pi includes an option that will automatically generate this proof for equivalences it discovers and proves for you. 2021-07-09 12:19:11 Then, in your proof, where your rewrite failed, but intuitively "should work": rewrite is_adjoint. destruct retraction_adjoint. and similarly for section. 2021-07-09 12:18:03 You can do that as follows: Definition retraction_adjoint := Adjoint.fg_id' f g section retraction. Lemma is_adjoint (b : B) : section (g b) = f_equal g (retraction_adjoint b). Proof. apply Adjoint.g_adjoint. Defined. and similarly for section. 2021-07-09 12:14:30 This proof lets you "adjust" the type equivalence to something that makes rewriting actually tractable. 2021-07-09 12:13:53 When do you need the adjoint equivalence? Roughly, when in any proof about a equivalence, you find that you are trying to rewrite by section or retraction (the proof that the two functions are mutual inverses), and it fails. 2021-07-09 12:10:41 Check out out, and use it to write some practical proofs without univalence using type equivalences :). 2021-07-09 12:09:48 The proof doesn't rely on univalence at all. It's ported from the HoTT book into vanilla Coq. It's not an obvious proof. I failed to port it myself. I used Jasper's proof of this in my most recent PLDI paper, Proof Repair across Type Equivalences. 2021-07-09 12:08:29 My former undergraduate student Jasper Hugunin proved this property in vanilla Coq, and it's now in the Coq standard library: https://t.co/Thzmm80SPn 2021-07-09 12:08:15 Type equivalences are useful not just in HoTT, but more generally. But there's a certain property in HoTT---that every equivalence induces an adjoint equivalence---that you need to use them in practice. 2021-07-09 11:49:55 RT @AxiomCrisis: After a year struggling I would recommend anyone interested in doing Coq based FV research reading the 1st, 2nd, 4th volum… 2021-07-09 11:49:14 RT @FiveThirtyEight: PSA: We’re sharing the data and code behind some of our articles and graphics. We hope you’ll use it to check our work… 2021-07-09 11:17:55 @AdamBatesOrg @IllinoisCS Thanks 2021-07-09 10:13:36 What are the logistics for faculty who do a year or a semester or whatever visiting at another institution before sabbatical? Just curious what this is like for students, teaching, and so on. 2021-07-09 10:05:03 Queer Israeli-American leftist Jews with PhDs? There's a community for that in Tel Aviv 2021-07-09 09:58:52 RT @noa_landau: מחקר מרתק של שי חזקני. שווה לקרוא 2021-07-09 09:43:36 Tel Aviv feels like the only place I've ever been where I can just totally and completely be myself, in all of my many complex dimensions. I love it and I hope to spend more time there later in life. Sabbatical in Tel Aviv? 2021-07-09 09:29:34 RT @jburnmurdoch: NEW: as we get more and more studies on vaccine efficacy, it becomes harder to distill their results into one clean summa… 2021-07-09 09:26:42 @matthew_petti Reported to the UN for War Crimes 2021-07-09 09:09:14 RT @geraldstratfor3: A beetroot is a lovely vegetable cheers https://t.co/IqqORmn5Vm 2021-07-09 08:56:35 RT @poisonjr: Yahoo searches hit so different https://t.co/hMQey0iu7V 2021-07-09 08:32:08 RT @bernpton: @HilaCodes @HerrDreyer @existsShachar Its generally the right approach to not trust machines to do anything consistently ever. 2021-07-09 08:28:54 RT @twisteddoodles: Frog life cycle. (cartoon for @newscientist) #science https://t.co/A8a20bbwGH 2021-07-09 08:27:27 @EremondiJoey Congrats! 2021-07-08 22:34:53 @gone_things This is so meta 2021-07-08 22:34:42 RT @gone_things: An #anachronism was a chronological inconsistency in some arrangement, especially a juxtaposition of persons, events, obje… 2021-07-08 22:32:49 RT @CompArchSA: Reminder: our third session is tomorrow from 2-3PM ET, and the night session is from 11PM-12AM ET. Attend one or both! (Reg… 2021-07-08 22:18:59 @adamCS I think it's often necessary not to be kind. But yes good to think about before acting 2021-07-08 22:17:41 @HHMusicOfficial Bad 2021-07-08 22:04:05 Mood https://t.co/PPH49JtulR 2021-07-08 21:39:44 RT @TaliaRinger: On this, I recommend the following exercise whenever you write anything negative about or to another person on Twitter : 2021-07-08 21:31:24 I don't want to leave Tel Aviv 2021-07-08 21:19:23 Now the taxi driver is asking me whether AI is going to control humanity and I keep telling him that actually it's a bigger worry it will repeat and amplify human mistakes in trusted contexts and I have had too many drinks for this 2021-07-08 21:18:11 Just attended a rooftop party and concert in Tel Aviv with a friend of a friend who I met through a mutual serial harasser, plus a Technion professor whose Tweets about the COVID Israel number methodologies I commented on two days ago, and what is life 2021-07-08 18:19:19 RT @TedOnPrivacy: Personal news! Today's my last day at Google. My "shields down" moment was the firing of @timnitGebru, then @mmitchell_a… 2021-07-08 18:18:35 חבל על הזמן https://t.co/DCz2UscVuR 2021-07-08 16:24:55 @JeffDean @PLMentoring @SigplanM @heyelbs @CompArchSA US citizens only, but absolutely amazing new NSF program with mentorship and funding and help getting into a PhD program when you have already finished undergrad and have been working for a couple of years: https://t.co/a5vocDSwYQ 2021-07-08 16:14:25 Ugh https://t.co/qR0gAxpVCf 2021-07-08 16:14:05 @edokonrad Ugh 2021-07-08 16:09:53 I don't want this guy in either country I have citizenship lol 2021-07-08 16:04:41 Man, just heard my first homophobic comments in Israel. In English. From an American. What a loser 2021-07-08 14:35:34 RT @TaliaRinger: On this, I recommend the following exercise whenever you write anything negative about or to another person on Twitter :… 2021-07-08 13:30:46 I'm gonna go out tonight in Tel Aviv with some other queer Israeli-Americans and this feels like the perfect outfit 2021-07-08 13:30:07 Siiide https://t.co/mtYJRUeP8s 2021-07-08 13:26:59 I'm so happy crop tops are in again 2021-07-08 13:26:38 New outfit feels fabulous https://t.co/eIi0qZz7Gs 2021-07-08 13:01:54 RT @satnam6502: Two white women refused to sit next to our two mixed race kids on sailing boat cruise out of Camden. My Californian kids we… 2021-07-08 11:37:03 RT @lmanul: After more than 14 (!) years at @Google, I am joining @Twitter. I'm also publishing volume 2 of my comics about life at Google.… 2021-07-08 09:22:27 Here's the email I'm talking about, including Meg's commentary on it, so that you can see in this example what I considered and where I went wrong, and use it as a reference. https://t.co/MxfQREl3pD 2021-07-08 08:52:03 I'm always willing to be a sounding board for these kinds of things. I just want people to think more about how they use the tool that is Twitter. It's an amazing tool for activism sometimes, but it can also be very dangerous. Please be careful. 2021-07-08 08:50:36 My only regret is not redacting "acting within existing systems," since I later learned that to be possibly or likely false, and so publicizing an old email without redacting it likely cause some minor harm to Meg. I feel bad about that mistake. 2021-07-08 08:47:18 This is easy when it's Sundar, given how much power he has. So I chose an easy example. In part because publicly writing about any other example may harm people in ways I don't want. But hopefully you can see how this works in practice. 2021-07-08 08:45:47 I think I used my influence quite appropriately to get people to think more about Sundar's role. https://t.co/WvCzwu200a 2021-07-08 08:44:55 The status quo, in particular, was intolerable. 2021-07-08 08:44:16 I thought about the idea of making myself potentially adversarial in a noticeable way toward one of the most powerful people in tech. I decided I don't care. 2021-07-08 08:43:25 And I accepted that even in the worst case, if Google does sue me at anything for any point, I'd honestly kind of be OK with that 2021-07-08 08:41:35 I also considered consequences for myself, like whether Google could try to sue me for anything I say. I did a lot of reading about defamation and public figures, and made my justification for posting likely being lawful explicit. 2021-07-08 08:40:09 I considered consequences for Sundar's direct reports, like Jeff. And I considered harm to Timnit and Meg, to any sources I spoke to at any point, and to the Ethical AI team. This is why I redacted parts of the email. https://t.co/Et6IgoceKd 2021-07-08 08:38:12 There is basically no sense in which I have any power over Sundar, except maybe the racial lens. That seemed like an acceptable cost considering the whole CEO of Google thing. https://t.co/1PlBOnULDj 2021-07-08 08:36:23 I decided to post publicly because I think one major reason we have seen no repair is that Sundar has managed to evade all criticism and accountability for his role in this. I want people to consider protesting directly to him in order to get that repair. https://t.co/1SNvZ2axIN 2021-07-08 08:33:02 I considered how Sundar may feel if he sees this being posted publicly, and concluded probably only mildly concerned about press picking up on some of my comments. He is not a part of the community from which I am criticizing him, which is noteworthy. https://t.co/vZVy0FOPkd 2021-07-08 08:31:18 I tried to send it privately to Sundar first. I wanted several months. I saw the same patterns emerge, and concluded that my email to Sundar had no impact. https://t.co/QwLBC6GOSX 2021-07-08 08:30:03 I'm going to walk through how I did this exercise with the recent post with my email to Sundar, now. 2021-07-08 08:14:24 With Twitter in particular, I've found that it's sometimes easy to assume the person on the receiving end of criticism or even shame and insults is unphased and unimpacted when this is not true. It's important to shatter this illusion so that you can be honest with yourself. 2021-07-08 08:11:05 In the end, I think public negative criticism and even shame can be justified. I also think not everything we do needs to be justified 2021-07-08 08:09:36 6. Consider your own personal influence, and the dynamics of others in your network commenting at the same time. Will this lead to dogpiling? What will the consequences be, if so? The more followers you have, for example, the more important it is to think about what you Tweet. 2021-07-08 08:06:15 5. Consider the consequences not just for the person you criticize, but also for other people, and for yourself. Are you putting other people in harm's way? Does the person have people close to them you may harm? Are you putting yourself in danger? Is that acceptable to you? 2021-07-08 08:01:34 4. Consider power dynamics. Public criticism coming from a position of power can be much more devastating, and hijacked much more readily. Anyone in any marginalized group likewise is particularly susceptible to hijacking by bad actors. 2021-07-08 07:59:41 3. Introspect a bit. Why are you posting this? Is it because you're angry? Because you want to change something? Because you want to influence public opinion or strategy? All are valid, but it can be informative to understand what you are doing. 2021-07-08 07:58:14 2. If you choose to post publicly, assume they will read it. Imagine what it is like to be them reading it. Consider how they feel. Ask yourself is this harm is necessary and acceptable to you. 2021-07-08 07:57:08 1. First ask yourself if it is a thing you can say privately to them directly. This has much better dynamics in that bad actors can't hop on it and hijack your criticism, and it can be less overwhelming for the person you are criticizing. 2021-07-08 07:55:49 On this, I recommend the following exercise whenever you write anything negative about or to another person on Twitter : https://t.co/ta4pjHDwJ5 2021-07-08 07:51:35 "But the structure of Twitter and the way it rewards a constant escalation of emotion makes it exceedingly difficult to just back down, to say, “I thought I was doing the right thing, but I hurt somebody very badly in the process.”" 2021-07-08 07:49:15 “like all weapons, it will do the most damage when aimed at the least defended, the isolated, those with no one to stand up for them, publicly or privately. And we must be careful with the temptation to use it inside our own houses to destroy shapes we think are intruders.” 2021-07-08 07:39:30 “I sought out and read everything written about the story. I couldn’t stop ... What if someone gave you a ledger of everything anyone’s ever said about you, anywhere? Who wouldn’t read it? I would read it 2021-07-08 07:28:19 "Sometimes, the path to your personal hell is paved with other people’s best intentions." 2021-07-08 07:28:09 "But what this story really symbolizes is the fact that as we’ve grown more adept at using the internet, we’ve also grown more adept at destroying people’s lives, but from a distance, in an abstracted way." 2021-07-08 07:26:20 https://t.co/xTFrWDD1B0 2021-07-08 07:22:30 RT @boarbarktree: @ryanqnorth https://t.co/ldEhTQ6vHk 2021-07-08 06:35:42 RT @Tyler_UCSC: We're lucky to have people like Heidy (with a strong formal PL background) working on developing and evaluating tools like… 2021-07-08 06:34:51 This paper looks like it is really good discussion of a lot of the questions I asked here the other day! Definitely lots of work to do as a community, and collaborations with ML & https://t.co/wNPxz82uQZ 2021-07-08 06:30:24 @HeidyKhlaaf I don't know how expressive the annotations are 2021-07-08 06:28:38 @HeidyKhlaaf One thing I could imagine, though, is that if the tool could help infer those specs interactively, maybe it could check the suggested code against the proposed spec too? And eventually use those specs to help with suggestions? 2021-07-08 06:27:23 @HeidyKhlaaf Very hard would be annotating the entire dataset with a human in the loop, though And I'm quite terrified of the idea of doing this fully automatically 2021-07-08 06:24:57 @HeidyKhlaaf One example of this is Python's static type annotations: https://t.co/JEcw5wTs5S I would not trust a tool to infer these fully automatically, but with a human in the loop, interactively, why not? 2021-07-08 06:21:03 @Miles_Brundage @HeidyKhlaaf @OpenAI Thanks! A skim looks pretty promising 2021-07-08 06:08:45 RT @Miles_Brundage: Oh, and we released the evaluation dataset + associated harness here: https://t.co/MJUzcGuIyC and the alignment dataset… 2021-07-08 06:05:52 RT @HeidyKhlaaf: I defer to @GretchenMarina, @PamelaMishkin, and @girishsastry on fairness, bias, economical, security impacts. There's a l… 2021-07-08 06:05:07 @HeidyKhlaaf With a human in the loop, I think you could enforce some kind of external static system 2021-07-08 06:04:00 RT @HeidyKhlaaf: For one, if Codex suggests code that only superficially appears correct, this negatively affects many programmers. But it'… 2021-07-08 06:03:53 RT @HeidyKhlaaf: It also struggles with complex programming constructs and relationships (again see Appendix D). So what does this mean in… 2021-07-08 06:03:51 RT @HeidyKhlaaf: But where it excels it can recommend syntactically and semantically incorrect (as many of you have kindly pointed out) or… 2021-07-08 06:02:23 RT @HeidyKhlaaf: Codex paper is out! I'm grateful to have led the Safety and PL workstreams for Codex/Copilot, working along Policy @OpenAI… 2021-07-07 21:47:16 RT @AdhdAngsty: the "I found out I have ADHD and now think I\'m broken" to "maybe being neurodivergent means I\'m smarter than everyone else"… 2021-07-02 22:21:02 God this hat makes me look so stuck up 2021-07-02 22:18:29 Building on the proof repair work lol 2021-07-02 22:17:51 TBH I'll blog this tomorrow if still motivated on an already extremely tractable example that you machine learning people might like because it could also help you with full static guarantees about tensor sizes by asking you a few questions, I think 2021-07-02 22:00:13 (Tools for concept alignment already exist) 2021-07-02 21:59:23 Then you can work on moving beyond limited contexts and get this workflow for more general proof automation, which is a harder problem IMO even if it's older, since you have less information to work with 2021-07-02 21:57:48 Hook in tools for aligning concepts across libraries, and you get to repurpose those repair tools for development 2021-07-02 21:56:44 Oh and yes it's true that I like hard problems, but honestly the technologies are developed enough that in limited contexts (for example, certain kinds of proof repair after incremental changes), this is like a 1-2 paper task for real 2021-07-02 21:47:59 @bisexualposibot It me 2021-07-02 21:47:38 YESSS https://t.co/yi37NBAWTO 2021-07-02 21:36:12 I just feel so impatient, I'm supposed to be on vacation but I want to make this happen *now*. I hate sitting still while the world moves right past me, it fills me deeply with anxiety https://t.co/OPFO6y9uEe 2021-07-02 21:27:36 @ShlomoArgamon I'd be interested in some studies on how humans think about code behavior, so that we can reconcile that with how automated tools "think" about code behavior and facilitate communication between the two 2021-07-02 21:26:26 Anyways please work with me thank you 2021-07-02 21:26:01 And explanations and transparency and help testing generated specs are *hugely* important because they make it clear the computer is just helping out but cannot be trusted without human vetting 2021-07-02 21:25:07 The human takes the driver's seat because the human is prompted with questions about everything interesting---the computer recognizes the human as the expert! But it takes seriously good proof automation or AI or whatever to figure out what questions to ask and how to ask them 2021-07-02 21:22:42 These questions notably should not just be I/O examples, but say questions about control flow, whether other lemmas ought to hold, and so on---think really broadly here 2021-07-02 21:19:33 I want to just totally reimagine UX here, and I think there are *extremely interesting* proof automation problems this naturally opens up 2021-07-02 21:16:55 If anyone wants to work on this kind of thing I'd likely take students interested in it immediately, it is directly relevant to some ongoing work 2021-07-02 21:15:33 @ShlomoArgamon Possibly! Can you elaborate? 2021-07-02 21:15:08 When I think about verification for software engineers, this is how I imagine it 2021-07-02 21:14:52 It should then ask the programmer questions to arrive at that input. Work with the human to understand both code and intent, as a team. Explain everything it thinks of, always make it possible to test things along the way 2021-07-02 21:13:23 The same holds for proofs. Good proof automation or AI or whatever you want to call it ought to be able to dispatch *most* of the proof, but along the way identify the minimal proof obligations it cannot prove alone, for which it needs user input 2021-07-02 21:11:49 And it should have ways of testing everything it potentially arrives at along the way to make sure it's reasonable, and explaining how it arrived at that spec 2021-07-02 21:11:00 Honestly I think rather than expecting programmers to write specs, the frame is totally wrong. From my view, the IDE should be *asking questions* to the programmer, from which it can converge with the programmer on a spec 2021-07-02 21:07:46 @emilymbender @alkoller I think right now it's naively syntactic, but I think the right people are collaborating with them such that in the next year or so, it won't be. At least I hope 2021-07-02 21:06:16 @va2lam 9 too many 2021-07-02 21:00:16 I love this so much https://t.co/yEIUDVV4Br 2021-07-02 20:47:43 @Impredicativity https://t.co/lJLbXS5y69 2021-07-02 20:42:54 I don't know if I'm just wired for sunny places, but the higher the UV index, the happier I feel. It's like this buzz all day after I spend any time outside, it's so wonderful 2021-07-02 20:29:58 She says my second PhD should be in Hebrew 2021-07-02 20:03:07 @emilymbender @alkoller So while I think it's somewhat irresponsible to release this tool in this state without immediate engagement on these issues, and to not discuss UX communicating lack of trustworthiness, the long-term situation looks potentially reasonable 2021-07-02 20:01:23 @emilymbender @alkoller I think you'd be happy to know though (sorry for spam I've been obsessing over this for days) that it seems from public Tweets that the Copilot folks are talking to neural synthesis folks at MSR, and those MSR folks I know are working on *neurosymbolic* synthesis 2021-07-02 19:55:13 @emilymbender @alkoller I also want to try this for natural language, if you're ever interested 2021-07-02 19:54:29 @emilymbender @alkoller But for that, you need type theory 2021-07-02 19:54:03 @emilymbender @alkoller It does mean IMO there's some hope syntactic methods can achieve semantic understanding, as long as there is first a preprocessing step to internalize the semantics and reify it in the syntax. I want to try this really badly and see what happens 2021-07-02 19:52:33 @emilymbender @alkoller One thing worth mentioning is that depending on the type system, more and more of the semantics of the programming language will show up in the syntax itself. But this is certainly not true for most languages people use, where much of the semantics is external 2021-07-02 19:50:47 RT @emilymbender: Watching all the #copilot discourse and hoping that people can see the analogies between such an approach to auto-generat… 2021-07-02 19:44:52 @mpiedrav It is probably net good for the world though 2021-07-02 19:39:21 @codydroux Cody, you're great sometimes! 2021-07-02 19:36:07 Dollar is so weak right now 2021-07-02 19:32:48 @codydroux Was honestly the first thought in my head and I'm kind of glad this interaction wasn't in person or I would have accidentally actually said it 2021-07-02 18:14:36 Tempted to respond to an email just now with "You are sometimes great!!" and had to laugh and stop myself 2021-07-02 18:13:35 @ncooper57 Thanks!!!! I'll get to it soon and chime in 2021-07-02 18:10:23 Huge, huge, huge congrats to @ChandrakanaNaN on defending her thesis today!!! She is an amazing person who was a great friend to me for all six years of grad school. She deserves amazing things and I can't wait to see what's next for her in life!!!! 2021-07-02 16:23:10 @johnregehr @MichaelKup @ethanhs @SeanTAllen I think I haven't personally encountered that, both in industry and as a researcher (or, if I have, I haven't noticed) 2021-07-02 16:22:35 @johnregehr @MichaelKup @ethanhs @SeanTAllen Ah I misunderstood what you mean by doc bugs, those sound really awful too 2021-07-02 16:21:09 @johnregehr @MichaelKup @ethanhs @SeanTAllen This means in lieu of the symbolic interventions that will likely eventually exist, the responsibility of adequately signaling "please don't trust me" is much, much higher for now 2021-07-02 16:19:53 @johnregehr @MichaelKup @ethanhs @SeanTAllen I think so precisely because of the technologies involved, which by definition will make the bugs extremely syntactically viable, and thus more deceptive. Most of the bugs in the examples are subtle and tricky 2021-07-02 16:18:23 Night out https://t.co/5J9Xk4kwlu 2021-07-02 15:44:02 @elfprince13 Want 2021-07-02 15:43:56 RT @elfprince13: @TaliaRinger One of my friends earlier this week showed me a picture of a shirt that was the anarchist symbol except the A… 2021-07-02 15:40:11 2021-07-02 15:39:54 @supersat @CDCgov To be fair I had to test before flying, then test upon landing, THEN do a serology test to enter Israel and not isolate as a citizen 2021-07-02 15:39:00 @johnregehr @MichaelKup @ethanhs @SeanTAllen I think this is only true for the particular kinds of bugs seen in toolchains right now 2021-07-02 14:58:02 @noneuclideangrl Illegal chocolate 2021-07-02 14:57:08 Yessssysysyzzyes https://t.co/GwTFmNrXrR 2021-07-02 14:53:39 Hearing Savta [grandma] talking glowingly about any progress she sees in trans rights is just the coolest thing 2021-07-02 14:28:38 It's always OK to call me "Talia" and not "Dr. Ringer" or "Professor Ringer" FWIW unless you feel very uncomfortable doing that. This holds for everyone regardless of seniority and so on, I prefer it but understand if it makes people feel weird 2021-07-02 14:18:55 @ShriramKMurthi 2021-07-02 14:00:58 @ShriramKMurthi Very strange haha. I think I *can think* in terms of logical systems but it's like one of many ways I think. Brains are complicated lol 2021-07-02 13:58:52 Hi @ncooper57 can I help with some of the concerns I had for Copilot that are related to my area of expertise? I'd be so happy to 2021-07-02 13:49:58 Someone please do this so we can get transparency and improve it. I'd be poorly suited to build this, but likely very well suited to help improve correctness, efficiency, and UX once built, and would if something open and transparent existed https://t.co/bMIb2n0j0X 2021-07-02 13:36:17 @hillelogram Like the food banks I volunteered at during COVID needed, more than anything else, people willing to risk exposing themselves to COVID to deliver food to people who needed it, not any technical help 2021-07-02 13:33:47 @hillelogram Honestly sometimes the help they most urgently need isn't tech help, but is help anyone could do, and IMO it's best to be willing to do that if you really care about the cause 2021-07-02 13:32:03 RT @hillelogram: In short: if you're a software developer and want to improve things, here's what you can do: 1. Find an org doing work yo… 2021-07-02 13:31:47 RT @hillelogram: I think that's the root of the problem. Tech is at its best when we're egoless and subservient, amplifying the abilities o… 2021-07-02 13:31:39 RT @hillelogram: It goes both ways, doesn't it? The cryptobros were trying to solve social ills without even the barest understanding of wh… 2021-07-02 13:28:47 Signed a lease for mid-August in Champaign! I expect to be in and out of town slowly getting settled until my real start date in October 2021-07-02 12:59:02 @sdumbrav I think she is from Galați 2021-07-02 12:40:23 @yoavgo I will probably go to Tel Aviv Sunday 2021-07-02 12:36:25 @yoavgo DM if you get bored and want coffee sometime, I'm in Ness Ziona 2021-07-02 12:34:56 @yoavgo Also I'm in Israel if you feel like hanging out sometime haha 2021-07-02 12:31:34 Are these feelings normal https://t.co/eqXOrOwnmv 2021-07-02 12:31:10 @yoavgo Hopefully the students I work with can help guide this a bit, and steer me toward new directions too 2021-07-02 12:30:00 @yoavgo Somehow it was easier to pick just one of the 10000 things I want to do than it is to pick just five 2021-07-02 12:27:58 @yoavgo Yeah true, I think it's a unique moment of anxiety knowing I'm about to pivot in my career, but not knowing exactly how as I await more students (I have one collaboration going on remotely, and will likely have one PhD student this fall, and then keep building) 2021-07-02 12:23:34 @yoavgo Also some of the fundamental design decisions rule out my ability to do some work I am really interested in 2021-07-02 12:23:11 @yoavgo Lean is a newish proof assistant very similar to Coq, but lead by Leo de Moura at MSR. Leo is incredible at building systems that reach real people (he is also behind Z3) but OTOH the communities for Lean and Coq have almost no overlap in contributors and users 2021-07-02 12:21:24 @yoavgo Ah yes, I'm not sure I want to though? I want my ideas in *both* systems haha 2021-07-02 12:17:22 @ShriramKMurthi I don't think it's "how we think," I think it's a collection of useful languages that we have for thinking, many of which are useful 2021-07-02 12:03:35 Overall it feels like in PL I'm often expected to be both a top-tier theorist and a top-tier engineer, and it's tiring, I would really like some engineering support. Maybe I'll hire a research scientist or something 2021-07-02 12:02:46 If my infra benefits any other Lean folks I think it's a worthy investment. But I have too many options for worthy investments. Likewise I'd love my work in coq/platform on the latest Coq version 2021-07-02 12:01:21 I suppose I'll see how motivated I feel when I get back from vacation 2021-07-02 12:00:19 I think part of the frustration I feel seeing PL technologies with possible negative societal implications not fully considered or mitigated is that it feels like it pushes me toward feeling obligated to rush industrial adoption immediately, moving me away from fundamental work 2021-07-02 11:59:09 And any time I spend doing that is time I could spend more on advances building on the infrastructure I already have, and I don't know which is more important. I think maybe the former, since it's a unique time in history 2021-07-02 11:58:28 As an example, I have a chance to contribute many of my ideas to Lean, which I think will do very well at industry adoption eventually. I'd love to be a part of that. But it means porting my infrastructure from Coq to Lean first, which is a huge investment 2021-07-02 11:57:30 One thing I'm struggling with a lot right now is that I have many concrete opportunities to contribute to adoption of the technologies I care about in industry, but this may come at the cost of time spent on fundamental research advances. I'm not sure how to weigh those options 2021-07-02 11:49:18 @clairlemon When it's an active decision, it's unfortunate, because it rules out a lot of useful systems of reasoning that can lead to important insights, much like speaking a different language gives you a new frame of thought 2021-07-02 11:47:51 @clairlemon The choice to ignore those logical systems entirely, but enforce a very particular logical systems ranges from negligence on the part of teachers to active political decisions to exclude entire useful lenses for reasoning about the world 2021-07-02 11:46:56 @clairlemon So there are entire well-defined, useful logical systems in which you can, with enough work, reason about things like positionality, like the background and assumptions of the person doing the reasoning. These logical systems are also useful for quantum physics and law 2021-07-02 11:44:34 @clairlemon But there are certain classes of logical systems that have nice properties for reasoning about reality. Like consistent logics and paraconsistent logics. Paraconsistent logics can handle quite nicely reasoning about true conclusions given false assumptions or biases 2021-07-02 11:43:22 @clairlemon This is not to say the choice of logical system is arbitrary. Some logical systems are really useless. Like a logical system that accepts everything as true is not very useful, nor is one that rejects everything as false 2021-07-02 11:39:56 Quillette is uniformly terrible, but yet again we have super failed to educate people if they think of "logic" as a single logical system https://t.co/89oK5jDGu6 2021-07-02 11:34:58 @clairlemon It's a failure of education that so many think of logic as the single system of reasoning you are accustomed to, namely first-order classical logic 2021-07-02 11:33:52 @clairlemon Logic is the study of different systems of reasoning, none of which completely capture truth and reality, but many of which can be used as lenses for viewing fragments of reality which lead to useful models and have nice properties 2021-07-02 11:32:33 @ShriramKMurthi In any case, Quillette is uniformly awful 2021-07-02 11:31:40 @ShriramKMurthi People tend to think of logic as "that one system of reasoning I learned in school, which is The Correct Way to think" and this is mostly a failure of education IMO 2021-07-02 11:20:49 @MichaelKup @ethanhs @SeanTAllen @johnregehr In that sense, I do think most compilers fail to do the right thing 2021-07-02 11:20:24 @MichaelKup @ethanhs @SeanTAllen @johnregehr Nope. The other solution is to signal your trustworthiness accurately, so if you may suggest bugs, you have a responsibility to use UX corresponding to how trustworthy you are/how often you introduce bugs, so people trust your tool appropriate 2021-07-02 11:18:23 @sdumbrav Yessss 2021-07-02 09:39:17 When I raise concerns without constructive criticism or solutions, it's because I'm exhausted from six years of having to do all of that work, and occasionally want other people to do some of it so I can rest 2021-07-02 09:36:27 @sdumbrav It used to be part of Moldovia 2021-07-02 08:43:06 It's impossible to explain the love of a Romanian-Israeli Jewish grandma. But it involves 10 minutes of arguing every morning that you shouldn't be able to go out for a run in this sun without one of those giant hats 2021-07-02 08:38:07 @va2lam My personality horoscope is extremely accurate https://t.co/ekFy0IrCUd 2021-07-02 08:35:13 @va2lam Interesting, not sure I believe in Myers-Briggs, but were any grad school personalities disproportionately weeded out? Also despite not believing in it I'll state these days I'm usually ENFJ lol 2021-07-02 08:09:35 Also, as faculty we escape some of the unhealthy power dynamics by virtue of gaining power ourselves and finding ourselves on the other end of them 2021-07-02 08:06:23 Even though mental illness is fairly common in academia, I think this is mostly a symptom of unhealthy systems and power dynamics triggering mental illness in those of us predisposed to it, and I'd guess that retention among those with mental illness is much lower 2021-07-02 08:05:01 My PTSD and severe depression had no benefits on the other hand aside from making me a more compassionate and empathetic person in the long run, dedicated to helping others in the community. The costs were massive though 2021-07-02 08:03:13 But these are only benefits when we get adequate support for our needs, and adequate accommodations in academia are a rarity. Also RSD in academia is hell 2021-07-02 08:02:19 ADHD has some concrete benefits for academia IMO. Like less inhibition diving into new problems, which manifests as creativity. Impulsivity, which manifests as fearless pursuit of possibly difficult topics. Hyperfocus which is a gift when it happens in moderation 2021-07-02 07:56:14 I know few faculty who openly have ADHD, but I suspect also that many are undiagnosed due to norms of earlier generations 2021-07-02 07:54:40 I don't have a good hypothesis here, but I'd be interested. Lots of ADHD folks and folks with depression that I know for example dropped out or went on leave due to inadequate support for their needs 2021-07-02 07:53:11 I know rates of depression and neurodivergence are probably both higher in CS PhD students than in the general population. I'm quite curious how those rates compare to the rates in those who go on to be faculty 2021-07-02 07:50:37 RT @teachingofsci: Short but thought-provoking. Whether to declare 'invisible' conditions as disabilities (chronic illness such as fibromya… 2021-07-02 07:12:05 RT @peard33: Google's latest annual workforce diversity report is out: Google reports soaring attrition among Black women https://t.co/HrV… 2021-07-02 07:10:06 OH: loud argument between my mom and my grandma in which my mom is saying in Hebrew "don't worry, I don't need cake!" over and over again while my grandma expresses deep concern about something 2021-07-02 07:05:35 RT @PTOOP: Given the political circumstances, I feel bothered by the need even to say that the presence of trans people in my professional… 2021-07-02 06:54:41 @agdakx @harriesadam Very cool initiative!!! 2021-07-02 06:54:32 RT @agdakx: @TaliaRinger @harriesadam Well, you might be interested in the objectives of the recently-granted EuroProofNet COST action: htt… 2021-07-02 06:51:33 Dad: I don't understand how you can sleep so much. I slept like 2 hours. Me: 6 years of not sleeping very much? Dad: Ah, OK, take it easy, you're on vacation 2021-07-02 06:45:56 RT @ZJAyres: I've been reflecting a lot on whether to check the "I have a disability" box on forms. Typically I do not, because when well m… 2021-07-02 06:40:24 @ncooper57 This makes some sense to me, given that errors in values are not captured by most type systems 2021-07-02 06:20:22 RT @mycoliza: this is tea. what you’ve invented is tea. https://t.co/uafsm0Wkr6 2021-07-02 06:17:46 @lysxia Yeah we are lucky enough to be in a language where incorrect answers are harmless though 2021-07-01 22:32:26 This is a funny solution https://t.co/7odXjwHpP9 2021-07-01 22:23:11 @TomChen17 @kurtsy I'm not pretending that at all? 2021-07-01 22:20:03 @TomChen17 @kurtsy Which is exactly what I wrote last time so I'm confused how it implies anything like that 2021-07-01 22:19:34 @TomChen17 @kurtsy Tools either need to be trustworthy, or need to signal adequately that they are not trustworthy 2021-07-01 22:10:19 @madanMus @kurtsy Probably the only people doing the right thing here are those actively accelerating neurosymbolic methods like a bunch of MSR folks but I'm personally absolutely terrified everyone will just ignore that work until it's too late 2021-07-01 22:07:56 @madanMus @kurtsy Then 10 years later they'll be like "oh yeah maybe we should make things less buggy" and reintegrate our work but we won't have jobs 2021-07-01 22:07:15 @madanMus @kurtsy Without that kind of UX, though, a tool like this worries me a lot. I also think that if things are syntactically plausible enough, nobody is going to care to fund research into how to make things less buggy, at least until all of us lose our jobs and funding like in NLP etc 2021-07-01 22:06:13 @madanMus @kurtsy But the temporary answer to this can be good enough UX to signal untrustworthiness corresponding to how common this is, while the field grapples with how to best integrate those symbolic interventions 2021-07-01 22:05:04 @madanMus @kurtsy There is something that seems notably scarier to me about introducing new syntactically plausible, very subtle bugs, of the kind that NNs without any symbolic intervention seem to be really great at producing---because this is a really misleading betrayal of trust 2021-07-01 21:58:32 @ErickGalinkin Thanks 2021-07-01 21:55:02 @madanMus @kurtsy That also seems Really Bad. I always look at merges in great detail but if that's the common use case, then semantic merge is prudent to develop, or better UX signaling this possibility 2021-07-01 21:53:48 @mdekstrand Planning some kind of interactive activity within SIGPLAN based on this right now (optional, with some reward for completing it) 2021-07-01 21:50:21 Vulnerability concern an appropriate response to 2.9! 2021-07-01 21:48:23 @mdekstrand It's a good code, it's too bad so few people even read it, let alone take it seriously 2021-07-01 21:47:47 RT @mdekstrand: @TaliaRinger What I love about the code is it specifically requires proactivity in considering possible harms, and does not… 2021-07-01 21:47:38 Licensing concern is an appropriate response to 1.5! 2021-07-01 21:46:50 2.7 sharing vulnerabilities and limitations of new systems is a *responsibility* we are beholden to 2021-07-01 21:45:32 2.5 hugely relevant 2021-07-01 21:43:07 See 1.2 Avoid Harm this means all of us tool developers 2021-07-01 21:41:49 Enjoy https://t.co/dQ5yTczbEJ 2021-07-01 21:40:36 BTW you all need to read the ACM Ethics Code which you violate every single time you don't seriously consider the impacts of your technologies on people 2021-07-01 21:37:11 @ian5v Godspeed, from what I've seen this is a difficult task 2021-07-01 21:36:34 @ian5v Follow @mmitchell_ai, @timnitGebru, @alexhanna, @AJLUnited, @schock, @geomblog, many others but that's a nice start 2021-07-01 21:31:07 Only in response to my Tweets. You all can do whatever you want on your own Twitter 2021-07-01 21:27:15 @lmeyerov @samth I'm also tired though and I reserve the right to express my depression about possible futures on social media 2021-07-01 21:26:38 @lmeyerov @samth This is a large part of how I'm hoping the ML + type theory workshop will help, with at least one focus group working on type theory for neurosymbolic methods 2021-07-01 21:25:52 @lmeyerov @samth "I feel depressed because I'm worried about this field going the way NLP did 10 years ago and PL experts losing our jobs before our expertise is reintegrated into products without credit" is a valid invitation to engage and tell me how we stop that from happening 2021-07-01 21:24:22 @lmeyerov @samth Yes but many people express constructive criticism informally on social media by complaining or expressing worries and expecting people to respond so that a conversation ensues converging toward progress 2021-07-01 21:23:01 I'm going to mute everyone else who scares me unless you're students sorry I can't handle it 2021-07-01 21:22:30 Honestly you all have me like 5 Tweets away from switching into Ethical AI research please stop espousing views on technology and our responsibilities in developing it that scare the crap out of me 2021-07-01 21:20:12 @kurtsy Yes, if your driverless car gave you the illusion of trustworthiness 2021-07-01 21:16:53 @lmeyerov @samth The answers don't need to be, "this is how we address everything, we're good." They do need to be "here's how we considered those things 2021-07-01 21:14:35 @lmeyerov @samth I agree, but that's the exact *point* of asking questions 2021-07-01 21:13:46 @lmeyerov @samth For released products that aren't papers, the responsibility to ask questions is much higher, because the product reaches real users immediately, and because lack of transparency means that we cannot just go find the answers ourselves 2021-07-01 21:12:10 @lmeyerov @samth Because some questions are informed by serious, serious mistakes in the past, and neglecting to answer them---even later, once asked, like in author responses---is a huge threat to the entire field 2021-07-01 21:11:09 @lmeyerov @samth I'm a very fair reviewer and spend like six hours on every review for every paper I reject making sure it's full of constructive feedback and pointers to things I think could help, pointing out everything promising about the work. But some questions need to be answered 2021-07-01 21:05:40 @kurtsy I am very upset that there are tool developers who don't feel this way, that scares me 2021-07-01 21:04:57 @kurtsy I think it's way, way more unethical for a tool to introduce a bug in someone's code than for someone to fail to find a bug in her own code 2021-07-01 21:00:37 @kurtsy And to anyone who cares about doing the right thing 2021-07-01 20:59:39 @lmeyerov @samth Everyone I've spoken to involved in the development of this has pointed out my concerns as *completely reasonable* so the weird attacks on me literally doing my job and questioning things so that we can make progress are not very reasonable 2021-07-01 20:56:18 @lmeyerov @samth Yeah we really ought to learn from standards that end in large-scale automated racism that's way better for the future than actually considering the impacts of our technologies before releasing them 2021-07-01 20:50:25 We've all brainwashed ourselves into thinking we should celebrate all new technologies by default but this is absurd 2021-07-01 20:49:28 @ian5v I did 2021-07-01 20:48:18 How is "maybe we should consider how our technologies impact humans and how to mitigate those concerns immediately, and expect that discussion to be upfront and transparent" at all controversial people scare me sometimes 2021-07-01 20:45:02 RT @SIGpwny: Experience the century to come at UIUCTF 2021! Take a well-needed break from the 2020s and jump into a premier CTF event betwe… 2021-07-01 20:43:43 @MichaelKup @kurtsy I would strongly guess that people tend to trust tools to be near perfect, and expect themselves to have bugs in their programs. I've definitely never met a programmer who expects her own code not to have bugs lmao 2021-07-01 20:40:34 @lmeyerov @satnam6502 I wouldn't signal them out, I'd reject any paper in which the authors refuse to discuss these things even in responses, given what we know now. It's extra important with a real tool reaching real users, and with no transparency into the dev process leading up to it 2021-07-01 20:38:49 RT @mgambogi: Talia’s been making this point all over but it’s a big one. And it applies outside of just dev tooling. People assume thing… 2021-07-01 20:38:35 @lmeyerov @satnam6502 Also, standard things like evaluation that doesn't just rely on just test I/O since in PL and SE it is widely accepted that this is really not sufficient to reason about tool correctness, and can mask bug introduction and so on. Also considered, apparently 2021-07-01 20:35:46 @lmeyerov @satnam6502 I'm told these are all things that have been considered, but we may not immediately get satisfying answers to them because companies are companies and censor everything important 2021-07-01 20:33:16 @lmeyerov @satnam6502 What are the ethical implications of a tool introducing bugs? Do humans trust the tool more than they trust humans, even when it introduces bugs? How do we signal not to trust it? How do we protect against data poisoning introducing security vulnerabilities? 2021-07-01 20:31:18 @samth The former is probably just a defensive fear response "I hope they're not coming for me too" 2021-07-01 20:30:36 @samth Disagree with the latter, it's exactly the kind of response researchers should have to any tool released without transparency into the process, *especially* when there are potential ethical implications that ought to be addressed sooner rather than later 2021-07-01 20:26:10 @samth Some people ask questions by stating concerns and expecting people to state things to counter them, it's just a conversation style thing IMO, I haven't seen any negative claims that weren't just extremely real concerns that warrant answers 2021-07-01 20:20:10 @samth Sometimes as experts we actually know the right questions to ask without yet knowing how things will go exactly because that's literally our job, and we shouldn't accept things uncritically until we have answers to those questions, or we are not doing our scientific duties 2021-07-01 20:18:16 @samth All of the questions I've asked so far have been extremely real concerns identified during development of the tool that are going to be discussed more publicly in more detail soon so sounds like healthy skepticism 2021-07-01 20:16:54 @Skiminok @HeidyKhlaaf I'm not sure I've heard about this in particular but maybe I missed it. I'd be very happy to see it 2021-07-01 20:12:54 (I admit I sometimes am skeptical in a way that isn't optimally kind, and I can work on that because it's important too) 2021-07-01 20:11:14 Please stay skeptical, please keep asking questions, please never accept a technology without criticism 2021-07-01 20:10:25 This is extra important when there isn't transparency involved in the development process, when PR and legal teams may gloss over very real issues to avoid getting sued 2021-07-01 20:09:05 Skepticism is how we identify the questions to ask that we need to answer to make progress. It's healthy. When I proposed my thesis topic to Andrew Appel a while ago, he said, "I'm skeptical," and I said, "well you should be, that's your job" 2021-07-01 20:07:54 I never want to be a part of a world in which computing technologies released without complete transparency into the internals and the development process aren't met with skepticism from computing researchers. Skepticism is important https://t.co/tUn6PAVltE 2021-07-01 20:06:10 @satnam6502 I've been told there will be some kind of post in the next few days addressing some of the skepticism and hinting toward future developments that can address it further, so the skepticism is contributing to healthy scientific dialog 2021-07-01 20:03:21 @satnam6502 It's a researcher's job to be skeptical, though. Skepticism is good. Uncritical acceptance of any technology is harmful to scientific progress 2021-07-01 19:57:53 @kurtsy This question is irrelevant. People tend to trust tools, so tool devs have a responsibility to not introduce new harm, or to adequately signal they should not be trusted 2021-07-01 19:42:56 @secretasianman @johnregehr @alastair_d_reid By making verification accessible, I don't mean expecting programmers to write specs from scratch, I doubt that will happen 2021-07-01 19:40:41 @SeanTAllen @johnregehr This would be irrelevant IMO. Programmers tend to trust tools more than themselves/other humans, IMO *any* bug suggested by a tool is a crisis, even if it's rare 2021-07-01 18:43:54 RT @normaldeviate: @TaliaRinger mao's slogan, "never forget the class struggle", is quoted in Gong's book on Java security. 2021-07-01 18:43:45 @normaldeviate Lmao 2021-07-01 18:34:31 @vanessainstem About a year after you've written the paper 2021-07-01 17:40:25 @_joaogui1 And the state monad is just as binding 2021-07-01 17:39:43 @_joaogui1 All those who seek freedom from state end up recreating it (by way of monads) 2021-07-01 17:37:54 RT @TechnoEmpress: @TaliaRinger https://t.co/J7E3NqW6mk I have stickers with this 2021-07-01 17:35:59 https://t.co/pOadnHuAoD 2021-07-01 17:35:45 https://t.co/Uu9tcyX3uj 2021-07-01 17:35:08 Properties are theft! 2021-07-01 17:32:50 @MonniauxD Yes a bit! I'm curious if it has seen success for this kind of thing yet 2021-07-01 17:31:46 @codydroux @harriesadam It's incredibly stressful, especially when you get so emotionally invested in the future of the field 2021-07-01 17:29:55 @codydroux @harriesadam Ah, do you even know me? 2021-07-01 17:29:17 @codydroux @bravenewprince @harriesadam @procmarco Hard problems are fun, though 2021-07-01 17:26:50 @codydroux @harriesadam ? 2021-07-01 16:27:25 @HeidyKhlaaf Wow that's fascinating, I wonder how that happened? Which vaccine did they get? That is surprising to me 2021-07-01 16:22:33 @HeidyKhlaaf Delta is scary but vaccines are still super protective against severe disease from it, no? Here in Israel at least they mostly seem worried about unvaccinated middle schoolers and Delta, but for vaccinated folks seem mostly unconcerned 2021-07-01 16:18:02 @HeidyKhlaaf Depends if you're vaccinated IMO 2021-07-01 16:16:42 Had to get an antibody test in Israel after arriving to shorten the isolation period, and it's cool to see concrete evidence of the success of the vaccine (I do have antibodies, levels are extremely high TBH) 2021-07-01 16:04:20 @harriesadam And we are actively working on a good decompiler for Coq, so hopefully that helps 2021-07-01 16:04:01 @harriesadam Yes, I think I saw a paper or two on this in other proof assistants. I think it's quite tractable for proof terms, but not for automation. However with good decompilation we could get tactic proofs from generated terms! 2021-07-01 16:00:54 @harriesadam @procmarco Yes, though if proof assistants could call out to verified libraries in PAIR or whatever we call it, that could work OK 2021-07-01 15:55:43 @harriesadam Should be fairly doable at the term level for most terms, but existing automation would be hard to port 2021-07-01 15:53:58 RT @Chacelounge: This looks bad https://t.co/bdp6aLkG5n 2021-07-01 15:50:42 @procmarco @harriesadam Yeah basically 2021-07-01 15:49:32 I mean yeah she's right this will still be a problem in a few days but also I care so much about this area I want things to be usable like right now 2021-07-01 15:48:41 Mom: What's wrong? Me: Well now that machine learning tools that may suggest buggy code are being released, we need to make verification easy as soon as we can so that we can make sure the suggestions are correct, and I can't stop thinking about it. Mom: Just for 1 or 2 days? 2021-07-01 15:38:16 @harriesadam Same with libraries, like I'd like to just get "mathlib but in Coq" fairly easily 2021-07-01 15:34:42 @harriesadam I'm not sure, it's a pretty open-ended question. I'd just like to build proof automation that is easy to port between proof assistants 2021-07-01 15:15:35 I really want to build some common infrastructure for automation and UX in both Lean and Coq. Is there a way to do this? I don't see much value in community segmentation between Lean and Coq, and I really want more crossover 2021-07-01 13:57:00 This was 2012 though. Alas, could have done this 9 years earlier 2021-07-01 13:54:46 Empathy encoded in the leadership principles was my parting wish at Amazon haha. Even wrote about it as part of a long email to Jeff Bezos https://t.co/mgdbyiXPpQ 2021-07-01 13:43:57 @vardi Like page 1 of my dissertation quotes it as inspiration haha https://t.co/YU89v7kNaf 2021-07-01 13:32:59 @vardi I think a lot of the costs of verification are not fundamental, and can be eliminated by way of better automation and user experience going forward. Social Processes was a great paper for future work ideas 2021-07-01 13:28:44 @vardi Do you think Hoare was influenced by Turing's 1948 paper? I was always curious whether Turing's paper was just too early, or whether it directly influenced later work on verification 2021-07-01 13:28:02 RT @vardi: By the mid-1990s, several model checkers had been built and adopted for industrial usage by semiconductor and design-automation… 2021-07-01 13:21:25 Honestly it's just so freaking cool that my only living grandparent survived the pandemic and I get to see her again every day for the next couple of weeks 2021-07-01 13:15:52 @schock Congrats! Also good for the world TBH 2021-07-01 13:15:31 RT @schock: Hello twitter friends, it's my last day at MIT. I'm excited to share that I'll be joining the Algorithmic Justice League (https… 2021-07-01 13:13:41 RT @jin_diesel: LinkedIn hacked again — personal info of 756 million users leaked https://t.co/c7JbW1WmW2 2021-07-01 12:41:24 RT @yihongz_bot: @TaliaRinger Even according to OOP, private properties should not be inherited. 2021-07-01 12:41:18 RT @_ntietz: @TaliaRinger Capitalism Java Too many factories 2021-07-01 12:10:55 RT @LauraViglioni: Communists functional programmers 2021-07-01 11:52:20 RT @Teddytroll: @TaliaRinger https://t.co/Dm5qqDKo4a 2021-07-01 11:38:51 @LauraViglioni Destroy inheritance 2021-07-01 11:38:40 @LauraViglioni Abolish the class hierarchy 2021-07-01 11:34:56 This Tweet is an invitation 2021-07-01 11:31:49 How has object-oriented programming terminology not led to socialist puns 2021-07-01 11:30:42 RT @zeigenvector: Who called it “turn off your air conditioners because otherwise there’s going to be a black out” and not “only fans”? 2021-07-01 11:30:23 RT @rachelmetz: if you've ever wondered how extensively the US government uses facial-recognition software, the answer is ... yes! yesterda… 2021-07-01 11:27:56 @HilaCodes @existsShachar That sounds absolutely wonderful. I'm around this upcoming Tuesday! I'd be really happy to stop by 2021-07-01 11:06:15 @HilaCodes Are you in Israel now? I'm in town until July 15th, would love to see you 2021-07-01 10:42:12 אכלתי את זה כשהייתי ילדה 2021-07-01 10:41:37 אני לא חושבת שיש משהו כזה בארה"ב 2021-07-01 10:39:29 גומיגם https://t.co/hKWnJeL7sO 2021-07-01 09:47:35 @Yasmeeno84 Congrats!!!! 2021-07-01 09:47:28 RT @Yasmeeno84: الحمدلله حتى يبلغ الحمد منتهاه بدأت هذه الرحلة بشتاء ٢٠١٠ و انتهت بصيف ٢٠٢١ بحصولي على درجة الدكتوراة بتخصص أمن المعلوم… 2021-07-01 09:45:49 @ian5v Hope I'm not giving the NSA any ideas lol 2021-07-01 09:38:50 Good to watch for the sake of knowing what elegant introduction of vulnerabilities looked like in the past: https://t.co/pyOIoJa7Mg 2021-07-01 09:37:39 Will be way harder to trace this time! 2021-07-01 09:37:12 Juniper Incident analog, but distributed across lots of companies and done by way of data poisoning of neural synthesis tools. Calling it in 2023 hahaha https://t.co/0ODiKhPh5Q 2021-07-01 09:32:33 @znmeb Oh and the NSA is going to love the opportunity to slowly poison datasets rather than have to ask for sequences of commits that introduce vulnerabilities (Juniper Incident) 2021-07-01 09:30:33 @znmeb Alex seems less worried about this, and has used the tool, so maybe they do an OK job at signaling "do not trust me" as is! I hope so 2021-07-01 09:29:20 @znmeb So I'd rather we put in the right HCI work to prevent that scenario before releasing this kind of tool, then invest much more in neurosymbolic methods and slowly remove the "do not trust me" user experience as warranted by improvements in technology 2021-07-01 09:27:45 @znmeb I think it's true, but my worry, which Alex does not seem to share, is that until we have good neurosymbolic tools, people are going to trust tools that suggest syntactically plausible but incorrect or even harmful code, rather than bother correcting them 2021-07-01 09:25:58 @krismicinski @0x21376B00 But it's rare for Alex to be needlessly optimistic, so I'll weigh his optimism accordingly and be a bit less pessimistic haha 2021-07-01 09:24:40 @krismicinski @0x21376B00 I agree with Alex on the conclusion that this is an opportunity for neurosymbolic methods. He's notably more optimistic than I am that 1) this won't harm people in the meantime, and 2) funding agencies and companies will see eye to eye with us on that 2021-07-01 08:55:26 @Skiminok I'd really love to see a user study of people trusting Copilot suggestions to not be buggy when told 1) they came from Copilot, or 2) they came from a human 2021-07-01 08:54:31 @Skiminok Hmm, on this point though, people have a tendency to trust machines more than themselves sometimes (consider the field of human-robot interaction). Are you worried about people using buggy generated code, but assuming it is correct since it came from a tool? 2021-07-01 08:52:28 I always value Alex's perspective on this work. He's notably both more optimistic and more informed than I am. Recommend reading the whole thread, not just the first Tweet https://t.co/RMo9xptqLX 2021-07-01 08:49:31 @Skiminok Hoping to organize a machine learning & 2021-07-01 08:48:20 @Skiminok I always love your perspective on this. I'm glad you're optimistic 2021-07-01 08:47:49 RT @Skiminok: Neural code generation models learn a *contextualized language space* for us, in which we now apply PL analysis/synthesis met… 2021-07-01 08:47:34 RT @Skiminok: Let me be precise. If your context/problem are unique, you will struggle with bugs in the current Copilot tech. And you will… 2021-07-01 08:05:54 @kazarmy It depends so much on the technology involved! But many technologies right now will not be accurate, and will also not be able to explain anything to you about why they suggested what they suggested 2021-07-01 08:02:26 If you're ADHD and my student, I'll see what I can do to advocate for you and help you get what you need 2021-07-01 08:00:46 True both for students and for faculty. If you're ADHD incoming faculty, and out about your disability, I strongly recommended negotiating either more administrative attention, or a bump in startup sufficient to hire your own part-time admin for planning & 2021-07-01 07:59:27 Things I think could help: - monthly or quarterly deadlines for the same venue, so that less rides on making a particular deadline, especially for more focused work that may not fit into other venues - disability accommodations that include help with executive functioning 2021-07-01 07:57:05 I think part of what bugs me about the "don't glorify overwork" thing is that overwork does fall out of the incentives set for academia sometimes, and it does so disproportionately for people with ADHD https://t.co/RVNV6XAPOl 2021-07-01 07:49:43 ADHD also makes it harder to plan, which makes it harder to enforce your own structure on unstructured schedules 2021-07-01 07:48:35 For NT folks: ADHD makes this hard because sometimes you get so into what you're doing, you can't get yourself to stop, so you may suddenly find that it's 2AM and you're still at work. This compounded by the fact that folks with ADHD disproportionately have DSPS 2021-07-01 07:46:51 Also I should note for the sake of not glorifying overwork that there are people who manage to maintain healthy sleep schedules during graduate school. I was not one of them. ADHD makes this very hard, and throw a pandemic on top of that and forget it 2021-07-01 07:42:14 My family was worried that if I slept too early I'd wake up too early, clearly not realizing that I have six years of sleep to catch up on hahaha 2021-07-01 07:07:09 אני רוצה לישון כל היום וכל הלילה 2021-07-01 06:53:17 So I know the analogous research in human-robot interaction. Do we have research on this yet for generated code? Some great potential user studies here https://t.co/D5P6vhPVxv 2021-07-01 06:50:24 @silvascientist I think they will, though, and this is human psychology. Human-robot interaction deals with this problem extensively. It also gives me faith that there are ways to build tools that deliberately appear less trustworthy 2021-07-01 06:47:40 There are incredibly good people thinking about and working on these important problems, I just hope they get the support and attention they deserve 2021-07-01 06:46:15 That and we need significant research on the HCI side of things for how to get a tool to adequately signal "please don't trust me" 2021-07-01 06:44:59 I think one thing that scares me so much about bugs in generated code is the way that programmers tend to trust tools by default, much more than they ought to sometimes. It feels like as tool developers, we have a very large responsibility to be correct 2021-07-01 06:34:07 RT @HeidyKhlaaf: I know folks have a lot of question about its limitations, PL capabilities, and safety implications, hopefully that's to c… 2021-07-01 06:17:13 RT @tian_yi_zhang: I will continue working on the intersection of SE, PL, and HCI at Purdue while also expanding my research and exploring… 2021-07-01 06:16:37 RT @tian_yi_zhang: Some (belated) personal news: Today is my last day at @HarvardHci after two wonderful years of postdoc with @roboticwres… 2021-07-01 06:06:02 RT @ElijahBoumaSims: @TaliaRinger My desire to be an academic is increasingly represented by this meme: https://t.co/7HjNzi4Twb 2021-07-01 01:34:26 RT @FrostScience: Our #LGBTQ+ scientist spotlight on the last day of #PrideMonth is Alan Turing. Considered the founder of computer science… 2021-07-01 01:25:54 RT @Russia_NC: https://t.co/OylHvo9F4N 2021-06-30 17:51:45 Anyways it's all depressing, and I don't want to hear about machine learning except as immediately relevant to my work for a while 2021-06-30 17:49:59 For companies building products, I guess they'll do whatever they want, it's annoying that the appropriate social norms don't extend to that 2021-06-30 17:48:38 For researchers, that's a matter of basic academic integrity 2021-06-30 17:42:09 Anyways, I don't want to be there for that. I'll do my best to connect the communities *now* rather than later but it's aggravating that this effort has to come from within PL, and that ML folks largely don't bother to talk to the communities they colonize and solicit their ideas 2021-06-30 17:38:14 By "becoming NLP" I mean replacing a whole bunch of techniques so folks lose funding and jobs for like a decade, then after hitting a wall with existing techniques, the ideas of those people already harmed reintegrated into the technologies. This is so harmful and it scares me 2021-06-30 17:35:47 I'm confused how it happens except by way of sheer size and power given how many readily admit to wanting to actually meet in the middle and collaborate and exchange knowledge. But I think it has to do with incentives to make money quickly or something 2021-06-30 17:34:07 The thought of PL becoming NLP, all of us losing our jobs, and then our knowledge later being reintegrated without credit is giving me deep, dark, unpleasant feelings I don't wish to visit at all, and I wish ML folks would stop effectively colonizing CS sometimes 2021-06-30 17:32:13 Arrived, slept 3 hours, woke up to my parents and grandma and uncle and cousins and their dog, a frenzy of hummus and gifts and cookies, back to bed now. It's fun, and I think for my own sake I really need to tune out of CS related things for a bit, except specific collaborations 2021-06-30 17:28:36 Right this belongs in main thread https://t.co/8QXlWYadNB 2021-06-30 17:15:52 @clegoues @acherm @Miles_Brundage Yeah the test-driven concern (possibly even introducing bugs because of untested functionality) is orthogonal to the vulnerability concern (a particular weakness of NN techniques) though I imagine both will compound each other in complicated ways 2021-06-30 15:50:52 @clegoues @acherm @Miles_Brundage Vulnerabilities are a particular concern for neural net techniques, and a reference in particular to attacks I linked to. This concern is orthogonal to the lack of meaning of just testing a tool on a suite of tests 2021-06-30 11:27:06 I remember Hebrew just well enough for my grandma to be able to point out in Hebrew that I have a PhD and I should be practicing Hebrew now whenever I make a grammatical mistake or forget a word lmao 2021-06-30 11:02:11 @BowToChris Love doesn't pay the bills 2021-06-30 10:51:04 @acherm @Miles_Brundage https://t.co/g8LXg0q8Rn 2021-06-30 10:49:31 @alcides @Miles_Brundage I'm asking about UX, it's not enough to just trust users to validate suggestions without also adequately signaling they are suggestions and could introduce bugs and/or vulnerabilities. This question is not for you, it is for Miles 2021-06-30 10:45:26 @alcides @Miles_Brundage Definitely don't tell me how program repair tools work lol 2021-06-30 10:44:37 @Miles_Brundage I just hope you spend a lot of time talking to lots of PL and SE experts about this kind of stuff 2021-06-30 10:40:32 @Miles_Brundage And probably my biggest worry was the note evaluating it based on test suite results. SE made this mistake with program repair, you can mislead yourself for *years* this way (but positive human feedback, which it sounds like you have, helps prevent that) https://t.co/Bn947S5cjT 2021-06-30 10:38:27 @Miles_Brundage See, for example (for attacks) https://t.co/NMzCxWxQhJ 2021-06-30 10:37:37 @Miles_Brundage Sorry for the immediate bombardment with questions, this just worries me a lot, there is consensus in the PL and SE communities that tests alone are not sufficiently informative for a tool like this. A human in the loop is, which you do :) but in that case, UX matters a ton 2021-06-30 10:35:19 @Miles_Brundage There are also a number of attacks recently poisoning datasets by introducing malicious code that, when trained on, will introduce security vulnerabilities. Do you protect against these attacks? How? 2021-06-30 10:34:28 @Miles_Brundage Program repair tools evaluated on inout/output tests often pass, but actually *introduce* bugs. How do you either avoid introducing bugs, or signal adequately to the user that a suggestion is nothing but a suggestion, and could be dangerous? 2021-06-30 09:59:10 הגעתי Let's hang out :) 2021-06-30 09:51:21 I like how all of you have slowly amended this to "I feel pretty depressed" haha 2021-06-30 04:55:01 I feel pretty depressed about the future of computing 2021-06-30 04:53:12 RT @RandomlyWalking: @TaliaRinger Oh there’s a paper on this https://t.co/14VO5kUxxb 2021-06-30 01:49:13 @EdNutting I think they're not ignored now, in that they can't be ignored---everyone who has been paying attention knows. But that doesn't mean that people are actually listening to the people putting out these ideas at all. More like active resistance 2021-06-30 00:44:49 For how many years were warnings of bias in and harm resulting from machine learning tools without proper protections ignored in pursuit of profit? This is a real question, so I know how to go about politicking in ML for PL 2021-06-30 00:41:18 @SouthernSummr ... because otherwise companies aren't going to bother, since they don't mind harming people, as evidence by *points to the last ten years of machine learning results in industry* 2021-06-30 00:38:21 @SouthernSummr These aren't things we don't know how to do. 2021-06-30 00:38:10 @SouthernSummr "human-in-the-loop with a UX that very clearly signals there is a risk of introducing bugs without intervention, verification, whatever" 2021-06-30 00:34:30 @ncooper57 I feel much more strongly and negatively about tools introducing new bugs and vulnerabilities in other people's code than I do about simple negligence on the part of the person developing the code 2021-06-30 00:29:08 @mmitchell_ai Who resigned? 2021-06-30 00:26:29 @PTOOP The Biden administration has a lot of really good tech policy folks advising, so if we want to act at least in the US we need to do it before Biden is out of office 2021-06-30 00:24:54 @PTOOP The right regulations will lead to investment in making technologies for certainty and good UX for user intervention just by virtue of it becoming necessary for growth 2021-06-30 00:24:10 @PTOOP We need to add the right regulations now before people get carried away using harmful technologies in our field too, and then spend the time making things cheap to do well, so that companies can continue to grow 2021-06-30 00:23:16 @PTOOP Regulations exist in most other fields, and when designed well help with many serious ethical concerns. Computer science has just gotten away with so little regulation, and it shouldn't have 2021-06-30 00:18:05 @PTOOP Yeah if you've had a heart attack you've probably tried a lot and might just want to chill and focus on your work. I'm young and healthy and for sure I'm going to use my influence to talk to people with any say about tech policy on this 2021-06-30 00:16:37 @PTOOP Those of us who understand the risks have a moral responsibility to get involved in the politics 2021-06-30 00:16:02 @PTOOP The disagreement modality 2021-06-30 00:15:05 @PTOOP We should make it much easier to be right. But we should also hold people responsible for unethical tool development. 2021-06-30 00:14:20 @PTOOP Oh I disagree. The machine learning world has shown that people don't really care much about doing harm if it means making money 2021-06-30 00:13:27 @PTOOP Paging @geomblog, who has a say on tech policy 2021-06-30 00:09:30 Regulators IMO should make tool developers liable for security bugs that our tools introduce in code in the absence of adequate protections (human-in-the-loop with a UX that very clearly signals there is a risk of introducing bugs without intervention, verification, whatever) 2021-06-30 00:07:04 RT @moyix: Summer afternoon project: fuzzing neural-network generated C code 2021-06-29 23:56:19 @catnaroek Eh, code is an explanation. Just, it might be totally wrong 2021-06-29 23:54:36 Important for anyone who develops or evaluates one of these tools https://t.co/EtInf4aYNT 2021-06-29 23:53:06 @duns_sc0tus @IgorBrigadir Well, I don't think we should have to be exact, I think the computer should be able to ask us the right clarifying questions when we're not exact 2021-06-29 23:48:52 @duns_sc0tus @IgorBrigadir Long term this is a promising route. Doing this without significant input from the PL and SE communities is a terrible idea and will end in pain for everyone 2021-06-29 23:47:20 Just now that this is happening, I don't want you all to waste 10 years chasing unhelpful results when it's completely avoidable, because we've already made these mistakes and this is how research works 2021-06-29 23:46:24 Likewise if someone submits a machine learning for programming paper to a machine learning conference or whatever, and the benchmarks are just plain old input/output tests, no enhancement of the suite, no specs, no human, IMO reject that shit---these are not meaningful results 2021-06-29 23:44:51 You either need specifications or a human in the loop or something to make sure you're not doing something not only unhelpful, but also outright harmful. The Github tool is still in safe territory because there's a human in the loop, but I hope it's honest about this possibility 2021-06-29 23:43:54 Program repair encountered this crisis a while ago: https://t.co/0eVrlFQp0r 2021-06-29 23:43:07 Right OK so if any of my friends go off making tools that generate or repair code, you should know that running a test suite (perhaps unless strongly augmented) on just code inputs and outputs is absolutely not enough to say anything about the tool---your tool may introduce bugs https://t.co/zQavMcTbBD 2021-06-29 23:40:37 @AlejandroPiad Code polarization 2021-06-29 23:35:33 RT @AlejandroPiad: @TaliaRinger Or when hackers push millions of lines of code smartly crafted to inject specific bugs into the model! 2021-06-29 23:35:30 @AlejandroPiad Oh my gosh this sounds fun 2021-06-29 23:35:00 @IgorBrigadir I like the idea of AI code autocomplete but I really don't think the technology is there to do it well, and I'm super super worried about the impacts of doing it preemptively, but I guess we'll see 2021-06-29 23:33:56 @IgorBrigadir Yes and then evaluate it on an underspecified input-output test suite and say it performs well and ignore the whole PL and SE communities when we say that's a bad idea and we've already tried that and it leads to tools that introduce new bugs 2021-06-29 23:31:07 Can't wait until a naive language model suggests the most popular security vulnerabilities to autocomplete your code 2021-06-29 23:23:35 RT @lastpositivist: Anyway the Soviet Union was so advanced they already did our culture war 50 years ago. This is why they won the space r… 2021-06-29 23:23:17 RT @lastpositivist: lmao just got sent an Anderson paper that begins with this little zinger. I know this is weird but everything I encount… 2021-06-29 23:06:59 @CheeToS_ Is having it sorted out just opening the bathrooms to everyone because IMO that's what make sense. Bizarre to me any other way 2021-06-29 19:18:33 RT @asajeffrey: @larsberg_ Does the AI report things like "I was only trained on MIT-licensed codesets, feel free to check in any generated… 2021-06-29 19:18:00 RT @larsberg_: Very curious about the license situation! E.g., is it like stackoverflow and CC-BY-SA and requires attribution? Or is the in… 2021-06-29 18:56:40 @_protz_ Man I simultaneously hate this functionality and want this functionality 2021-06-29 18:47:56 @PTOOP Part of my deliberate coupling of TT and ML is a belief that they will help each other significantly, but part of it is also a belief that if people don't realize this soon, the same thing will happen to us 2021-06-29 18:46:53 @PTOOP I think the worry is if people lose jobs during that time, only for their knowledge to later be valued. I think about symbolic AI and about classic NLP, linguistics, and neuroscience 2021-06-29 18:43:46 The whole thing makes me anxious, and I just try to close my eyes and pretend that capitalism isn't a thing 2021-06-29 18:42:15 But then when industry gets buy-in they often make things secret for a competitive advantage, so it's hard to make sure they're talking to the right people to develop something that won't mislead people and take away jobs only to later reveal serious limitations 2021-06-29 18:39:26 But it's really hard to do that without support from a large team of engineers, so it's all kind of unfortunate 2021-06-29 18:35:46 This is one of my biggest anxieties about the future of programming, and I sometimes feel like I ought to pour time into racing the approaches I know will have limitations and keeping up so that people don't falsely buy into those systems as superior 2021-06-29 18:34:17 I don't often feel afraid that all of the interesting problems will be solved. I do often feel afraid that people will falsely believe all of the interesting problems are solved only to realize the fundamental limitations to their approaches 10 years later, after serious damage 2021-06-29 18:26:27 @MarisaVeryMoe I'd prefer stalls to astroturf though 2021-06-29 18:11:31 @lambdatotoro Ah yes, the three genders 2021-06-29 18:06:48 Why is gender so hard 2021-06-29 18:06:42 Update: when choosing between the bathroom for men, women, and animals, I wandered into the animals one 2021-06-29 17:57:38 @PTOOP @andrejbauer I'm fine with that when it's done in a way such that you don't lose consistency with univalence. Consistency with univalence is really useful for building automation even externally 2021-06-29 17:50:14 @PTOOP @andrejbauer I think provable UIP is nice, but I don't like the general principle, both for philosophical reasons and for practical reasons (inconsistency with univalence, which I think has more long-term potential for building good automation) 2021-06-29 17:47:30 @ilyasergey @icfp_conference Well gee Ilya it's not like it's the International Conference on Reusable Programming now is it 2021-06-29 17:45:21 @andrejbauer UIP is sad, though 2021-06-29 17:45:02 @andrejbauer Undecidable definitional equality 2021-06-29 17:43:17 @PTOOP Ah so what I'm asking for just amounts to primitive projections and nothing more 2021-06-29 17:36:10 @PTOOP Maybe that itself isn't decidable though lol 2021-06-29 17:35:42 @PTOOP I think I'd like to know the class of types for which we get decidability, and the class for which we lose it. Then one could in theory have good ways to expand definitional equality while sometimes providing propositional hints 2021-06-29 17:27:51 @PTOOP But beta/iota don't feel annoying to me, they seem to make sense 2021-06-29 17:27:25 @PTOOP Ugh what should we do 2021-06-29 17:18:17 @PTOOP I feel weird about classifying something with well behaved definitional eta the same way as something with poorly behaved definitional eta 2021-06-29 17:16:46 @PTOOP I just want to know *how much* and why, I think 2021-06-29 17:11:46 RT @josecastillo: #Pride2021 advice: do what the fuck you want. Your queer sexuality is good and beautiful and full of a hell of a lot more… 2021-06-29 17:10:22 Why is coordination so hard 2021-06-29 17:10:03 Update: was reviewing a paper and drinking tea on the plane, and spilled the entire cup of hot tea on my lap 2021-06-29 16:57:21 Thanks to Seattle for the practice https://t.co/egNg8kJPSM 2021-06-29 15:56:28 @codydroux I'm always a fan of reification 2021-06-29 15:52:40 Like something about eta feels quite fundamental to equality, it's very sad to add a new inductive type but lose something so fundamental 2021-06-29 15:49:40 @codydroux I have nothing useful to contribute, though, haha 2021-06-29 15:49:27 @codydroux That feels unsatisfying to me 2021-06-29 15:48:34 @codydroux They're the bad boys of the type theory world 2021-06-29 15:47:17 @codydroux I think they're cool, but I don't think they're nice 2021-06-29 15:45:59 It almost feels like if eta is poorly behaved, maybe our notion of inductive types is just wrong 2021-06-29 15:45:03 It's sad to me that eta doesn't always behave nicely 2021-06-29 15:44:29 Definitional eta simplifies reasoning for a lot of types, so I'd love to know the dependently typed intensional type theory with definitional eta as often as possible for inductive types 2021-06-29 15:42:01 Yes this is what I think about on airplanes deal with it 2021-06-29 15:41:20 Sigma and records seem OK? Anything more interesting? 2021-06-29 15:40:50 Definitional eta for inductive types in general makes a type theory extensional, right? Is there a well-defined class of inductive types for which definitional eta does not make a type theory extensional? 2021-06-29 15:37:38 Categories with Friends 2021-06-29 12:55:29 It's an LGBT programming languages tattoo, and more broadly a symbol of being authentic to myself 2021-06-29 12:55:06 The lambda tattoo is one I got before grad school. I'd just come out for the second time as bi, and also decided to go to grad school. Lambda is a historical gay rights symbol and a symbol of my research area. Purple gradient in the middle for both the bi flag and UW colors 2021-06-29 12:51:36 @almostmeenal But professors are just normal people who get weirdly obsessed with some topic and want to study it very deeply. Wide range of people but lots of very cool humans 2021-06-29 12:50:46 @almostmeenal I got the tattoo before starting grad school, it's a combination PL and LGBT tattoo. Purple gradient in the middle for both UW colors and the bi flag it was the time I decided both to come out for the second time, and to pursue my dream job 2021-06-29 12:48:53 @asadjaved889 You can apply through the standard UIUC application process, but if you're interested in my work then it makes sense to talk about that in your statement 2021-06-29 12:46:34 @EliSennesh @NuritPeleg Technically we're also in space 2021-06-29 12:45:20 RT @typer_uma: Queer icon https://t.co/SIFjzb0tdO 2021-06-29 12:43:44 @EliSennesh @NuritPeleg It was over 40C in Seattle. Nothing over 40C should exist 2021-06-29 12:17:16 @almostmeenal Yesss 2021-06-29 12:10:51 I'm so excited for the loud complaining and the aggressive insertion of opinions into all conversations it's like white noise after growing up in an Israeli family, and I find it genuinely soothing, like I can let my guard down and be myself 2021-06-29 12:05:16 @Arvind_0602 Yeah, I usually don't wake up at 3 AM, just have to catch an early international flight. 4 hours is the minimum I can handle, but I usually go for 8 2021-06-29 12:03:44 When I was younger this stuff used to set me into total panic, honestly, but I'm secure enough now to just kind of shrug it off as I wipe up the entire contents of a @Starbucks Grande Pink Drink from the airport floor with some napkins 2021-06-29 12:01:38 @Arvind_0602 I think a little after 11 PM? I didn't expect a full night of sleep though. I'll try to sleep on the flight a bit 2021-06-29 12:00:15 I'm sticky and smell like strawberries. Accidental perfume I guess 2021-06-29 11:59:40 Solid start: - wake up at 3 AM - snooze but accidentally set alarm to 3:20 PM - wake up at 3:40 AM - forget to take ADHD medicine - miss shuttle, walk to airport - check in - buy refreshing cold beverage - spill refreshing cold beverage everywhere 2021-06-29 11:47:28 @larsr_h It's a kippah 2021-06-29 11:33:18 On my way to Israel to see my family there for the first time since January 2019! 2021-06-29 05:37:05 @sonicisdead I was genuinely mad, I was like, "smanpeko can't be more than smanyonyo, because smanyonyo is more than whatever you say, so if you say smanpeko, smanyonyo is more than smanpeko" 2021-06-29 05:35:57 @sonicisdead When I was six my best friend and I made up the word "smanyonyo" which we defined to mean "more than whatever you say" so that we could win arguments with everyone. It worked quite well until another friend defined "smanpeko" to mean "more than smanyonyo" 2021-06-29 05:17:04 @wilton_quinn Petition to rename "computer" to "compoopter" temporarily denied---the committee is too pooped to vote 2021-06-29 05:15:36 @wilton_quinn ... All opposed? 2021-06-29 05:13:21 @wilton_quinn All in favor? 2021-06-29 05:10:30 People think I reveal everything I think, but in reality they just underestimate how chaotic my mind can be 2021-06-29 04:57:03 I'm an adult 2021-06-29 04:55:24 Don't tell anyone but for six whole years during my PhD every time someone said "computer" I thought "compoopter" and laughed in my head and never revealed this to anyone ever 2021-06-29 04:52:31 @BowToChris I hate almost nobody haha 2021-06-29 04:50:18 @nzqrcs It's a hard sport for sure. I couldn't run 38:56 these days anymore. I was training very hard back then, running 75 miles per week 2021-06-29 04:42:30 Our team swept the race, all top 8 places, because we were all good at suffering I guess. Good times 2021-06-29 04:42:02 In the end I finished in like 41 minutes or something, it was not at all anything close to good for me back then. But I finished in eighth place. So I scored a single point at nationals. I was so proud of myself 2021-06-29 04:41:06 So I suffered really hard until I passed her, and then I kept suffering to make sure she didn't pass me. My friends were cheering really hard, it felt really good, or at least as good as it can when you're flopping around a track 25 times in 100 degree weather as fast as you can 2021-06-29 04:39:14 But then something weird happened. The girl in front of me, the girl from another team, sure started to fade, and she got closer and closer to me. And I thought, holy shit, I can actually score a point 2021-06-29 04:38:15 So I stopped thinking about times and just raced. It was slow and terrible, and I spend half of the race squeezing ice water sponges down my sports bra. People lapped me and I pretended we were on the same lap, tried to use them as rabbits for a bit. I was in last place 2021-06-29 04:36:37 8 people score at nationals. So this was an opportunity for me. I've never been a national-level athlete, but I've always been really, really good at suffering. If I could suffer better than just one person, maybe I could score a point 2021-06-29 04:35:33 Only 9 of us were crazy enough to run it in 100+ degree weather. 8 from my club. 2021-06-29 04:34:18 I don't know if you've ever run a track 10K, but honestly it turns out it sucks. You have way too many chances to drop out, and the fastest athletes might lap you 2021-06-29 04:33:00 It was club nationals, though, so we obviously didn't do that. My coach had me pack my hat with ice and wear it on my head, so that the ice would melt all over my face. The officials set up misters and fans and sponges in ice water every 200 meters, as governed by USATF rules 2021-06-29 04:31:31 Come day of, though, it was over 100 degrees. We had a big heat warning and everything. It said to avoid all outdoor exercise 2021-06-29 04:30:52 I signed up for the 10K. The track 10K. My first track 10K ever. I really wanted to try it, I was very excited. My road PR back then was 38:56, and I was hoping I'd be able to run a similar time 2021-06-29 04:29:55 This was USATF club track nationals. I forget the year, but it was in Tacoma, WA, which meant my club could send pretty much any of us who wanted to race. Why not 2021-06-29 04:28:37 In the spirit of it being too hot to exist, a quick story about the first and probably the last time I ever scored points at a national track & 2021-06-29 04:24:16 You'd think at one point I'd stop assuming everyone hates me but it's hard 2021-06-29 04:23:25 My RSD is so strong that I thought a well-known researcher in my area hated me for months but he actually admires me and shares the same long-term vision for the future of the field. lmao 2021-06-29 04:07:35 RT @Antipodal_Ivo: @TaliaRinger New professor just dropped comes equipped with a calculus 2021-06-29 03:26:07 RT @Weather_West: Portland has now officially hit 115F (!!!). Salem, OR hit 117 earlier (all-time record) 2021-06-29 03:25:28 @janadoppa Yeah, this is not an undergraduate I worked with at UW though, just someone I relate to 2021-06-29 02:41:41 I think my room is 110 degrees or more right now. It hurts my skin to stand in my room, and when I walk downstairs to the 90 degree living room, it feels cool 2021-06-29 01:54:02 Thanks to my parents for paying for the photoshoot :) 2021-06-29 01:49:39 @palvaro If I had to describe myself in a single photo, this would be it. It's the most "me" I've ever seen captured, though maybe a bit less awkward than usual 2021-06-29 01:40:17 @palvaro It was so hard to explain to the photographer what the lambda tattoo meant and why I needed photos with it haha 2021-06-29 01:26:56 It was a nice way to say goodbye to UW, all things considered. Onward to professor life! https://t.co/Ff9P07QQNE 2021-06-29 01:25:47 No applause, and no hooding, but everyone who passed by yelled "congratulations!" so the smile here is genuine https://t.co/JC5k1rKNpd 2021-06-29 01:24:30 I celebrated this last week with a photoshoot, since I didn't get to walk for graduation this year https://t.co/bJeQlYsucU 2021-06-29 01:21:37 I'm a computer doctor https://t.co/bN3WtT3gLy 2021-06-29 00:37:58 @TobyMeadows If interested. Feels like asking for free labor though. They should be paid for this kind of thing IMO 2021-06-29 00:33:13 @sibinmohan Yes I only want students to work with me if they're convinced it's what is best for them, given the best options they could possibly have 2021-06-28 23:27:34 @HerrDreyer @stuckintheory @samth @jonmsterling @jer_gib I have very high risk tolerance in general, and lack fear compared to most people, and I still struggled a lot with this internally. I can understand a lot of students would want to take a more standard route 2021-06-28 14:59:52 RT @KSeattleWeather: Officially 90 degrees at Sea-Tac. At 8 am. 2021-06-28 14:59:48 RT @KSeattleWeather: HISTORIC HEAT WAVE PEAKS TODAY High temperatures are likely to top out around 110 degrees late this afternoon in Seat… 2021-06-28 05:30:22 RT @Halfastorian: @EricHolthaus @Noahpinion https://t.co/wAZch5hak9 2021-06-27 20:11:28 RT @djg98115: After a day of intermittent effort (?), I am proud (??) to provide some original (???) comedic (????) to get us through the S… 2021-06-27 20:11:07 a double dose of hot pandemic humor https://t.co/QLQ8sCgt7x 2021-06-27 20:00:32 RT @dinoman_j: https://t.co/XedlU2QOki 2021-06-27 19:47:22 @mmitchell_ai yayyyyyyy https://t.co/NQ4G6KoqyA 2021-06-27 19:02:49 This is reasonable though https://t.co/8d8ySoTS2O 2021-06-27 19:01:51 RT @reduct_rs: I Am Not Defined by My Failures Despite the Trombonist Who Follows Me Around Making the Womp-Womp Sound Whenever My Commit i… 2021-06-27 19:01:39 RT @reduct_rs: Influencer? This Woman Spends All Day on Twitter and Has Not Committed Anything to the Repo in Weeks 2021-06-27 18:35:48 OTOH I love that there are so many of you at the ND-queer intersection. Hi! You're fantastic 2021-06-27 18:34:10 @evijitghosh Haha sorry 2021-06-27 18:31:26 @wilton_quinn I'd be interested in broader population studies about this, but couldn't find any when I last tried 2021-06-27 18:28:25 I don't know what to infer from this, but I find it interesting 2021-06-27 18:26:35 Half of neurodivergent folks who voted are queer, in contrast with fewer than a quarter of neurotypical folks who voted https://t.co/g4NkWjQnUA 2021-06-27 18:16:10 @gwillen But also like as an ADHD person, slow exposure to sun resulting in a natural tan is going to protect me way more than sunscreen with no tan because I'd have to legit remember to bring sunscreen everywhere, and if I mess up once I'd burn from lack of a tan haha 2021-06-27 18:14:32 @gwillen Oh, interesting! Where did you see that? Could be a good read 2021-06-27 18:14:08 @gwillen I do know lack of sun exposure is highly correlated with other cancers, but I don't know anything about causation. Vitamin D is a serious concern though and I don't know why folks downplay it so much 2021-06-27 18:11:55 But yeah citing a study of British adults like it's universal truth is just terrible science 2021-06-27 18:08:37 And it's technically true that a tan, like a burn, is your body's response to damage from UV radiation, but your body is always repairing and protecting itself in response to damage 2021-06-27 18:07:06 This only bugs me because of advice I've seen in the last few years to avoid basically all sun always, as if your life is about optimizing to prevent skin cancer, when sun exposure of an amount that varies by background and for each individual can have lots of positive effects 2021-06-27 18:04:48 It's probably a reasonable guideline if your skin type is similar to the kind commonly seen in that population, but for example a lot of pale middle eastern folks naturally tan so dramatically that I'd expect the SPF gain to be much higher 2021-06-27 18:02:41 The quote that a sun tan only provides an SPF protection of at most about 3 is, from what I recall, based on a single study of only British adults, FYI 2021-06-27 17:46:14 RT @marileezafari: All. Even the sweetest men who I would call genuine allies underestimate it. Not as bad, but their own biases and experi… 2021-06-27 16:59:24 RT @moskitos_bite: yes, powerful people, please do! 2021-06-27 16:48:57 RT @NuritPeleg: @TaliaRinger "things will be fine if you'll just believe in yourself" is an AWFUL promise. Because when things aren't fine,… 2021-06-27 16:46:56 It's a good time for systemic change, with so much changed forced upon us by the pandemic. I hope you'll all think about what power you have and how to use it to build a better community. Stay safe and healthy and cool in the meantime 2021-06-27 16:45:14 I'm using it right now 2021-06-27 16:44:24 Sometimes power isn't being in charge of decisions yourself. Sometimes it's having your opinions valued by people in charge of decisions. This is a form of power that I have, for example, and that I had even as a senior graduate student. It's a form I take very seriously 2021-06-27 16:41:37 But when you do that, make sure to talk to and listen to people who have been through these things, so that you can also make sure to build systems that keep us safe from retribution and other concerns you may not think of 2021-06-27 16:40:16 I want you to think about every incident you've read about or been party to, and how to build systems that in analogous situations discourage harm and---when harm does occur---encourage repair and reform 2021-06-27 16:38:41 A lot of people who read my Twitter are in positions of power and have influence over the design of those systems. I want you to think about this thread a lot 2021-06-27 16:36:02 This is hard because it's often systems enabling or encouraging harm, and systems dictating how if at all harm can be repaired. But it is people who make decisions about how those systems operate---though often not the people you'd expect 2021-06-27 16:33:02 To alleviate this effect, one must first repair the harm, then work to reform systems to prevent that harm from recurring, and finally only then expect that we will feel happy with and not angry with signals of caring about our participation 2021-06-27 16:31:12 Actions to the opposite effect are most hurtful when they come from the very institutions who recruited us, which is why signals of caring about our participation in computing---even when genuine and not performative---can feel like a punch in the gut, when coupled with harm 2021-06-27 16:27:49 We should obviously keep recruiting, but we should do so honestly, while working transparently to build safe environments for the people we recruit 2021-06-27 16:25:32 And not discussing how we can change the systems and norms that enable or even encourage that kind of behavior, and how to bring in powerful allies to help us 2021-06-27 16:24:04 It does feel a bit like a betrayal to be brought in actively by diversity programs promising me things will be fine if I just believe in myself, but not telling me I'm about to face huge systemic barriers, that people may treat me very cruelly in a professional context at times 2021-06-27 16:22:33 I am grateful to have the career I have, but I wish I could have it without the PTSD 2021-06-27 16:07:43 We are doing people a disservice if we recruit them into an unsafe environment without actively working to make the environment safer 2021-06-27 15:54:54 I think the difference for the rest of us is that it seems like basically everyone has a story, and that should be extremely alarming 2021-06-27 15:54:04 Of course sometimes straight white men are subject to it as well by way of other power dynamics like advisor/advisee relationships 2021-06-27 15:52:02 I think most straight white men in CS strongly underestimate the amount of harassment and abuse that occurs both in industry and in academia, disproportionately targeting the rest of us 2021-06-27 15:24:20 RT @noraj_williams: Fun fact: all genders have the capacity to learn to cook and clean distributed equally! We know this because when cooki… 2021-06-27 15:21:15 @TheCodePixi @_joaogui1 https://t.co/ycDE8wzCD3 2021-06-27 15:12:32 Too late though https://t.co/BPvZDFotf9 2021-06-27 15:10:25 RT @Reductress: How to Support Saving the Bees Even Though They Were Responsible for the Death of Your Childhood Best Friend: https://t.co/… 2021-06-27 15:03:13 @elazar_g Like after this event we'll probably see lots more forest fires, unbreathable air again, and so on 2021-06-27 15:02:00 @elazar_g But extreme weather events become more common much more quickly than that 2021-06-27 14:56:34 @TheCodePixi @_joaogui1 And they may have HONKED LIKE GEESE 2021-06-27 14:55:51 RT @TheCodePixi: I’ve been avoiding Twitter for a bit but have found a reason to show up because I must ask: What is your favorite dinosaur… 2021-06-27 14:55:39 RT @TheCodePixi: Also a big fan of dinos that we’ve learned had lots of feathers and looked like big goofy chickens. Look at this silly T-R… 2021-06-27 14:53:49 RT @LamontLilly: 6 years ago today. So much love and respect for this sister! Thank you, @BreeNewsome: "I do commemorate that moment, as I… 2021-06-27 14:37:30 @elazar_g How? 2021-06-27 14:36:22 RT @ProjectDrawdown: We have the solutions to stop #climate change. Dive into the need-to-know science with global leaders in Climate Solut… 2021-06-27 14:36:07 RT @ProjectDrawdown: "A lot of companies have made dubious climate commitments using accounting tricks — usually relying on problematic ‘ca… 2021-06-27 14:23:31 RT @littmath: spending billions of dollars building a supercollider to smash integers together at sub-light speeds and break them into thei… 2021-06-27 14:18:04 RT @agdakx: @TaliaRinger I just want to share the following articles to show that in fact, yes, there are solutions and it is not all hopel… 2021-06-27 06:24:53 @KimayaBedarkar Yeah, I often wonder what things we'd study if granted full control 2021-06-27 06:09:02 Brain: You are complicit in the bombing of children abroad Me: Can we please do this in the morning 2021-06-27 06:08:02 @Kaushik171 IIRC nope it's the DoD 2021-06-27 06:03:14 Brain: The DoD funds more cancer research than anyone else in the US Me: It's bed time though 2021-06-27 06:02:21 Brain: The internet was a DARPA project Me: OK I get it though 2021-06-27 06:01:29 Me: OK time to sleep Brain: Funding for basically all science in this country comes from the war machine Me: Please stop 2021-06-27 05:39:43 Based that at almost 31 I can tell my mom I'm hanging out with hackers and she says "good!" 2021-06-27 05:26:33 Conversations with my mom get more interesting as I get older: "I'm staying at a friend's house. He has AC." "Which friend?" "The hacker friend. The one that lent me the CPAP." "Oh good!" 2021-06-27 05:24:08 @ezyang Yeah I was depressed for a while and honestly it doesn't seem that bad personally, but do I want to bring kids into this world? I'd thought so, but now I worry 2021-06-27 05:07:21 @0xadr1an 8 or 9 miles total 2021-06-27 05:01:57 RT @EthicsInBricks: Not everything that counts can be counted, and not everything that can be counted counts. https://t.co/jPdy9YE9gW 2021-06-27 04:59:52 RT @EthicsInBricks: "All things are subject to interpretation. Whichever interpretation prevails at a given time is a function of power and… 2021-06-27 04:57:47 @__tpm__ I used to think that in an urgent emergency, humans would all come together to save each other. Then the pandemic happened, and now I think we'll just argue and sabotage each other until we all die 2021-06-27 04:54:39 Is there any hope for a future? 2021-06-27 04:51:16 @marki_a Only about a week was unliveable. It can get way worse 2021-06-27 04:48:13 Already last year I had to flee the state to Utah just to breathe because of fires. Is that just going to become normal even in the PNW? 2021-06-27 04:47:21 How many years will it take before summers are just totally unlivable in huge swaths of the world? 2021-06-27 04:46:17 RT @KSeattleWeather: Probably goes without saying, but tonight should set the record for *hottest night* in recorded Seattle history. Cur… 2021-06-27 04:40:43 @amyjko My house was 86/100 so I'm crashing at a friend's lol 2021-06-27 04:40:16 @amyjko So jealous 2021-06-27 03:54:16 This is before I go see my family in Israel, so I guess at least I won't burn there from the sudden transition 2021-06-27 03:53:15 Color from a run at 7AM today before it even got that hot lol https://t.co/yp8MVCOqdo 2021-06-27 03:50:54 @miniapeur It's 100 outside too 2021-06-27 02:42:38 @CheeToS_ Milk 2021-06-27 02:16:14 @LockerRoomSac Yummmmm 2021-06-27 00:56:22 @soldni I think milk, watermelon, and ice are the only three you need, but you can add any other fruit you want 2021-06-27 00:54:40 Housemate says this is a Korean hot weather dessert. It's soooo tasty https://t.co/Ug0lrmmaRG 2021-06-26 21:41:11 We're so screwed https://t.co/xGxOeYo5FJ 2021-06-26 21:40:03 RT @NWSBayArea: 2 pm temp update: 61 in downtown San Francisco and 101 at Pinnacles National Park. 40 degree spread from coast to about 2… 2021-06-26 21:29:54 I have another place to sleep through this all though 2021-06-26 21:29:38 I think my bedroom is already 100 degrees 2021-06-26 21:02:14 RT @siggraph: SIGGRAPH is honored to spotlight stories that illuminate the contributions of the LGBTQ+ community to our industry and world… 2021-06-26 20:37:05 @codydroux Cody 2021-06-26 19:37:32 I want to collaborate with every single one of you, you're so interesting. Why do I have such finite time 2021-06-26 19:26:58 @elazar_g That is an encoding that is actually fairly syntactically useful 2021-06-26 19:21:04 @lindsey But if it is a Dagstuhl, I can say "logic and ML," but then select only the classical FOL people who actually talk to type theorists, and similarly in the opposite direction haha. Just don't want it to be UNSAT 2021-06-26 19:18:48 @lindsey Hmm I'll have to think. Nothing wrong with classical logic, I think it's just a matter of not having to bring together 3 communities who don't like each other. 2 is easier 2021-06-26 19:16:50 @elazar_g 2 also sucks 2021-06-26 19:16:20 @lindsey But there are lots of other cool logics out there, and if it is a Dagstuhl I could be careful to represent interesting logics 2021-06-26 19:15:27 @lindsey I just don't want to get too many classical FOL folks, they don't usually like type theorists, and it's too many communities to bridge 2021-06-26 19:14:19 @elazar_g The syntax erases the semantics, which we are left to infer externally 2021-06-26 19:13:31 @TechnoEmpress I wish 2021-06-26 19:12:05 @lindsey I'm glad :) the type theorists are going to be more stubborn 2021-06-26 19:08:49 RT @TaliaRinger: Which of these best describes you? 2021-06-26 19:08:13 Hi I'm still bi. Happy pride you wonderful wonderful people https://t.co/qUAR3bSVl2 2021-06-26 18:56:47 Syntax Is All You Need, But Not That Syntax Because It Sucks 2021-06-26 18:46:50 @YieldsF Is there any work in making NL semantics explicit in extended syntax? 2021-06-26 18:45:40 @YieldsF Not as much, but interested. Kind of skeptical that the logical systems that we knew about before, say, the late 1970s could be useful for expressing substantial parts of natural language at all, but wouldn't be surprised if concepts were developed multiple times in parallel 2021-06-26 18:34:31 @roboguy20 Essentially, or really thinking of it not as learning over natural language, but learning over an extension of natural language 2021-06-26 18:27:46 And of course you can erase those in the end because they're redundant, otherwise unencoded information 2021-06-26 18:26:03 I think it gets extra fun too because while you could start with heuristic scripts, like reifying arithmetic, you could later maybe learn theories themselves, then reify those theories in the syntax 2021-06-26 18:22:35 (Type theory can of course get way, way fancier than reifying just arithmetic, just that's a simple example) 2021-06-26 18:03:36 Just a fun curiosity, I think, but I want to know what happens 2021-06-26 18:03:14 So it would be interesting to me to preprocess text to reify external theories humans have in the syntax, then just train over that instead. I just want to know what happens so badly (but lack the infrastructure to try this, though I could write some preprocessing code) 2021-06-26 18:01:52 What I like about type theory is that you can take external things and make them internal. In fact if you're clever you can represent them explicitly in your syntax 2021-06-26 18:00:43 A really simple example of this is just reading sentences like "John has two apples, and Mary gives him three more 2021-06-26 17:58:47 While natural language is very different in that many things lack structure, or we have no idea how to ever explain them, I think there's a natural analogue when there is structure. 2021-06-26 17:57:41 In type theory there's this distinction between things that are "internal" (in the type theory or logical system itself) and "external" (structure that exists in some logical system you can reason about in an external tool, but not explicitly in the system you are in) 2021-06-26 17:38:53 @mraginsky @EliSennesh @jaycech3n @cHHillee @PTOOP @pigworker @osazuwa Maybe we can find a student in the next recruiting cycle or something, either way we should talk when I get back in town 2021-06-26 17:37:33 RT @mraginsky: @EliSennesh @TaliaRinger @jaycech3n @cHHillee @PTOOP @pigworker @osazuwa This is where we synthesize ML, control, and type t… 2021-06-26 17:36:33 @mraginsky @EliSennesh @jaycech3n @cHHillee @PTOOP @pigworker @osazuwa Do you have students interested in this kind of thing? 2021-06-26 17:33:56 @XandaSchofield @krismicinski @cbarrett I was intrigued when my friend told me about "golden tickets" 2021-06-26 17:20:06 @EliSennesh @jaycech3n @cHHillee @PTOOP @pigworker Yeah @PTOOP is his other account 2021-06-26 17:19:18 @EliSennesh @jaycech3n @cHHillee Inductive types in particular 2021-06-26 17:19:08 @EliSennesh @jaycech3n @cHHillee Oh, I think "description" is a technical term 2021-06-26 17:18:18 @cbarrett Oh man @geomblog was it you who first invoked the principle that all discussions of ethical AI lead to dismantling capitalism because 2021-06-26 17:17:18 @EliSennesh @jaycech3n @cHHillee Yeah I'd talk to @PTOOP / @pigworker about separating structure from the rest he is quite good at that 2021-06-26 17:15:25 @EliSennesh @jaycech3n @cHHillee Conor McBride's descriptions? That was what Jasper Hugunin suggested to me yesterday but IDK 2021-06-26 17:14:16 @EliSennesh @jaycech3n @cHHillee I wanna collaborate with you one day 2021-06-26 17:13:13 @cbarrett @krismicinski Regulate tech companies 2021-06-26 17:12:35 @cbarrett To me your points don't say "don't use ML," they say "ethical AI matters even for type theorists" 2021-06-26 17:12:07 @cbarrett I mean that everyone doing CS should need an ethics education and should need to apply it all the time as we work, it's not a comment to you but to the theoretical "you" applying ML to answer fuzzy PLT questions 2021-06-26 17:09:53 @cbarrett @krismicinski Learn ethics yourself too 2021-06-26 17:09:38 @cbarrett @krismicinski Hire ethicists 2021-06-26 17:08:44 @EliSennesh @jaycech3n @cHHillee Approximate type inference hehe 2021-06-26 17:07:27 @cbarrett @krismicinski Only if you abandon guarantees about the things that need guarantees, which you don't need to do when you have a rich type system 2021-06-26 17:02:55 @EliSennesh @jaycech3n @cHHillee Explaiiiin I want to know more :) 2021-06-26 17:00:15 @o1lo01ol1o That would be fun! 2021-06-26 16:56:12 @cbarrett @krismicinski "I want my code to have good style" "I want my code to be easy to understand" "I want my code to do this thing, but I'm not quite sure how to tell you what this thing is formally" "I want my proofs to be elegant" "I want my updated code to look like my old code" 2021-06-26 16:54:08 @jaycech3n @HarshaSomisetty @PTOOP The CoqHammer guy, nice 2021-06-26 16:52:52 @cbarrett @krismicinski Because humans use these things, and humans have weird soft constraints about what they want from their code, but that they can't often easily express 2021-06-26 16:51:08 @XandaSchofield @krismicinski @cbarrett Do you have papers I can look at from the pure ML perspective? 2021-06-26 16:49:24 @jaycech3n @HarshaSomisetty Other low hanging fruit (I probably will try these): - NNs for proof synthesis will probably get better if you first transform fixpoints to eliminators - maybe you could even encode inductive types as descriptions a la @PTOOP (Jasper Hugunin's idea) 2021-06-26 16:46:47 @jaycech3n @HarshaSomisetty Sure yeah, it's for symbolic automation, but I'd look at the proof term transformation in Chapter 4. The categorical explanation of it in my defense (on YouTube) is also useful 2021-06-26 16:43:51 @jaycech3n @HarshaSomisetty I'm weird though 2021-06-26 16:43:38 @jaycech3n @HarshaSomisetty I actually think we want to use it earlier, to pass over data with implicit, external structure and make it explicit and internal 2021-06-26 16:42:41 @m_misamore @jaycech3n @HarshaSomisetty I would say yes, and that is immediately tractable. Every program has external structure, you can reify that explicitly, I want to do this ASAP 2021-06-26 16:41:29 @cbarrett @krismicinski ML can make PLT immediately useful, so that it reaches real people instead of being viewed as a weird niche thing nobody will ever use, which is how people view cubical right now even though it's one of the most practical things ever IMO with good enough automation and UX 2021-06-26 16:39:05 @cbarrett @krismicinski Path finding is everywhere in ML, and also everywhere in HoTT & 2021-06-26 16:38:17 @cbarrett @krismicinski Like inverse problems in ML are approximate decompilation. This is why we decided to build a language model into our proof term to proof script decompiler 2021-06-26 16:37:32 @cbarrett @krismicinski Faculty interviews got me reading like 50 of the most interesting sounding NeurIPS and ICML papers from the last few years, and honestly it was a really useful exercise, because the concepts overlap quite nicely with PL if you let yourself be imaginative 2021-06-26 16:32:55 @cbarrett @krismicinski What if you have a bunch of changes you've made over time, and your IDE is tracking them, and you want your IDE to suggest some useful abstractions that would have usefully captured previous iterations without breaking? I'd probably go with something neurosymbolic for that 2021-06-26 16:31:14 @cbarrett @krismicinski And once you start reifying semantics into syntax, I think, then pretty much all of the ML techniques that are good at syntax will carry over fairly nicely to PL, but in a way that actually makes sense and isn't terrible 2021-06-26 16:30:10 @cbarrett @krismicinski What if you want to align concepts across different proof assistants and libraries? IIRC the work that exists on this is all learned 2021-06-26 16:28:32 @cbarrett @krismicinski Ah a skeptic in that direction. Interesting! So the Andrew Ng class is simple and just an intro. But ML is going to be interesting whenever we have soft constraints---like what if I want to not just synthesize proofs, but also match a human-friendly style of tactics? 2021-06-26 16:26:12 @krismicinski @cbarrett Easiest place to start with this would be for program synthesis, but I'd bet it carries over to any domain in which there is some implicit structure that the model just can't use well even though it's obvious to humans---in type theory terms, it's external 2021-06-26 16:25:17 @krismicinski @cbarrett Before even doing something that fancy, you could just run a few heuristics to reify well-known concepts syntactically, then run the same workflow 2021-06-26 16:24:29 @krismicinski @cbarrett What if you first align concepts across a large corpus of data with type theory knowledge, then reify those concepts syntactically, then train over extended syntax, then erase the reified syntax before inference? 2021-06-26 16:22:43 @krismicinski @cbarrett Type theory is all about capturing structure and making it explicit, general, and understandable. Even for really, really complicated structure not captured by old-school logics. At a point I'm TT there's no distinction between syntax & 2021-06-26 16:20:52 @krismicinski @cbarrett And robust. I think a lot about LLMs and how they still manage to not understand basic things like addition after parsing the entire internet. That makes sense---we understand math metatheoretically, but don't represent it syntactically. But we could. That's a simple example 2021-06-26 16:19:51 @krismicinski @cbarrett I actually think it will help with few example learning, by way of reifying implicit semantics hidden by syntax as syntax, making it possible to leverage existing NN techniques but actually capture semantic data they miss. And this will also make models more accurate 2021-06-26 16:17:38 @scheidegger Ah yes, I think I've had to meet those constraints to organize other things like PLMW, SIGPLAN-M, and the POPLmark Retrospective Panel 2021-06-26 16:15:58 @geomblog I think it'd be hard to get most ML people to come to a PL conference, so maybe holding it at one of the ML conferences is the way to go 2021-06-26 16:15:17 @EliSennesh @cHHillee My thesis haha 2021-06-26 16:14:50 RT @EliSennesh: @TaliaRinger Talia, you're my friend, but please stop telling everyone the secret to my career. https://t.co/hEAZeMWEnD 2021-06-26 13:40:42 @five9a2 I think that is just one application of TT for ML though. It should also be amazing for neurosymbolic learning 2021-06-26 13:28:20 @deckeresq @tiarkrompf Nice 2021-06-26 13:25:19 @m_misamore In a language without univalence, I already use it as the metatheory to describe and understand proof repair 2021-06-26 13:24:17 @m_misamore IMO, in its implementation (cubical type theory), the most natural generalization of congruence to dependent types. This allows for beautiful and elegant automation of transport across equivalences, allowing for amazing proof reuse tools. Just need a student to help implement 2021-06-26 13:15:16 @jaycech3n @HarshaSomisetty Do you think this question is tractable? 2021-06-26 06:57:11 I've heard the opinion that type theory probably can't offer much to ML twice now and it just means you're not thinking aspirationally enough, and you're accepting limitations as inevitable and reasonable when they are not, IMO https://t.co/noiwKFcset 2021-06-26 06:54:22 @umuti5ik Sigh https://t.co/noiwKFcset 2021-06-26 06:50:47 Small sample but interesting https://t.co/neAXncbbuc 2021-06-26 06:48:44 Which of these best describes you? 2021-06-26 06:42:34 @scheidegger Interesting, like what? 2021-06-26 06:28:32 RT @StrahdVonZ: Please, before you joke about the heat wave in the Pacific Northwest, remember that almost all of us are without central AC… 2021-06-26 06:20:50 @kylewadegrove I want it to spin off into like 4 or 5 collaborations between people who otherwise would not have ever spoken to each other about research. And track those collaborations and retrospect on the results in a few years 2021-06-26 06:18:11 This is how I approach most of type theory https://t.co/4UGXVRqTzf 2021-06-26 06:16:00 Heck yeah https://t.co/kG5uYkYiws 2021-06-26 06:06:29 If this sounds fun to you and you're a machine learning expert who thinks the intersection of machine learning and type theory is promising, and you're interested in proposing a Dagstuhl, super please let me know 2021-06-26 06:05:36 Dagstuhl seminars literally involve hanging out in a castle in a forest in Germany, drinking beer, bringing together junior researchers with famous people, and fostering super cool collaborations between different research areas 2021-06-26 05:59:12 @dlowd Ah I just assumed helping because I assumed most people are good and want to help each other and are willing to do the work to try, just our social norms incentivize selfishness, which is contrary to human nature 2021-06-26 05:57:29 @dlowd Certainly, though, power dynamics should be considered when intervening in other people's problems 2021-06-26 05:56:58 @dlowd Plus it means accepting that they have as much to offer to you as you to them, a frame that colonization rejects 2021-06-26 05:56:19 @dlowd IMO post-hoc, but the root cause is just people thinking they're the best. Intervening and helping people means actually listening to them, not just imposing your norms and values on them 2021-06-26 05:54:21 I'm good on type theory and on seeing connections between the areas, but a good ML expert would help a lot 2021-06-26 05:53:17 If I organize a Dagstuhl for machine learning & 2021-06-26 05:52:10 @rakamaric @dagstuhl True!! I think my worry is that there's such a backlog of seminars that were delayed during COVID. It would take a long time to be able to do it, probably. 2021-06-26 05:48:48 RT @fund_my_startup: please fund my startup that will reinvent cats by integrating space travel 2021-06-26 05:47:54 RT @fund_my_startup: looking for funds to revolutionise Pokémon by integrating OCaml and Electron 2021-06-26 05:47:11 RT @fund_my_startup: please fund my startup that will disrupt pop music with artificial intelligence and an API 2021-06-26 05:45:35 RT @fund_my_startup: pivoting to reinvent goats using facial recognition, native apps and neural networks 2021-06-26 05:45:28 RT @fund_my_startup: pivoting to disrupt tomatoes using Lisp 2021-06-26 05:45:12 RT @fund_my_startup: seeking money to reinvent crisps using Haskell and A/B testing 2021-06-26 05:44:23 RT @fund_my_startup: planning to revolutionise lip balm with CoffeeScript, the GPL and OCaml 2021-06-26 05:40:19 @techpractical I find it usually improves my internal state to focus on the system honestly 2021-06-26 05:37:59 @techpractical IME, the world is so messed up that the external situation is almost always totally fucked, and there is almost always low-hanging fruit to improve the situation, if not for yourself then for everyone after you 2021-06-26 05:35:48 Because acting with transparency is one of the most important things you can do IMO 2021-06-26 05:35:32 I cross the boundary unless the person has told me to never contact them again explicitly 2021-06-26 05:35:04 More! Because we are again best equipped to help one another. So by focusing too much on our own problems and not on other people's problems, we solve way fewer problems, and there is more pain and suffering in the world 2021-06-26 05:34:12 External situation. Most often my feelings are reasonable reflections of the world around me, so improving the world is a good use of my energy, and helps not just me but everyone after me. If this is impossible or the feelings persist after, then my internal state 2021-06-26 05:33:15 Definitely not. Feelings should reflect reality. My bad feelings motivate me to improve the world, so I don't want to lose them! 2021-06-26 05:32:47 Stoicism is bad for the same reason unwarranted enthusiastic positivity is bad. It's important to show feelings that reflect the external situation, so that people don't feel terrible for feeling bad inside 2021-06-26 05:31:57 More for others because I believe people are generally better at helping one another than at helping ourselves, so I want to build a world in which all of us are always helping each other, functioning like one big team 2021-06-26 05:31:02 IMO harmful specifically because it misleads people around me to think that positivity is warranted, which can make them feel much, much worse for feeling bad internally. 2021-06-26 05:29:56 Finally, even though there are few of you, I'm quite fascinated by my friends with lawful alignment, who say unjust laws ought to be followed, authority ought to be respected, and so on. I'd love to hear more about that worldview https://t.co/LejK2yu46w 2021-06-26 05:28:32 Transparency versus boundaries. How do you navigate this tradeoff in practice? Like say acting transparently means crossing a boundary, talking to someone who prefers not to talk to you. What do you do? Why? https://t.co/Gv62jQLQOY 2021-06-26 05:26:41 Should people intervene more or less in other people's problems? Why? https://t.co/63CfCwIPJT 2021-06-26 05:26:02 I AGREE with most of you that giving an ultimatum is more respectful than withholding information. But I'm surprised this is a popular opinion. Can you explain? 2021-06-26 05:25:26 This is not evenly split, but it still surprised me, so I want to hear about. In work contexts, I often hear ultimata framed as inherently disrespectful, in many cases a fixable offense. I know people who end romantic relationships rather than issue them https://t.co/M5UlSpivDf 2021-06-26 05:24:05 Better to start with the external situation or your internal state? Why? https://t.co/lsPXBVVIav 2021-06-26 05:23:34 Would you or would you not choose to be happy all the time if you could, in spite of the world around you? Why? https://t.co/IXR6EBpBiY 2021-06-26 05:22:48 This one is EXACTLY 50/50. Wow! Being stoic. Do you think this is generally a good thing or a bad thing? https://t.co/Fw7Md0hDju 2021-06-26 05:22:09 Do you optimize more for yourself or more for others? Why? https://t.co/kDAaJm7Mf4 2021-06-26 05:21:29 OK, the interesting ones to discuss are the ones that had almost 50/50 answers. So let's do that. Do you view enthusiastic positivity as more helpful or harmful in a bad situation? Why? https://t.co/5NOrYdtNh5 2021-06-26 04:58:50 RT @DinnerMayBeLate: This is, I think, the central thing. 2021-06-26 04:58:15 @sharad24sc Yes :) 2021-06-26 04:57:20 @jwthickstun Cool!! I appreciate it, it should help with drawing the right bridges 2021-06-26 04:42:13 @notkavi With enough type theory you could suffer through this as long as you accept you will not understand everything. Embrace confusion, aim for intuition, trust your feelings, don't worry too much about details IMO https://t.co/grtd5bw8ww 2021-06-26 04:38:37 @cHHillee Ah OK, well, HoTT going to be useful for any situation that involves reasoning about points, spaces, and paths in arbitrarily many dimensions 2021-06-26 04:36:11 I would put a lot of money on homotopy type theory being useful for machine learning if you get the right people in a room together https://t.co/zPvPA9UCKJ 2021-06-26 04:32:23 @cHHillee In as many dimensions as you want 2021-06-26 04:31:54 @cHHillee https://t.co/FMzsvLK7J5 2021-06-26 04:30:59 @cHHillee https://t.co/Ao7VA9jbQY 2021-06-26 04:18:16 @cHHillee You want few-example learning? Robustness? Guarantees? Truly explainable AI? Talk to type theorists, they'll be able to help as long as you first get on the same page and establish shared vernacular 2021-06-26 04:16:27 @cHHillee Of course, not everything can be explained this way. But that's the interesting part---how can they meet somewhere, how can you get the best of both worlds? 2021-06-26 04:15:48 @cHHillee And for the parts that can be explained this way, type theory is amazing at generalizing in one or two examples to something fully general and fully robust, possibly even with a proof nothing will ever go wrong 2021-06-26 04:14:12 @cHHillee Type theory is good for understanding semantics erased in syntactic data, and reifying it explicitly, perhaps even directly in the syntax itself 2021-06-26 04:12:54 @cHHillee Type theory is broader than that. Symbolic search problems in rich type theories are often simple to do with between zero and two examples, and extremely strong guarantees 2021-06-26 04:08:54 @miniapeur Yeah, this is part of why I want a workshop so badly 2021-06-26 04:07:54 @brendanzab Both!! 2021-06-26 03:43:29 @miniapeur I find the best way to learn these things for me is often to have conversations with experts 2021-06-26 03:42:40 @miniapeur Undergrad in math & 2021-06-26 03:34:23 @miniapeur It's an incredibly hard read. I struggled with it a lot the first 2 times I read it. I've read it 3 times now. I think it gets into a lot of details you don't need if your goal is just to get a sense of what's happening to build automation 2021-06-26 03:29:37 @miniapeur A quick note on state-of-the-art type theory: I don't recommend ever going into one of these books or papers expecting to understanding everything. It's really important to embrace the feeling of confusion, let it stew, and come back to it later on 2021-06-26 03:27:34 @miniapeur If you are extremely familiar with topology, I would recommend reading this: https://t.co/jh5qNmSAYn. I then recommend washing it down with this: https://t.co/BIiInVp5Tq. If not, though, I don't recommend that at all 2021-06-26 03:21:47 @miniapeur To be honest, I think this is the problem right now. The newest type theory doesn't have a lot of good educational resources that are reasonable to understand. Do you know topology very well? If so, I do have recommendations, but if not, not as much 2021-06-26 03:20:19 Thank you based stalkers and harassers I appreciate you for helping me become the researcher I am today lmao 2021-06-26 03:19:50 It's a lot like how my first research experience was made possible by a stalker who loudly yelled "hi Talia!!" every time I came late to class, so that my professor started to notice me 2021-06-26 03:17:52 I'm so glad I got harassed by that dude. It opened up my Twitter network to include tons of machine learning people, which then made me obsess over new problems that I think can be much more impactful than anything I'd thought about before 2021-06-26 03:04:24 @miniapeur Automatic transport in type theory is typically finding a path, where paths encode equalities or equivalences (which are bijective). So if you start adding soft cost constraints, the problems might coincide interestingly 2021-06-26 03:01:00 @miniapeur I've been curious for a while whether there's a relationship between transport in the type theory sense and in the machine learning sense hahaha. Like if you add soft constraints to automatic transport problems in type theory, can they be viewed as optimal transport problems 2021-06-26 02:59:37 @miniapeur Yeah fair 2021-06-26 02:44:53 How do people start workshops, find sponsors, convince the right people to come, and so on? I know the structure I'd want. Not talks, but working groups on specific focus areas, with the goal of spinning off new collaborations by the end of the workshop 2021-06-26 02:43:01 @cbarrett Both fields would advance significantly if the two communities spoke to one another 2021-06-26 02:42:47 @cbarrett The things machine learning researchers find hard are things that type theorists could help with pretty much immediately. The things that type theorists find hard are things that machine learning people could help with pretty much immediately 2021-06-26 02:25:44 @mvanier42 I honestly think more resistance will come from the type theorists 2021-06-26 02:24:24 @cbarrett That's exactly why. The things that each community is bad at are exactly the things that the other community is good at. Perfect team 2021-06-26 02:23:03 "But we don't need that structure" bro you literally have to train on the whole internet to get results that look syntactically reasonable but still even after processing the entire internet struggle to do basic math 2021-06-26 02:22:20 "But continuous math is scary" bro you all invented cubical type theory which literally axiomatizes a real number interval in order to describe equalities as paths between spaces SMH 2021-06-26 02:19:22 It's a match made in heaven. Talk to each other please 2021-06-26 02:19:03 Machine learning people are amazing at efficiently encoding and processing large amounts of data, and at capturing the soft constraints that can't easily be explained by the kind of structure type theorists love. 2021-06-26 02:16:53 Type theorists are amazing at discovering and describing beautiful, elegant, general, rich, and informative structure that provides strong guarantees and is easy to explain. 2021-06-26 02:11:04 @rbhar90 Yeah, I think it could be great, and can think of a few focused work groups or tracks that could go extraordinarily well (type theory for machine learning frameworks, neurosymbolic learning) but I'm not sure how to navigate the social challenge of getting the right people to come 2021-06-26 02:04:56 I really, really, really want to create a machine learning & 2021-06-26 02:03:00 @mdekstrand It's so annoying that in my quest to make machine learning need fewer examples, I need the infrastructure to learn from ridiculous numbers of examples hahahaha 2021-06-26 02:01:12 @znmeb @arclight Actually an extremely interesting use case, because one cool application of this technology is porting legacy code 2021-06-26 02:00:38 @mdekstrand Ah, got it, I'll talk to the machine learning folks at UIUC, probably 2021-06-26 01:59:49 RT @nikitab: @TaliaRinger @certifiablyrand Nice! We converge at Poisson, what are the odds 2021-06-26 01:51:41 @mdekstrand I have enough startup to buy extra storage when I start. How much does 10TB cost? 2021-06-26 01:50:54 @mdekstrand Haha only like 1GB I have so little space. I can't imagine having 10TB, that's local? Man everyone else's research sounds computationally expensive AF 2021-06-26 01:45:55 But I'm a doofus and have no clue where I'm supposed to store all of these things so I can browse them lmao 2021-06-26 01:45:03 So if I can observe behavior in simple programming languages that can be described externally by the same beautiful structure, it should be possible to reify that structure syntactically, and that would be a very beautiful merger of type theory and neurosymbolic synthesis 2021-06-26 01:43:55 But as I do this for proof search, I want to see what I can move back out to help my program synthesis friends, and the cool thing about simpler programming languages is that a lot of their behavior can be described externally in terms of more expressive type systems 2021-06-26 01:40:41 Constructive dependently typed languages are so expressive that I've never needed more than 1-2 examples until now and honestly I think eventually when it comes to constraints that aren't soft, it will eventually be easy to learn from very few examples for most of programming 2021-06-26 01:38:00 My friend sent me a few (noting limitations of all existing datasets right now) but like, some of them are really big. I just don't have space on my computer for that kind of thing, to even look through them just to brainstorm. Do you all store data externally? 2021-06-26 01:36:54 So as I brainstorm some ideas for neurosymbolic automation, one thing I want to do is manually look at existing code datasets to see what semantic structure is not yet encoded syntactically (probably most TBH). Naive question: 2021-06-26 01:24:13 RT @djg98115: Your therapist really should understand that your behavior depends on your environment https://t.co/WyOtwIyHbF 2021-06-26 01:24:08 RT @EmilyKager: i was complaining to my therapist about needing an answer for functions that have access to local scope variables and she t… 2021-06-26 01:17:10 @certifiablyrand Wow he was coadvised by Reynolds and Bob Harper, wild 2021-06-26 00:56:43 @nicoleradziwill @johnregehr @certifiablyrand @GMorrisett It also just includes names I've been a huge fan of since my math undergrad. Like Euler oh my gosh. And then in modern times, Church & 2021-06-26 00:47:54 @johnregehr @samth @certifiablyrand We need math genealogy project with existential variables. clearly 2021-06-26 00:45:01 @certifiablyrand @djg98115 I just submitted myself 2021-06-26 00:44:34 @samth @certifiablyrand I really like it, it speaks a lot to the history of the field 2021-06-26 00:44:01 RT @samth: @TaliaRinger @certifiablyrand It's a weird feature of CS that some people's academic genealogy looks like that, back to 1700s, a… 2021-06-26 00:32:46 @certifiablyrand This is Dan's academic line. So good https://t.co/05Wr4GwoDX 2021-06-26 00:31:27 @certifiablyrand Dan's line is amazing, happy to inherit it 2021-06-26 00:31:08 @certifiablyrand ? Do I have to put myself in there manually? 2021-06-26 00:08:44 If you really love looking at programs and proofs and finding structure within them---then using that structure to build cool new automation 2021-06-26 00:07:32 If you're generally not afraid of ideas that tread totally new ground, and if you think most things that are hard are only hard because we haven't worked on them enough yet, not because they ought to be hard 2021-06-26 00:05:36 Or if you think that, in an ideal world, formal proof about program correctness really shouldn't be much harder than writing unit tests 2021-06-26 00:04:42 You'll like working with me if you think types are beautiful, but you want them to be both beautiful and useful, and you think the way to do that is to build good automation and user experience, meeting software engineers where they are 2021-06-26 00:03:27 @_joaogui1 Right now I'm looking at machine learning frameworks, and helping people easily and with strong guarantees rule out tensor shape errors at compile time. I don't know if that will flesh out, it really depends on student interests, but there are industrial partners 2021-06-26 00:01:38 Also have room for lots of work continuing along the proof repair line, and want to do some more mixed-methods verification work using that. Will plan that in July. Have capacity for 1-2 Ph.D. students who care about proof automation. Plus 1-2 with any overlapping interests 2021-06-25 23:59:28 NS = neurosymbolic DTT = dependent type theory ML = machine learning (in this particular context) cubical = cubical type theory 2021-06-25 23:58:32 Summer research thoughts & - NS proof term decompiler (ongoing) - DTT for ML (planning, need interested students) - e-graphs for automatic transport in cubical (planning, need interested students) - reifying semantics as syntax for NS proof search (thinking) 2021-06-25 23:11:24 @tianyin_xu @taoxiease Like @JeffDean, the famous object-oriented compiler optimization guy 2021-06-25 23:10:19 @polikarn Then they go on complaining about how it's actually recursion, but there's a well defined transformation between induction and a limited form of recursion :) 2021-06-25 23:09:56 @polikarn Oh haha, yeah, quite the nit. Especially given the eventual development of inductive types 2021-06-25 23:08:50 @tianyin_xu @taoxiease It's an influential list of people for sure 2021-06-25 23:04:04 @tianyin_xu ominous, which ones 2021-06-25 22:58:23 Moved myself into the alumni list on the PLSE website. heck yes https://t.co/Qvi4HKpcR7 https://t.co/PhrA9dNJ3s 2021-06-25 22:44:45 Love you @somelaurachick 2021-06-25 22:41:42 My friend's robot is hanging out in the hall here, but it's triggering the part of my brain that wants to pet cute doggies, so I keep wanting to run over and pet the robot which is weird 2021-06-25 22:37:02 So then we laughed and got pizza 2021-06-25 22:36:53 Remembering fondly the time I was supposed to hang out with one of my friends, but forgot because of ADHD, so I messaged her frantically apologizing, at which point she admitted she'd also forgotten, also because of ADHD 2021-06-25 22:33:55 @andrewarana Hybrid! 2021-06-25 22:32:32 @jameskjx Abstinence-only media ed 2021-06-25 21:30:38 @mydamnhandle1 Honestly my best practice is finding a friend with AC and crashing there 2021-06-25 21:22:03 https://t.co/lWKAqRVja4 2021-06-25 21:20:39 https://t.co/5p9eSKGDI3 2021-06-25 21:18:05 Recall that AC is a rarity here 2021-06-25 21:17:27 @KSeattleWeather well that's bad 2021-06-25 21:16:44 110° holy fuck 2021-06-25 21:15:56 I'm sorry what https://t.co/QnkogsESgp 2021-06-25 20:57:48 RT @schemeprincess: I'd love for this to eventually develop into Rising Stars but instead of discussing job hunting, we tackle systematic i… 2021-06-25 20:53:33 @schemeprincess I do fit the criteria, though, if you end up wanting my help. 2021-06-25 20:50:56 @schemeprincess Eventually, that is, I'm sure you have your goals for how you plan to start 2021-06-25 20:49:48 @schemeprincess I'm in 2021-06-25 20:49:15 RT @schemeprincess: Thanks to a suggestion from my sister @_wangsharon, who is starting her PhD in clinical psychology on sexual violence a… 2021-06-25 20:40:07 @polikarn Turing really did this way too early, the world wasn't ready 2021-06-25 20:39:48 @polikarn Those comments are from way later, right? Because this was 1948 IIRC and they mention Floyd which was late 1960s 2021-06-25 20:36:55 RT @polikarn: @TaliaRinger I really like the "conference discussion" about this paper (see https://t.co/595whdiWom). Basically, everyone wa… 2021-06-25 20:35:17 Very well deserved!!!!!! https://t.co/RXUGAhFgoy 2021-06-25 19:51:12 @soldni I don't really know what it is but if I get ephemeral documents I'll be happy 2021-06-25 19:50:28 @soldni I monitor news about Google lol. At some point saw news about this https://t.co/NXMWeXDyna 2021-06-25 19:44:50 @soldni Isn't everyone moving to Workspace? Maybe it will be available then? 2021-06-25 19:32:02 @amyjko Yeah I need something like that 2021-06-25 19:26:30 Do professors get media training? 2021-06-25 18:48:48 @polikarn Turing the OG verification guy 2021-06-19 07:30:41 Actu4l1y, 4h75 6o54d w3rk 3e99 2021-06-19 07:13:17 Or you can teach them to write de Bruijn indexed notes, that could be fun 2021-06-19 07:11:21 Teachers if you catch a kid doing this, don't scold the kid, use it to teach the kid really cool math 2021-06-19 07:09:57 I just wish I'd learned cooler cryptosystems like RSA back then, that would have been fun 2021-06-19 07:09:35 In middle school I thought of a cute primitive cipher to pass notes. Basically the letter that occurred in position n became the number n in all subsequent occurrences. This got me through middle school without teachers reading my notes, surprising 2021-06-19 07:05:07 RT @va2lam: @TaliaRinger @nikitab Sorry! 2021-06-19 07:01:01 Canada needs to start exporting apologies. I'll pay https://t.co/1eO711s7uL 2021-06-19 07:00:32 @va2lam @nikitab It's cool Canada has apologies which the US largely lacks 2021-06-19 06:50:49 @nbhansen Yeahhhh let's get tacos and make faces 2021-06-19 06:50:39 RT @nbhansen: @TaliaRinger one of the most normal symptoms for ADHD is actually tacos and silly faces :) 2021-06-19 06:48:50 Tacos and silly faces all day 2021-06-19 06:48:36 ADHD squad 2021-06-19 06:46:48 RT @mariana057: Wait…. What??? https://t.co/5UqB5frwug 2021-06-19 06:43:24 "I'm sorry, I feel so bad, I haven't seen you in so long with this whole pandemic" "It's cool, it happens, like once every hundred years or so" 2021-06-19 06:17:41 @nikitab Does Urbana-Champaign have good tacos anywhere? 2021-06-19 05:57:22 @dimvar @SchuCodes Thanks so much! 2021-06-19 05:55:52 I'll miss these friends https://t.co/Hm1Kwg4wYk 2021-06-19 05:55:14 I have friends https://t.co/aeLugIMbSp 2021-06-19 03:26:33 Seeing friends I made playing dancing games long before grad school kinda puts things into perspective when they're like "hahaha holy shit you're a professor, let's get tacos." Same 2021-06-19 02:44:41 @dimvar @SchuCodes If you sign up, you'll stay in the system for future matches :). ML/PL intersection is basically always in demand 2021-06-19 02:04:19 - help me plan my training for the Boston Marathon 2021-06-19 02:02:19 - help me book a COVID test before I fly to Israel - help me go through old clothes and donate some - help me figure out how to find new doctors where I move - help me figure out when I need to book a mover - help me parse the overwhelming list of contacts at UIUC 2021-06-19 01:58:47 @wilton_quinn That would be so amazing I benefit so much from talking to someone as I do that 2021-06-19 01:57:52 - go through my mail with me - help me clean my room - do some chores for me - cook for me and make enough for leftovers!!! - help me figure out how to get my automatic payment on my car lease to start up again 2021-06-19 01:56:44 Alternatively: - a load of groceries, with lots of small prepared food items - a meal when I'm tired - choose where we will hang out and tell me when to be there - help me unpack from my trip several weeks ago - do laundry with me 2021-06-19 01:53:41 Please everyone who wants to do something for me for graduation, sit down with me for 30 minutes to an hour and help me talk through my tasks and emails and calendar events and future faculty life 2021-06-19 01:53:10 My mom asked what I want from graduation and I was like "help organizing tasks and planning" but I think her friends are going to send me gifts and then I'll have to convince myself to open them and throw out the boxes and ahhh 2021-06-19 01:50:17 I desperately need help organizing things, doing general executive functioning stuff, planning, and so on this summer. I have extra admin help once I start (asked because of ADHD) but in the meantime my executive functioning skills are so dead I can't plan dinner 2021-06-19 01:28:04 @samth @TensorFlow @PyTorch So let's do a cool thing, let's learn from the good things about mathlib's community building without also excluding people, erasing history, and making blatantly incorrect and hyperbolic claims about people's entire life's work 2021-06-19 01:26:42 @samth @TensorFlow @PyTorch Instead the community seems quite open to developing new languages and frameworks, and readily concedes to all of the limitations, and welcomes people who want to come and help and develop new things. This is an example of a strong and tightly knit community that isn't harmful 2021-06-19 01:25:47 @samth @TensorFlow @PyTorch But if the TensorFlow community gave me some nonsense about how all languages outside of Python are useless artifacts of history and yelled at me whenever I asked them questions about what future tools could look like, I'd probably be annoyed 2021-06-19 01:24:04 @samth I think in the end the answer to Lean's success among mathematicians is the same as the answer to Python's success among machine learning experts. Good libraries and frameworks---@TensorFlow, @PyTorch, and so on---and good communities tightly coupled with them 2021-06-19 01:19:52 @samth This also often manifests as erasure of history, and I think success that comes at the cost of others is not success we should aim for 2021-06-19 01:18:56 @samth And there are downside's to mathlib and the Lean community in their defensive rejection of and anger toward all other proof assistants, in particular that makes it absolutely impossible to talk to anyone about common work to improve both tools, which I often want to do 2021-06-19 01:18:18 @samth Yeah, but everything before mathlib being wrong is another one of those hyperbolic and harmful statements. There have been a number of other communities like this that have widely adopted proof assistants 2021-06-19 01:17:07 @jaycech3n @samth This also confuses me, given that it's not even a classical logic? Lean's support for classical math is exactly because of its library support---again, a really good lesson for other proof assistants, but one that is always missed. Some mathematicians even think Lean is classical 2021-06-19 01:15:56 @samth The hyperbole harms everyone else, though 2021-06-19 01:15:35 @samth I think mathlib is a wonderful success story for how to build a really great and invested user community with a good voice to proselytize to potential users, and use that to continue to grow a really wonderful library and automation in that library 2021-06-19 01:14:10 @samth And shitting on Voevodsky is just incredibly low, the dude died, of course his contributions stopped. He had a fields medal of all things, is that not fashionable? It's strange to me to define fashionable math as in particular only the math Kevin Buzzard cares about 2021-06-19 01:11:07 @samth Literally to have good math libraries you need a good community of math users, then you can grow the community, but Coq never had that community 2021-06-19 01:09:49 @samth This part doesn't make sense, because the HOL-based proof assistants didn't do that 2021-06-19 01:07:45 @samth Though Buzzard helps, people pay attention to him so much they forget the rest of us exist 2021-06-19 01:07:19 @samth It's because of the mathlib community building efforts I think that's pretty clear 2021-06-19 00:56:25 @Someody42 @XenaProject Historically whenever new and interesting math has been formalized in proof assistants, it has usually just moved out of math and into other fields by way of community norms 2021-06-19 00:55:33 @Someody42 @XenaProject That seems like a tautological statement though 2021-06-19 00:53:23 @jaycech3n It also makes me want to throw up a little when people pass off Voevodsky's work as niche and uninteresting, and neglect to consider it a useful mathematical contribution, when the dude literally won a field's medal and then died a tragically early death 2021-06-19 00:51:07 @jaycech3n Yes, but the very important takeaway there is that Lean has done a really amazing job at growing and fostering a community of active and interested mathematicians, and that takeaway is actually _lost_ if you neglect to discuss the context and history 2021-06-19 00:42:36 @jaycech3n Yeah, just seems like this keeps happening. How hard is it to do the basic duty of journalists to make sure you're crediting a field appropriately? And last time I spoke to Kevin about this, I told him to also point the interviewer toward other relevant interviewees 2021-06-19 00:39:58 RT @Someody42: @TaliaRinger Tbh they are not even doing a good job at paying homage to lean itself : Leonardo de Moura is only mentioned as… 2021-06-19 00:39:54 @Someody42 lmao good point 2021-06-19 00:36:24 For example, they completely revolutionized mechanized metatheory and, to a large degree, programming languages research by proxy. One of several examples 2021-06-19 00:35:29 "Proof assistants have long had their fans, but this is the first time that they have played a major role at the cutting edge of a field" this is incorrect @XenaProject, or at the very least, far too broad of a statement 2021-06-19 00:34:57 @jaycech3n "Proof assistants have long had their fans, but this is the first time that they have played a major role at the cutting edge of a field" this is also just blatantly wrong 2021-06-19 00:31:23 @jaycech3n It also just seems intellectually honest when you mention one new language with a long legacy to be like "and here are the things that made this possible 2021-06-19 00:30:08 @jaycech3n I don't know, even the 4CT history glances over the fact that it was actually formalized inside of a proof assistant after not just done by brute force automation. Also there's a fairly big community of mathematicians in Europe who work with Isabelle/HOL or the HOL variants 2021-06-19 00:28:20 @bakirtzisg I'd look here for related work up to 2019: https://t.co/KpsAsw5geb And then look at the things that cite that work to get updates to 2021, and hopefully you'll see where the community is centered for your style of work 2021-06-19 00:11:40 @sonicisdead Yeah it seems extremely intellectually dishonest at this point 2021-06-19 00:09:23 RT @sonicisdead: @TaliaRinger it bothered me a lot when I attended SUMS (a math conference for undergrads) and being the only CS/Math perso… 2021-06-19 00:04:02 @samth Refine your comment so that you're making a point and not trolling, please 2021-06-19 00:03:47 @samth It's still very silly 2021-06-19 00:02:23 @samth Then maybe just don't make purposely inflammatory comments? Be more specific. Lean has seen a lot of success in one particular community, and there is a lot to learn from that, but it's standing on the shoulders of giants 2021-06-18 23:58:57 @samth Sorry to be harsh but this is the kind of comment I feel like you could do just a little bit of reading and understand why it's wrong, but it also endangers literally my entire field to make it so then I'm forced to respond to argue and I have things to do 2021-06-18 23:56:48 @samth And from the history, and how impossible Lean would have been without those decades and decades of prior work, and from the mathematicians working in other proof assistants (fairly common in Europe, though not on this scale), and all of the other ongoing proof assistant work 2021-06-18 23:55:46 @samth I think this is the sort of nonsensical comment one can only make if totally disconnected from most of the proof assistant community 2021-06-18 23:42:50 @tummycom I feel like this is the responsibility of the reporter, @dcastelvecchi 2021-06-18 23:41:55 Anyways, maybe one day a scientific reporter will notice that I exist for my actual expertise in the entire history of proof assistants and not just to write articles about me getting harassed 2021-06-18 23:39:19 So far every pop sci article about Lean for math has either 1) blatantly erased this history, or 2) discounted it as old and irrelevant. This means we as a community don't get to learn the right lessons from Lean's success, and it's quite bad 2021-06-18 23:37:16 Lean has done incredibly well at building a tightly knit and dedicated community of mathematician users and contributors, with libraries and automation to help them out. Other proof assistants could learn from this. But Lean got to learn a LOT from other proof assistants 2021-06-18 23:36:06 It still feels a little dishonest to not at all discuss the history of proof assistants when talking about Lean, given just how many decades of work Lean builds on. I don't really know why this happens with every pop sci article about Lean. https://t.co/KpsAsw5geb 2021-06-18 23:35:09 @tummycom I mean I wrote a whole book on it 2021-06-18 23:31:20 I mean that earnestly, you may have fun talking about future possibilities together 2021-06-18 23:30:57 "Proof assistants can’t read a maths textbook, they need continuous input from humans, and they can’t decide whether a mathematical statement is interesting or profound — only whether it is correct, Buzzard says." @XenaProject you may enjoy talking to @ChrSzegedy 2021-06-18 23:29:34 I also long for the day when someone writing one of these articles so much as asks me for my expert opinion lmao 2021-06-18 23:27:51 Mixed feelings as usual: It's amazing and wonderful to see proof assistants get this much coverage, but I still really wish any of these articles would pay better homage to the history of the field, the mathematicians using other proof assistants, and so on. https://t.co/vamn5xHEMf 2021-06-18 22:48:29 RT @SigplanM: Help us match our 370th mentee! Make a difference in someone's life! And lure them into the wonderful world at the intersecti… 2021-06-18 22:43:49 If you're interested, you can sign up here: https://t.co/YdYAk14tfq Please tell all of your ML for PL and PL for ML friends. You're popular right now. How often do you get to be this popular? 2021-06-18 22:41:59 We have lots of great mentors at this intersection already, like @ChrSzegedy, @Skiminok, Mangpo, and @sanchez_stern. But we need more! This area is so popular right now, and so many people want to go into it 2021-06-18 22:40:39 I know lots of people following me care about both machine learning and programming languages, because boy do those conversations get lively. Are any of you willing to mentor? We are still short on a mentor for a mentee who is interested in ML & 2021-06-18 21:41:45 RT @OwoTizusa: I hope this won't hurt people https://t.co/enfDKXJwGx 2021-06-18 20:47:46 I agree there are more things he should apologize for. I hope this is the start of a healing conversation, though. For now, I think it's encouraging, though my opinion matters little since Matthias has never harmed me in any way. https://t.co/om6DdFhDMk 2021-06-18 20:46:21 So I hope that this turns into a conversation that helps the community heal and Matthias grow as a person. For now, the show of good faith is something I strongly appreciate in a community that often has an unapologetic culture. But I was not harmed, so my opinion matters little. 2021-06-18 20:44:06 I think accepting the apology in good faith---unless the person has a history of using apologies abusively to gaslight, then repeating the abuse---is generally a good thing to do. But it's extremely valid to expect more. The ideal is for it to turn into a conversation. 2021-06-18 20:43:08 Importantly, though, this also means that an apology is still worth it even if it isn't the once-and-done slam dunk you're looking for to heal your relationship with the other person, to help the other person feel better, to learn to be a better person next time. 2021-06-18 20:41:14 Of course, in the end, the apology should always be for the other person, and not for you. That means that the other person may not forgive you, and should not be expected to forgive you, and you should accept that upfront. It is still the mature thing to do. 2021-06-18 20:40:16 More importantly, I think, it means that an apology should be open to revisions. If someone reads it and says, "actually, I'm also hurt by this, and you didn't address this," you should be willing to talk to them about that, reflect on that, and apologize for that, too. 2021-06-18 20:39:15 That means, of course, that the apology loses its weight if it is followed by actions that directly reverse course and inflict more harm. (This is formalized in the Jewish system of apologies, in which our higher being only forgives you when you do not repeat the mistake.) 2021-06-18 20:37:57 But it's unrealistic to expect those guesses to be perfect. So I think it's better to look at a good apology like a hypothesis---this person has really thought hard about what they have done, and wants to signal that they want to make things better and change going forward. 2021-06-18 20:37:16 When apologizing, you sometimes have to make some guesses about how other people are hurt, especially if the people hurt do not want to talk to you about it. In that case you should take a lot of time to do your best at this 2021-06-18 20:36:08 No apology is perfect from the start. But a good apology signals good faith and opens up the conversation toward internal change, healing of those harmed, and repair of the relationship and community. The better the apology, the better the signal. 2021-06-18 20:35:35 A tangent from Matthias' apology, some quick thoughts on apologies. 2021-06-18 20:31:45 @yawaramin It takes a lot of time and effort to get better at writing genuine and complete apologies---for a lot of us, it's a skill we learn over the course of our entire lives. Hopefully this at least starts a conversation that opens a path toward changes and healing 2021-06-18 20:30:14 @yawaramin But these days I'm just so happy to see any apology attempt at all. There is such a huge culture in computer science of doubling down, getting defensive, gaslighting, or just expecting people you hurt to let go while you quietly move on with your life. 2021-06-18 20:29:25 @yawaramin I think it's counterproductive to expect a perfect apology from people who often don't apologize the first time around, and I think he really tried hard. I can understand if the person harmed does not forgive him 2021-06-18 20:10:26 @sonicisdead @16kbps Yes, this is also something for which he ought to apologize and change his behavior 2021-06-18 20:07:47 @16kbps An apology from the rest of the team for not stepping in sooner is also warranted, and I hope we see that soon 2021-06-18 20:05:49 @16kbps Yeah, I assume it won't be the end of the conversation, but it's a start and a show of good faith 2021-06-18 19:34:12 People reach out to me fairly often about replicating @SigplanM for other fields. I'm always really excited about this, and hope these replicas flesh out soon. I was thinking of writing a blog post about what it was like to start the program, and lessons so far 2021-06-18 19:21:09 @fenderglass @samth Thanks! 2021-06-18 19:15:49 @wilton_quinn It is!!!! It's so good! 2021-06-18 19:13:33 So wonderful and mature I love this 2021-06-18 19:11:32 Matthias Felleisen apologized! This is wonderful. Apologies are great https://t.co/LuJj7sEBee 2021-06-18 17:34:23 A human in the loop is another form of complementary information, because the human can vet the patch and have an internal understanding of why it works 2021-06-18 17:33:00 @mustach_io It's a good paper, just often misinterpreted to mean "program repair is dead" when it really means "it is prudent that program repair tools consider some information beyond just passing tests" 2021-06-18 17:31:58 Domain knowledge is another form of explanation that can be leveraged, which is why program repair often is tractable in very specific domains even in lieu of specifications or examples. But tests alone with no other information are a terrible oracle 2021-06-18 17:29:12 @CoderPankaj @michael_w_hicks In practice, this is rarely true 2021-06-18 17:28:20 https://t.co/vb6rqkjdCh 2021-06-18 17:28:02 More formally: https://t.co/GyBEJdDHKw https://t.co/bs2ibiFJzH 2021-06-18 17:27:25 @michael_w_hicks The lesson being precisely that just changing things to pass some tests isn't good enough and can be outright harmful. You want new information/explanations beyond that (example patches, specifications, etc) 2021-06-18 17:25:50 @michael_w_hicks A good way to make this lesson official is by way of the famous review of patches in test-based program repair: https://t.co/GyBEJdDHKw 2021-06-18 17:10:00 RT @afd_icl: The PLDI conference site is now ready: you can log in and go hang out with people in Gather! I just went in but was the only o… 2021-06-18 15:47:23 RT @dataandpolitics: MOTION: let’s also make Stonewall Day (June 28) a national holiday. 2021-06-18 08:12:13 Oh and he joined before it was an official committee, just as a volunteer. I gave existing volunteers first dibs at being committee members when we became official. But for sure none of them did this for the glory, haha 2021-06-18 08:08:55 Yanze joined @SigplanM as a committee member before he even started his Ph.D. And his help has been absolutely indispensable. @ubc_spl is lucky to have Yanze! https://t.co/XR6mXgDj7k 2021-06-18 08:03:29 @liyzunique @ubc_spl Congrats to UBC! They are lucky to have you 2021-06-18 08:02:58 RT @liyzunique: Super excited to share that I'll join UBC @ubc_spl for my PhD this fall! 2021-06-18 07:59:37 Let's make all conferences hybrid from this day forward https://t.co/AO4efU9SKi 2021-06-18 07:46:49 The famous logic programming guy obviously 2021-06-18 07:43:10 Second time I'm on a list with Sergey Brin https://t.co/3SQy4Oeh3I 2021-06-18 07:34:30 https://t.co/kAbL3u46LD 2021-06-18 07:31:25 Hey cool someone at @umdcs highlighted me on the alumni page . Honored and grateful to have been introduced to computer science by such wonderful people at Maryland to begin with https://t.co/5abr06AkGQ 2021-06-18 07:21:58 @mjskay If it feels moral I'll just do the thing it's hard for me to stop and think about myself because it's hard for me to stop and think at all, and my feelings about what things do for other people are much stronger 2021-06-18 07:21:01 @mjskay I think the problem is that this extends very far out of my research and into things like, say, the incident that led to me getting harassed 2021-06-18 07:16:21 In the ADHD sense I just do what feels right always 2021-06-18 07:09:52 The whole ops team @completelysound @samps @mathur_umang @HeyHeiko @liyzunique and I have joined together to do these matches. Team effort!! 2021-06-18 07:08:27 I could use some fear or inhibition internally for at least the years leading up to tenure honestly 2021-06-18 07:07:11 More @SigplanM emails going out tomorrow. Only two PLDI mentees not yet matched! Will recruit tomorrow, then revisit the waitlist and remaining rematches. Goal to have nothing pending after PLDI at all. Godspeed though 2021-06-18 06:54:37 RT @ZJAyres: We've got to leave these hugely outdated stereotypes about mental illness and people lashing out and being violent. Yes, some… 2021-06-18 06:10:42 @ShikshaWrites The weekly allowance thing was weird. I mean I like being paid for my labor but that's a bit presumptuous haha 2021-06-18 06:05:57 Where do these people come from holy shit https://t.co/B6tH0GZe7b 2021-06-18 05:57:10 RT @GabriellBirchak: BREAKING NEWS: 2021-06-18 01:41:41 RT @WholesomeMeme: https://t.co/XNdvavInQN 2021-06-18 00:52:52 RT @SigplanM: You know who else is in super high demand? Mentors at the intersection of programming languages and machine learning. Super p… 2021-06-17 23:59:02 RT @TravisoGilbert: @NerdyAndQuirky https://t.co/KHRVCfBnw3 2021-06-17 23:43:24 @PTOOP Please do 2021-06-17 22:22:32 @PTOOP Maybe you know others you'd recommend? :) 2021-06-17 22:21:59 @PTOOP What if you and the mentee agree to just never communicate by email? 2021-06-17 22:21:25 Hi @carloangiuli, I think you would make an absolutely amazing mentor 2021-06-17 22:20:53 Hey @PTOOP I think you'd make a good mentor, even though you probably don't think that of yourself 2021-06-17 22:17:59 Come on this is literally my community haha https://t.co/chIpNR8Xeo 2021-06-17 22:17:45 RT @SigplanM: We need a few more mentors who are interested in dependent types! Please sign up below if interested, and spread the word!… 2021-06-17 21:53:41 Open offer: If you're looking for an advisor match in PL/SE research, or for an internship, and Tweet what you're looking for and DM me, I'll retweet (and tag anyone I think is relevant off the top of my head). 2021-06-17 21:09:38 @roeschinc @satnam6502 @apaszke @ChrSzegedy @jeanqasaur @atombeast @ezyang @byroncook @Skiminok Is there an official process, or should interested students just contact you? 2021-06-17 21:05:37 @satnam6502 @apaszke @ChrSzegedy @roeschinc @jeanqasaur @atombeast @ezyang @byroncook @Skiminok Oh my gosh Samuel Gruetter you are quite lucky . Let me know if anything opens up in the future. Through SIGPLAN-M, students ask me about this kind of thing pretty often 2021-06-17 21:02:15 @atombeast @satnam6502 @apaszke @ChrSzegedy @roeschinc @jeanqasaur @ezyang @byroncook @Skiminok Does V-SPELLS have potential slots for interns budgeted? 2021-06-17 20:54:14 Oh and @wilcoxjay @SagivMooly 2021-06-17 20:52:54 I'm curious about: @satnam6502 @apaszke @ChrSzegedy @roeschinc @jeanqasaur @atombeast @ezyang @byroncook @Skiminok And more, of course. 2021-06-17 20:47:26 What industrial teams are willing to take off-cycle interns for students with exceptional circumstances? Especially teams interested in programming languages, compilers, dependent types, proof assistants, formal methods, proof automation, program synthesis, and so on? 2021-06-17 19:19:39 From Jubi: “[SIGPLAN-M is] not only helping mentees, [but] also helping mentors become better human beings.” 2021-06-17 19:18:05 Mentoring is great not just for the student, but also for you. Don't take my word for it, though---take @jubitaneja's word for it instead https://t.co/NNRyCxIt8I 2021-06-17 19:16:38 We would still benefit from more long-term mentors going into @PLDI! We now have over 400 interested mentees. Help grow the community! https://t.co/MgMJLTQLFh 2021-06-17 18:40:44 It would obviously be ideal if I didn't need power to cede it or lift people to begin with, though 2021-06-17 18:39:10 What I honestly enjoy about having power is the ability to: - cede it to people who deserve power but don't have it - lift people who deserve power but don't have it - easily build connections between people - connect good people with opportunities - improve the community 2021-06-17 18:31:50 @TheGradient @infrared_ether @kheimerl Also, @dj_hci, I am not sure who your advisor is. And @Anne_Spencer_ for you directly 2021-06-17 18:30:02 @TheGradient I think this would be good for @infrared_ether's nonprofit, @kheimerl 2021-06-17 18:26:26 @TheGradient https://t.co/8lW2kugjOp 2021-06-17 18:24:47 @miniapeur Good point, I should clarify that. It means that if students tell me about specific things within UIUC, like sexual assault, I will be obligated by the university to report even if the student does not consent to it. It's a really harmful policy 2021-06-17 18:21:49 https://t.co/h6djcX37Tq 2021-06-17 18:20:30 I now discuss my history with mental illness directly on my website, to make it clear I'm willing to talk to students (modulo mandatory reporter status within UIUC starting in the fall, unless I manage to push reform on that) 2021-06-17 18:08:50 @GoogleAI @mmitchell_ai With informed consent https://t.co/ttGaMWMMeK 2021-06-17 18:07:15 Meg does make a good point though https://t.co/Vpw2TmgJ2V 2021-06-17 18:06:36 @mmitchell_ai @GoogleAI Yeah I'd be happy to talk more, we can move to DM if it is better 2021-06-17 17:56:21 @mmitchell_ai @GoogleAI I think the right thing to do is ask you whether you'd like credit, then oblige. 2021-06-17 17:45:53 RT @lindsey: really nice dissertation on hs-to-coq by @fancytypes's student, Antal Spector-Zabusky https://t.co/jgGbmQKhIn 2021-06-17 17:45:02 RT @VeryBadLlama: sorry I missed your call, 95% of the calls I get are from robots trying to steal my credit card information so this entir… 2021-06-17 17:27:26 @GoogleAI @mmitchell_ai Context: https://t.co/I1ChEXVuR0 2021-06-17 17:20:53 Integrity & 2021-06-17 17:10:47 @GoogleAI @mmitchell_ai From https://t.co/W4XXoUNM00 2021-06-17 17:06:46 RT @natematias: You want your research to make a difference, but you're unsure how to balance that with the demands of tenure as an assista… 2021-06-17 16:56:26 RT @NathanEYates: @JenniferJJacobs Biden's list of 16 industries that Russia is forbidden to hack: 1. Ice cream 2. Ice cream 3. Ice cream… 2021-06-17 16:53:19 @satnam6502 @krismicinski @peter_sewell_ From a teaching review I got for undergrad compilers, in 2016: "Talia talks like she's never excited about anything" 2021-06-17 16:51:24 RT @AllezLesBoulez: the moral panic about critical race theory is just this, basically https://t.co/sQuQ4gEbIg 2021-06-17 08:24:01 @GoogleAI Scientific excellence means giving credit to @mmitchell_ai for her work. https://t.co/MvbqL11CiG 2021-06-17 07:38:27 #adhdtwitter 2021-06-17 07:38:02 Making phone calls https://t.co/REGQsWk5rr 2021-06-17 06:48:37 RT @marileezafari: This is why Yang's comments infuriate me. Mentally ill individuals are walking among you. We exist. We don't need to be… 2021-06-17 06:46:06 Remember when that harasser guy called me "unhirable" for my mental illness? Hahaha joke's on him https://t.co/cukvoYanVX 2021-06-17 06:43:23 OK I'm taking off the title to signal that it was just celebratory and it is cool to call me "Talia" (I quite like my name) 2021-06-17 06:32:46 @Davidbrcz @GrammarTable This is what my friend does in real life, and he started before GoT hahaha 2021-06-17 06:31:41 Talking about my mental health has put me in danger before too. But we will destroy the stigma one voice at a time. (Better harassment laws would be nice, though) https://t.co/SOGXoRTcbB 2021-06-17 06:29:02 @ZJAyres I appreciate you https://t.co/W3ohi62Wvt 2021-06-17 06:28:36 RT @commieactivity: Maybe a hot take? But I absolutely need the phrase "how can you expect anyone to love you if you don't love yourself?"… 2021-06-17 06:23:40 @GrammarTable I said "less" until a college friend corrected everyone around me to say "fewer," and now I hear his voice whenever someone makes that mistake 2021-06-17 06:22:30 RT @GrammarTable: If you would pick "fewer" in the sentence below, at what age do you think you started doing that? "I have _______ (less,… 2021-06-17 06:21:26 RT @MalvikhaM: This pandemic has had a huge impact on the mental health & 2021-06-17 05:55:50 I think they just experience things much more slowly than we do 2021-06-17 05:53:08 I believe plants have qualia and these sunflowers are probably having a pretty bad time https://t.co/yp5NZdWmqZ 2021-06-17 05:45:39 @CaseyHo Overall point stands, I just do think hospitals matter... just in addition to everything else, and with good treatment and so on 2021-06-17 05:40:02 @CaseyHo Beds are really important though, in addition to everything else, at good mental health hospitals. I've been hospitalized before, I needed it 2021-06-17 05:30:26 RT @anthonyocampo: this you? https://t.co/RKs3tzVHM7 2021-06-17 05:30:11 RT @kingdeficit: this you? https://t.co/svErujBdPQ 2021-06-17 05:28:52 RT @zei_squirrel: they actually did the "our concentration camp guards are diverse" meme 2021-06-17 05:28:07 RT @LambdaLegal: Remember Roxsana Hernández—the trans woman who died in ICE custody after not receiving proper treatment for her HIV? We do… 2021-06-17 05:27:30 RT @actupny: Roxsana Hernandez and Johana Medina León were two trans HIV+ migrants that died in ICE custody. Still no justice. 2021-06-17 05:20:04 @ICEgov It's not cute when our concentration camp organization appropriates pride. Gay crimes against humanity are still crimes against humanity 2021-06-17 05:17:23 It's 2021 and even our concentration camps are flying the rainbow flag. GDIAF ICE 2021-06-17 05:15:23 RT @ggjalliance: @ICEgov so far this administration has “rejected calls to protect one of the most vulnerable groups of trans people: migra… 2021-06-17 05:15:16 RT @RainbowLegion: @CaseyExplosion @ICEgov https://t.co/f7iy7bMFks 2021-06-17 05:15:04 RT @AlexandraErin: @ICEgov It's a very touching message you have here, that queers bleed in all colors of the rainbow. This Pride season,… 2021-06-17 05:11:09 @luna_lun_a @ICEgov Thanks 2021-06-17 05:09:59 https://t.co/fqt8g0th0a 2021-06-17 05:09:25 LGBTQ immigrant rights are LGBTQ rights https://t.co/q1wdDv2WIg 2021-06-17 05:07:31 @OptimistsInc I had the even greater privilege of my parents being able to pay for a state school 2021-06-17 05:06:18 @ICEgov Abolish ICE https://t.co/mTaRBuQiVV 2021-06-17 05:05:31 @ICEgov How many of your asylum seekers are LGBT human beings seeking better lives? How many of them do you lock up in dangerously tight quarters during a pandemic? 2021-06-17 05:04:08 @ICEgov Abolish ICE https://t.co/AVcE81zk0T 2021-06-17 05:02:51 @ICEgov Abolish ICE https://t.co/wNx0nL96w3 2021-06-17 05:01:46 @ICEgov Abolish ICE https://t.co/il0pNOvppN 2021-06-17 05:00:50 @ICEgov Abolish ICE https://t.co/aOoM87GPMp 2021-06-17 04:58:48 I need help finishing cake before it goes bad 2021-06-17 04:53:34 @OptimistsInc It's so nice to know we can get a late start and still do well. I must confess this required money, though. There was no way I was getting a scholarship to college 2021-06-17 04:51:54 RT @_pdarragh: if you think you have the right to be a leader in your community, then you also have the responsibility to be a POSITIVE lea… 2021-06-17 04:51:46 RT @_pdarragh: ugh, this is all just upsetting. toxic people shouldn't be tolerated. you shouldn't have to tell newcomers to your community… 2021-06-17 04:34:07 @OptimistsInc Sneaking into the Hookah bar, getting drunk at older friends' houses. Ah, the teenage years 2021-06-17 04:31:27 @OptimistsInc I loved skipping class. To go play DDR, to go get coffee, to go protest in favor of same-sex marriage. I loved the freedom and rebellion 2021-06-17 04:29:52 Then got an ADHD diagnosis, found computer science, became an extremely serious student, did very well 2021-06-17 04:29:18 I made all of my friends through competitive Dance Dance Revolution instead where people were mostly older, and also basically everyone was gay and/or neurodivergent like me. Then we'd go and party together, my party phase of my life was ages 16-19, then it was out of my system 2021-06-17 04:27:30 RT @OptimistsInc: @TaliaRinger Mine was from when I skipped class and went to walmart or something with my friend where I bought a pet fish… 2021-06-17 04:27:17 @OptimistsInc Hahahhahahaha I love it 2021-06-17 04:23:19 I profited first though 2021-06-17 04:23:08 My favorite detention was for running an underground illegal candy dealing business at school when RI banned the school from selling candy 2021-06-17 04:21:01 Also got in trouble literally all the time and most of my teachers hated me 2021-06-17 04:20:24 I had neither the diagnosis nor the challenge though so I just skipped school and got bad grades until college 2021-06-17 04:19:04 Twice Exceptional feels https://t.co/qP0a24LDW1 2021-06-17 04:18:46 @ravenscimaven Yessss Twice Exceptional kids 2021-06-17 04:17:37 RT @ravenscimaven: I made a YouTube video about what we're talking about here https://t.co/ehx4tyrpCP 2021-06-17 04:17:32 RT @ravenscimaven: Yep we are alike in that way. Staying busy keeps me "focused" but I only put things that I am passionate about. I say "n… 2021-06-17 04:17:20 RT @ravenscimaven: I have attention deficit hyperactivity disorder and it manifests in many ways including this https://t.co/5X19Kcbh6C 2021-06-17 04:11:53 @mmitchell_ai I suppose maybe that could get your friends in trouble if you know and answer that, so maybe better not to do that publicly actually 2021-06-17 04:10:24 @mmitchell_ai Do you know if your name was initially on here, like have you asked your friends you worked with on this? Was it removed by PR? 2021-06-17 04:03:00 @satnam6502 These look good to me honestly 2021-06-17 03:59:49 @mustach_io @krismicinski @samth Honestly though the hardest parts of grad school weren't grad school, they were an abusive colleague and then a pandemic immediately after. I need to make sure my students all have external systems of emotional support (I'll help too but shouldn't be the only one) 2021-06-17 03:57:30 @mustach_io @krismicinski @samth Yeah, remembering what it was like to be a first year will be very important 2021-06-17 03:55:29 @krismicinski @mustach_io @samth Yeah, and I'm going to care enough to want that, that's my personality 2021-06-17 03:53:58 @krismicinski @mustach_io @samth The part I'm scared of is teaching 2021-06-17 03:53:19 @krismicinski @mustach_io @samth I think there will be new and interesting twists for every new student, but I think it comes much more naturally to me than actually being a student, which was really hard 2021-06-17 03:51:01 @mustach_io @krismicinski @samth Yes 2021-06-17 03:50:24 @mustach_io @samth @krismicinski Also every meeting with a student is the most joyful and amazing part of the job, so I'm not really sure the time is time I wouldn't want to spend 2021-06-17 03:49:50 @mustach_io @samth @krismicinski I feel like I'm going to do all of this research regardless of whether I have students working with me, because I'm obsessed, so students actually save me time, after an initial investment period which varies by student. They may steer me to different work though 2021-06-17 03:47:57 Oh man suggesting any change in funding models, even something totally complementary to existing things, sure always starts arguments 2021-06-17 03:46:26 @samth @mustach_io @krismicinski Because if I screw up with funding the student can have a really bad grad school experience and I don't want that 2021-06-17 03:46:01 @samth @mustach_io @krismicinski I'm not saying I'm going to take on bad students, I'm saying that I would otherwise probably take on only exceptionally good matches, and way fewer of them, early on without the extra funding. I'd be nervous to even do that, I'd be so afraid of making a mistake 2021-06-17 03:44:49 @samth @mustach_io @krismicinski If you have enough money, that's true 2021-06-17 03:43:28 @mustach_io @krismicinski @samth I think with "just" startup, I'd be extremely conservative about students early on 2021-06-17 03:42:47 @mustach_io @krismicinski @samth But a dedicated program should be flexible about the time frame within which the money is spent 2021-06-17 03:42:14 @mustach_io @krismicinski @samth Going through this now, and with the extra funding, I'm a bit less afraid of making a mistake taking on a student, so I actually think it helps me get a student. Plus "can I coadvise your student, I'll pay for a year" is a better sell than "can I coadvise while you pay" 2021-06-17 03:38:40 @mustach_io @krismicinski I generally think funding independence from your employer is almost always ideal 2021-06-17 03:36:57 @mustach_io Yeah, a program meant for incoming faculty probably should be blinded 2021-06-17 03:35:35 @mustach_io @krismicinski Yeah just one that depends strongly on your particular employer 2021-06-17 03:34:31 @samth @krismicinski @mustach_io Yep this is another dimension along which startup varies dramatically by institution 2021-06-17 03:33:24 @krismicinski @mustach_io Yeah but the advantage at the high end is real 2021-06-17 03:31:58 @mustach_io @krismicinski I think having good partners in the application is important. But a program dedicated to this like the NIH one could just be individual, and could even provide program-affiliated assistants to help individuals apply 2021-06-17 03:31:00 @mustach_io @krismicinski Thanks, I'm below in some other ways though (no best papers, just one PLDI/POPL paper) so I really appreciated the chance to shine where I'm best (ideas, planning projects) 2021-06-17 03:28:53 @krismicinski @mustach_io So I'm strongly in favor of some outside funding that isn't school-dependent. But mostly I think more money helps you treat students more fairly, and that's the most important thing 2021-06-17 03:27:51 @krismicinski @mustach_io Schools I heard from ranged 3-6 IIRC. The range is much larger than you would think. 2021-06-17 03:26:02 @krismicinski @mustach_io I wouldn't recommend a DARPA grant for most people, I'd recommend a dedicated program. In my case I deliberately formed a team with Galois since they can help a lot with deliverables and administration 2021-06-17 03:24:12 @mustach_io I'm confused what prestige has to do with it. Blind applications by university, maybe? I formed the team for the grant application myself, and wrote the UW portion of the grant application 2021-06-17 03:22:33 @krismicinski @mustach_io Obviously negotiated but not everyone can do that 2021-06-17 03:22:12 @krismicinski @mustach_io It can differ pretty substantially. It did in my offers 2021-06-17 03:20:24 @mustach_io It just lets me use my funding much more justly, giving my students a lot of security 2021-06-17 03:19:19 @mustach_io "Enough" is often not enough to, say, handle unexpected adversity, pay your undergraduates, or not worry about forcing your students to TA or intern more than they actually want to 2021-06-17 03:17:32 @samth I think it requires being PI somewhere within some time frame, but lets you start your own lab somewhere new entirely or something. I don't remember. @fenderglass do you remember what program it was? 2021-06-17 03:14:04 @mustach_io That's dependent on your university, though. And I don't want to talk specifics publicly, but you can get a lot of security to help your students if you pad that with external funding 2021-06-17 03:11:42 @krismicinski @xwangsd Yeah this is part of why I think dedicated programs are better 2021-06-17 03:09:55 @samth I think this is a different one, it was specifically meant to help people skip postdocs, I helped a former partner apply for it 2021-06-17 03:08:52 Even if you don't get a faculty position in that time frame, it should be pretty easy to rig a postdoc somewhere when you come with guaranteed funding lol 2021-06-17 03:06:50 In the NIH one, basically funding is contingent on becoming a PI within some time frame after the award. I think this is a nice model 2021-06-17 03:05:20 @xwangsd (It wasn’t a program like that, just a special situation with really wonderful people advocating for me to get my fair share) 2021-06-17 03:04:31 @xwangsd In my case, I was so strongly involved in the grant that my work was budgeted in, and now I switch to PI at my new institution and the money moves there 2021-06-17 03:03:19 @xwangsd Typically in programs with this explicit goal, the funding is conditional on *eventually* becoming a PI somewhere, within a year or two 2021-06-17 03:02:11 @xwangsd Yeah so NIH has one you can take with you anywhere to help build your lab, like padding on a startup. I like that concept 2021-06-17 03:00:58 If acceptances are early enough, it gives senior grad students who don't have as many top papers an additional way to signal skills useful to being faculty on an application. And something to talk about during interviews 2021-06-17 02:57:23 I also think it's a good way of letting senior grad students test the waters of applying for something like that before the stakes of failure are higher 2021-06-17 02:55:56 I'm not aware of anything in CS like that 2021-06-17 02:55:24 It's really nice starting as faculty with a grant aside from my startup already. I think we should start more opportunities that make that possible. It's very freeing early on. I remember NIH has something meant solely for incoming faculty, I'd like to see that more 2021-06-17 02:21:43 @SMukherjee89 From a position of power, tone matters, but speaking up or laterally in a power hierarchy, I don't believe in policing tone, it's used to silence the legitimate complaints of people from marginalized communities too often 2021-06-17 02:11:58 Also, after finishing extreme exercise like a long race in the summer heat, you should immediately find someone who didn't race and try to have a conversation with them. You won't know if you're incoherent, but they will 2021-06-17 02:04:53 Yay I can walk again 2021-06-17 01:55:52 Oh and you can also do things like put a bag of ice in a hat and let it melt on you as you cycle or run, or soak a sponge in cold water and squeeze it all over yourself when you get too hot, or pour water on your forehead 2021-06-17 01:52:04 Anyways, kids, drink water and wear sunscreen when you're cycling in the summer heat 2021-06-17 01:51:16 After some time they let me use a bag of ice to cool down further, and started slowly giving me liquids. I luckily recovered 2021-06-17 01:50:07 There they felt my skin, took my temperature, and seemed extremely concerned. So they took some towels, soaked them in water, and put them on my forehead and that was the most blissful and amazing feeling imaginable 2021-06-17 01:48:51 Then I collapsed and he called the medical tent guys, who put me on a golf cart and asked me questions like what my name was and how old I was, which I couldn't answer. I had no idea, I just wanted the ice, why were they taking me away from the ice pool? 2021-06-17 01:47:17 He looked confused, then I repeated myself "ICE. ICE" 2021-06-17 01:46:54 Eventually I walked up to someone to ask him where the ice pool was. But I couldn't form sentences. So I just looked at him and said "ICE?!" 2021-06-17 01:46:20 So then I wobbled around more thinking it was elsewhere, but it wasn't there either 2021-06-17 01:46:00 With determination, I wobbled off into the distance toward where I thought the pool was. But when I got there, it wasn't there 2021-06-17 01:45:21 I was also positive that there was a children's pool full of ice cubes out in the distance. So I decided I was going to find that goddamn ice pool 2021-06-17 01:44:27 I thought I was totally coherent but I was hallucinating. At that point I just felt so unimaginably hot, in a way I can't even explain 2021-06-17 01:43:44 By 3 miles in the only thing I could think about was water, I kept jogging to each water station, it was so painful. I don't remember much but I finished, and then wobbled off to the side 2021-06-17 01:42:29 Had water before the run, at which point I felt kind of bad, but didn't understand why. Started to run, and it was so, so incredibly slow, and felt worse and worse every mile 2021-06-17 01:41:24 It was an Olympic distance triathlon. Swim was great, had some water before I got on the bike, then just did the entire bike ride with no sunscreen and no water 2021-06-17 01:40:38 Age group nationals were in Omaha, Nebraska, in the middle of August, right through peak sun. It was well into the 90s. No idea why they did that but whatever 2021-06-17 01:39:41 But I was so new to cycling that I never learned two very obvious things: 1. The need to wear sunscreen 2. How to drink water while cycling 2021-06-17 01:38:52 I guess while I'm recovering, that is an interesting story. I didn't learn to ride a bike until age 24, then qualified for triathlon AG nationals in my first triathlon. I decided to do it because why not https://t.co/nJgbO6zvc3 2021-06-17 01:36:04 Slowly improving with water and dried fruit while lying down, seems like a dehydration thing 2021-06-17 01:26:01 @PrivacyLawyerD @HHMusicOfficial In contact with a friend monitoring me so it's OK now 2021-06-17 01:24:58 @HHMusicOfficial @PrivacyLawyerD I think it's OK honestly the muscles are just not good, I have water and dried fruit and I'll eat it and wait 20 minutes, if I cannot stomach it or cannot get up after that then I will call 2021-06-17 01:24:08 @anshrdas Thanks, good call 2021-06-17 01:18:44 @PrivacyLawyerD @HHMusicOfficial I'm not sure I can walk far enough to get salt 2021-06-17 01:13:50 With that said if these Tweets are incoherent please let me know. When you are incoherent, you usually don't realize it 2021-06-17 01:13:15 I've only ever had my muscles just stop working once before and it was when I had heat stroke during triathlon age group nationals. But then I was not coherent, I was hallucinating and unable to form sentences. I feel coherent and it's not nearly as hot out 2021-06-17 01:12:09 I was out on a run and got dizzy, nauseous, and weirdly my muscles stopped working so much I can't walk up stairs. Is this a dehydration thing? Not sure how worried to be. The group walked me back to the office 2021-06-17 00:25:42 (Note that being kind is not the same thing as being "nice." Kindness is about compassion and treating each other like humans, apologizing when you do harm people, and so on. Being "nice" is mostly about tone policing) 2021-06-17 00:07:59 @PTOOP If you undergraduates cotaught you kindness, does that mean you taught them kindness? :P 2021-06-17 00:07:07 When they show up later, faculty and especially those who are safest and not often tasked with this kind of labor really need to step in and teach people like this how to be kind and, if that doesn't work, 1) assure safety for those impacted, then 2) don't tolerate the behavior 2021-06-17 00:04:56 Often the warning signs show up early, when the researcher is a student, in which case the advisor really needs to teach kindness alongside other research skills. Though not everyone will listen 2021-06-17 00:03:01 I maintain that kind mediocrity is way better for the field than cruel genius. Ideally those in a position to do so need to teach even the most brilliant researchers in our community to be kind, though---that is a basic research skill 2021-06-17 00:02:07 I don't know who wrote this, but it's yet another example if the irreparable harm we do not just to other human beings but also to the field as whole whenever we tolerate smart jerks https://t.co/BXbOO3NaC1 2021-06-16 23:50:43 When you can't decide between working and relaxing https://t.co/UVkfkBfHjw 2021-06-16 23:36:16 Elba is a blessing to the community always good to see funding go to great people https://t.co/eSng1n82O3 2021-06-16 22:21:10 @znmeb I think I'm more interested in paying fairly/competitively adjusting for cost-of-living, so that it's just 2021-06-16 22:20:25 RT @ISCAConfOrg: The 2021 TCCA Young Computer Architect Award goes to @samps “for contributions to approximate computing and hardware synth… 2021-06-16 22:20:20 RT @luisceze: This is so well deserved! I could not be more proud! On top of being a world class computer systems architecture researcher,… 2021-06-16 22:10:43 @ccanonne_ I think at the least I'd take the undergrad rate from my institution and adjust it to cost-of-living where the student lives. But yeah, those are good considerations, given the timing 2021-06-16 22:07:32 @ccanonne_ But, I think it's a good idea to ask them what they usually pay as a start 2021-06-16 22:07:13 @ccanonne_ It's a bit of a strange situation, hard to explain, but he is not currently affiliated with a university 2021-06-16 22:06:13 Is there any data on undergraduate researcher hourly pay for computer scientists available anywhere? So I know I'm paying competitively? 2021-06-16 22:04:35 When people pay undergraduate researchers hourly, how much do they pay? Trying to figure out a summer rate and just totally unsure where to start or what is reasonable 2021-06-16 21:19:05 So even if we don't end up working together, chances are I'll be able to connect you with other interested people working on the same problems, if we have overlapping interests 2021-06-16 21:18:31 I'm tracking all of the ideas I have and all of the people who are interested in them (students, faculty, industrial researchers) so that I can build connections really quickly and pull teams together as soon as one starts to materialize 2021-06-16 21:16:44 RT @lindsey: @ezyang I think it actually doesn't matter so much where you start in this cycle. If you look at tools first you'll get to pr… 2021-06-16 21:16:38 @lindsey @ezyang Yesss going back and forth along this cycle is my entire research career so far 2021-06-16 21:15:54 If you're interested in working with me in the near future, this summer is a really good time to talk before I get too wrapped up with junior faculty life. I've basically been in idea generation and connection building mode all week, and it has been really fantastic 2021-06-16 20:58:39 @yisongyue 2021-06-16 20:57:32 @yisongyue That sounds really hard too. Do you mean after working together? A primary decision factor for me in choosing a job was making sure there were enough faculty doing close things that if a student doesn't work out, it's easy for them to find a different advisor 2021-06-16 20:52:44 I think the hardest part about being faculty is going to be containing my excitement about doing all of the cool things while I also teach and so on 2021-06-16 20:22:35 You know when brain threads merge? Just tried to order neurosymbolic boba on Uber Eats 2021-06-16 18:51:33 @mdekstrand what is a Dynalist 2021-06-16 18:50:52 Meta: I have a lot of ideas right now, and I'll have students in the near future, and I need a way to manage my ideas and relevant resources so that I can easily come back to them when students are interested. Any recommendations for this? 2021-06-16 18:49:54 @_ddjohnson coooooool~ 2021-06-16 18:39:15 @AriKatz20 If anything comes of it, let me know. I'd be curious about moving from e-graphs in FOL to e-graphs in cubical type theory and so on, and seeing whether the techniques carry over, and if not, what becomes difficult 2021-06-16 18:37:28 Or rather, e-graphs are data structures upon which you can easily build symbolic path solvers 2021-06-16 18:32:13 There's a deep foundational connection between path finding and rewrites along equalities 2021-06-16 18:28:28 How does that interact with NS learning? I have literally no idea, but I sure as hell want to know 2021-06-16 18:28:00 If it gets hard, it might help minimize fundamental problems, like though it's possible to deal with bindings in e-graphs (and solved for two interesting dependently typed logics), doing so probably slows down performance, and requires proof generation 2021-06-16 18:17:46 If not, I think it might be worth trying this with classical FOL first, because from those results you could figure out if the basic representation is useful, then you could try moving to higher-order dependently typed logics just by changing the representation of e-graphs 2021-06-16 18:14:15 @pascalhitzler @monireheb @mdksarker @DaSeLab @federicobianchy Yeah, if you end up interested in that, let me know. I really love building symbolic reasoning techniques that reuse existing information in constructive proofs to derive new proofs. So it would be really fun to try to move that into the neurosymbolic world and see what happens 2021-06-16 18:13:18 Is there any work on using e-graphs as a representation for neurosymbolic learning? Just wondering because these are now available for lots of cool logical systems, even higher-order dependently typed logics like cubical type theory 2021-06-16 18:00:23 I'll check these out! There's a PL-specific one coming out at some point but it isn't public yet https://t.co/75SQuTCcdd 2021-06-16 17:55:47 @DaSeLab @pascalhitzler @monireheb @mdksarker @federicobianchy What do you mean by where to go? 2021-06-16 17:54:47 @pascalhitzler @monireheb @mdksarker @DaSeLab @federicobianchy Another thing I think is interesting is that in logics like cubical, you actually have a real number interval as a type, and you have higher-order paths 2021-06-16 17:53:26 @pascalhitzler @monireheb @mdksarker @DaSeLab @federicobianchy With this you get equality solvers for higher-order dependently typed logics, from which you can implement type-directed search given a starting point and arriving at an end point, following a path of equalities, as a rewrite system. It's quite neat, want to implement this summer 2021-06-16 17:52:39 @pascalhitzler @monireheb @mdksarker @DaSeLab @federicobianchy Representing variable bindings is a challenge for symbolic techniques too. I wonder if recent work on that would help? For example, with e-graphs (the data structures used for congruence closure): https://t.co/UsIxDxyGUP https://t.co/acIQpx23dn 2021-06-16 17:51:53 @pascalhitzler @monireheb @mdksarker @DaSeLab @federicobianchy Ah, I actually think constructive logics should require much less backtracking, since there is so much information contained within proofs that you can reuse. 2021-06-16 17:45:41 At this point, if you want some updates on related work, I'd also recommend my thesis related work section: https://t.co/XlBfGkP8QZ 2021-06-16 17:44:46 Or on my website: https://t.co/KpsAsw5geb 2021-06-16 17:44:29 At this point, though, it is available for free online: https://t.co/evMqkaUApe 2021-06-16 17:43:43 Sometimes people ask me how to read the QED at Large survey paper on proof engineering. It's pretty big and overwhelming! So far that, I wrote an extended reading guide and Q& 2021-06-16 17:00:04 @techno_ben Does Twitter long form count? 2021-06-16 16:55:46 So far (major WIP) https://t.co/ogBVkPkPKH 2021-06-16 16:53:44 Midwest PL/SE folks, please remind me of all Twitter accounts to put on a Midwest PL/SE list! Should make it easier to follow relevant people and build a community. @plfmse @madpl1 @certifiablyrand @samth @purdue_pl @roopshasamanta @GhostofBendy etc 2021-06-16 16:36:10 RT @plfmse: We are working with 41 amazing undergrad and high school research interns for this summer from all over the world (Argentina, C… 2021-06-16 15:16:59 RT @codydroux: @LianaThinks @TaliaRinger @EliSennesh https://t.co/PKhJlOqNAv 2021-06-16 15:14:03 RT @QuantaMagazine: Biologists recently discovered crystals in the shoulder feathers of leafbirds that form perfectly ordered bubbles refle… 2021-06-16 15:11:16 RT @GeologyTime: fossilized Opal Wood From Indonesia Photo: Syena Septiani #fossils #minerals https://t.co/y1rH1XBA7C 2021-06-16 14:57:44 "But unlike their U.S. counterparts, Israeli leftists don’t have the privilege to be ideologically maximalist." Sums it up https://t.co/dslCUxaXCi 2021-06-16 14:46:24 RT @Stpolishook: Really REALLY important piece. ⁦@arash_tehran⁩ it actually reminded me of your similar assessment of how key it is to supp… 2021-06-16 14:36:46 RT @newscientist: Dogs that detect seizures may be sniffing out the scent of human fear https://t.co/Sv9FIAId4O https://t.co/Ol2TwDi760 2021-06-16 14:25:35 @AnneOgborn This will scare people 2021-06-16 14:19:45 @Rubberduck203 They aren't, though, in my field 2021-06-16 07:23:25 RT @backdatbairup: Yo 2021-06-16 07:10:41 @DomSteinhoefel Yeah S4 is good 2021-06-16 07:09:05 RT @Russia_NC: https://t.co/7ctLFtfDMF 2021-06-16 06:44:21 The least realistic part of the show though is that there are sometimes consequences for abusive white men 2021-06-16 06:41:56 Good show 2021-06-16 06:41:21 The Handmaid's Take season 4 theme: standing up to your abuser once safe and in a position to do so is way, way more cathartic and wonderful than the gaslighting "just move on" nonsense we are always told to do 2021-06-16 05:07:04 RT @hourlyFox: https://t.co/aEqFc0FfCC 2021-06-16 05:00:25 RT @ksmckinley: @sigarch Please join the SIGARCH/SIGMICRO CARES committee @ISCAConfOrg on Thursday at 10:00 ET and 20:00 ET to discuss ma… 2021-06-16 04:58:41 RT @icatshouldnt: https://t.co/9e42kWMs4g 2021-06-16 04:47:43 https://t.co/sTAeK8Kb8u 2021-06-16 04:42:44 Deleting dating apps to meet someone the old fashioned way (Well, somebody has to arrange the matches, Young people can't decide these things themselves. She might bring someone wonderful---- Someone interesting---- And well off---- And important---) 2021-06-16 04:10:34 In programming languages I have found that these days, most reviewers are reasonable at adapting expectations to problem difficulty, too 2021-06-16 04:08:26 When you get partway down that path, people start to want to help you, so it becomes easier to get the rest of the way there 2021-06-16 04:07:54 The field is so, so, so young that I think it's extremely likely that most researchers could make reasonable progress on most hard problems. And sometimes it's a matter of breaking it down into a path to solving the hard problem, then solving problems on that path 2021-06-16 04:06:39 I think people discount research problems for being hard way too often. Probably because it's legitimately scary when your success or your students' success hinges on good results. But I think it's rare that hard-sounding problems in CS are intractable 2021-06-16 04:00:45 @pascalhitzler @monireheb @mdksarker @DaSeLab @federicobianchy I also think they're particularly promising for practical automation: https://t.co/5uMDoikU01 2021-06-16 03:54:58 @pascalhitzler @monireheb @mdksarker @DaSeLab @federicobianchy Proof assistant world is full of symbolic automation for absolutely wild logics if you're ever interested in talking to folks who care about those use cases. (I'm in that world) 2021-06-16 03:52:27 @pascalhitzler @monireheb @mdksarker @DaSeLab @federicobianchy Interesting, what's tricky about it? Just curious. I do very symbolic things on higher-order dependently typed logics, recently complementing that with a bit of machine learning, and am interested in the merger at some point 2021-06-16 03:49:21 I have my startup and a joint grant already so it really isn't a rush, but I want maximal security for my students 2021-06-16 03:47:51 What are all the grants and awards people apply for as early CS faculty? I have no urgent need, but want a sense of what to pay attention to and plan for early on 2021-06-16 03:42:41 RT @tuna_suma: … # https://t.co/a4lpaF8axI 2021-06-16 03:40:28 @pascalhitzler @monireheb @mdksarker @DaSeLab @federicobianchy Any attempts so far at any higher-order logics? 2021-06-16 03:19:31 RT @willie_agnew: @mmitchell_ai @_KarenHao @timnitGebru Turing as a role model is hard and complicated because I feel the CS community is s… 2021-06-16 03:12:40 RT @emmavaninwegen: deleting my dating apps because I want to meet someone the old fashioned way (by defending both her and my buddy’s hono… 2021-06-16 03:06:13 @ravenscimaven You look so happy!!! It's contagious 2021-06-16 03:03:26 RT @hourlyFox: https://t.co/9LBCV3VJ15 2021-06-16 00:49:05 RT @OpenSexism: "Women face social challenges that include: missing having other women around (lack of peer parity), insulting expletives i… 2021-06-16 00:48:47 RT @OpenSexism: "In summary, the primary studies reported women’s participation ratios ranging between 4% to 14% ...across different measur… 2021-06-16 00:31:20 Like when would I ever optimize for myself rather than for the other person or for our relationship? Almost never! And when would I dishonestly communicate more than I actually want to manipulate the other person into taking an action? I can't remember ever having done that! 2021-06-16 00:30:25 When I say those assumptions are implicit, though, I mean that I first found out that people don't like ultimata in the US when I was 28 years old, and when someone explained to me why, I was disgusted by his assumptions, which sounded patently false in most situations to me 2021-06-16 00:24:21 I'm going to save the ultimata people issue to me in different communication styles and hopefully come back to this later with examples. I hope you'll pay attention and realize that you probably issue ultimata literally all the time both at work and in relationships though 2021-06-16 00:22:57 @PTOOP In Israeli culture, if you want to be a jerk, you'll just directly be a jerk lol 2021-06-16 00:22:30 @PTOOP British culture is completely impossible for me to navigate, I gave up a long time ago 2021-06-16 00:22:00 @PTOOP https://t.co/5c2VhpaMCZ 2021-06-16 00:20:52 @PTOOP Yeah, but writing this way in the US will get you in a lot of trouble when you lack power, and will make people afraid of you when you have power 2021-06-16 00:20:08 These assumptions are implicit in many cultures, so that all of those extra words just aren't necessary 2021-06-16 00:19:44 Politeness in phrasing signals you are not optimizing solely for yourself, so that the ultimatum is OK. Leaving the inference that Z implies doing Y to the recipient signals that you want to honestly give full information, but without sounding harmful and manipulative 2021-06-16 00:18:19 Rather than "if you don't do X, I will do Y," often it's "hey, hope you're doing OK! I just realized you haven't done X yet. That's fine, and I know you're busy, so I'm just curious if you'll get to X soon. If you don't get to it soon, I'll assume Z" where Z implies doing Y 2021-06-16 00:15:18 Even Americans, who eschew ultimata enough to consider them completely taboo at work or in relationships, actually issue ultimata to each other at work and in relationships all the time. They just phrase them very, very softly, so that they don't sound like ultimata 2021-06-16 00:14:36 In many cultures, and to many individuals, that completeness of information is valued much, much more than politeness in itself 2021-06-16 00:13:40 Incomplete information can be really frustrating! If you'd known something to be true, and were not optimizing solely for yourself, perhaps you would have acted differently 2021-06-16 00:13:00 If you are not being dishonest, then an ultimatum is not going to manipulate the other party 2021-06-16 00:11:29 If you are not trying to win, but are each optimizing for the benefit of the other agent or for your relationship, then issuing an ultimatum doesn't put the other agent in a bad situation. It just gives them more information to make an optimal decision for you or the relationship 2021-06-16 00:10:10 This is something that comes up often and I rarely hear people talk about. Ultimata are not inherently bad, that makes assumptions about the game played (individualistic, trying to "win"), the honesty of the issuing agent, and the value of complete information versus politeness https://t.co/TO3rIhFDBN 2021-06-16 00:06:11 @AnneOgborn Yeah it's definitely an ultimatum, it's a limited time within which the recipient must act before I will cede their spot to someone else. I'm the one doing the ceding. So I need to be upfront about this without it sounding like an ultimatum. I think I've done OK but it was hard 2021-06-16 00:03:51 @LianaThinks @EliSennesh That would also be amazing!! Then I can finally find out if people are angry at me or not 2021-06-16 00:03:01 @PTOOP I requested additional admin support for emails through my university. You may be able to do the same 2021-06-16 00:01:16 @wilbowma @ionathanch Yeah and I'd rather make some people tired than make a different set of people scared 2021-06-16 00:00:11 @PTOOP I guess but all they need to do is say whether they want to stay in the program and, if so, whether with the same mentor or a different mentor. Also have you considered that you may have ADHD? I say this as someone with ADHD 2021-06-15 23:57:27 @AnneOgborn Americans seem to really hate anything that smells like an ultimatum. It's a really strange cultural assumption I think based on raw individualism (there's no harm in an ultimatum if you're optimizing for each other or for your relationship, not for yourself), but it's real 2021-06-15 23:56:13 @AnneOgborn I disagree, at least in the context of American communication norms, which are not my default. My default scares people and I had people respond by saying they would do things I really don't want them to do just for a mentoring program because they were scared 2021-06-15 23:54:25 @CheeToS_ No, this is an email to the mentee 2021-06-15 23:52:47 Even the "I will assume you are no longer interested" is a Americanism I learned from Twitter a while back. I used to just say I would give away their slot haha 2021-06-15 23:49:24 This is me code switching from the way I naturally communicate (default when I'm tired) to the way I communicate when I think really, really hard about how to sound American 2021-06-15 23:46:19 "Spots are a bit tight these days, so if I don't hear from you in the next week or so, I will assume you are no longer interested (I will check in one more time, though, to make sure). Thanks! < 2021-06-15 23:46:03 "Hi < I hope you are doing well! Your mentor through SIGPLAN-M hasn't been able to reach you recently. I am curious if you are still interested in the program. Would you like to continue? If so, would you like to continue with the same mentor, or switch to another mentor?" 2021-06-15 23:45:19 New template for email: "Subject: Your mentor hasn't heard from you in a while (time sensitive)" (1/3) 2021-06-15 23:44:49 "Spots are limited, so if I don't hear from you in the next week, I will assume you are no longer interested (I will check in one more time, though, to make sure). Thanks! < 2021-06-15 23:44:20 "Hi < Your mentor through SIGPLAN-M hasn't been able to reach you recently. Would you like to continue through SIGPLAN-M? If so, would you like to continue with the same mentor, or switch to another mentor? ..." (2/3) 2021-06-15 23:44:08 Old template for email: "Subject: Your mentor hasn't heard from you in a while (action required)" (1/3) 2021-06-15 23:42:27 Wanted: A tool in my email that translates the English Israelis and Israeli-Americans write into the more polite and indirect English many Americans write https://t.co/m9ASwbLU8w 2021-06-15 23:38:01 @PTOOP Twitter: free dopamine machine, cool research ideas, fun, socialization, activism, procrastination, crowdsourcing papers, arguing with Conor McBride 2021-06-15 23:37:00 @PTOOP Work email: personal work communications, keep inbox under 10 at all times, move to tasks if need to take action Personal email: spam, mailing lists Slack: active research collaborations and synchronous committee communication Zulip: ahhh another thing?? help omg 2021-06-15 23:30:22 @PTOOP Zulip and Slack make sense for me for synchronous collaboration and discussion, like planning with the committee members, but not for any important communications because then you're making someone pay attention to yet another medium all the time 2021-06-15 23:29:24 @PTOOP if someone asks me to join another Zulip or Slack I just won't, it's too hard to keep track of all of these things, while email is all in one place 2021-06-15 23:26:57 @PTOOP I find it quite ideal for coordinating the mentoring program. I give mentors and mentees a choice of how to communicate with each other, though. 2021-06-15 23:25:52 @PTOOP I can't decentralize my efforts onto multiple communication media, too hard to run a program that way, and email works for so many other parts of the program. There will be a Slack soon, but only for informal chitchat, not for official communication with the committee 2021-06-15 23:24:31 There are about 400 mentees and 200 mentors, plus an active waitlist, so I can't afford to really personalize this too much 2021-06-15 23:23:57 I think I scared a mentee so I'm trying to soften this email template. 2021-06-15 23:22:20 This is in particular for mentees that mentors cannot reach for several months, to prompt whether they want a rematch, to keep the current match, or to cede their spot to someone on the waitlist (default after a week, plus a follow-up email, plus a few more days). 2021-06-15 23:20:45 @cbarrett I like that one 2021-06-15 23:20:33 @PTOOP https://t.co/eYp7fRDJBi 2021-06-15 23:19:44 @PTOOP Unfortunately not possible, managing about 400 mentees and 200 mentors from all over the world 2021-06-15 23:17:58 "(action requested)"? 2021-06-15 23:16:15 What's a less imposing version of "(action required)" in an email header coming from a position of power when I need a response within a week or so for the recipient's own benefit, but if the recipient wants to choose not to respond that will only hurt them? 2021-06-15 22:45:04 I did it I did it I finally filed all of my reimbursements from second visits. Except for all of my meals because god knows I'm not going to figure out how I paid for all of those and find proof of payment 2021-06-15 22:42:13 RT @ezyang: python changed their tracebacks from reporting line number of the end of the statement to the beginning of statement and this b… 2021-06-15 22:19:44 I'm quite proud to be involved in the first attempt at a truly hybrid PL conference, along with @JAldrichPL, @potaninNZ, @stevemblackburn, @yuyuchazuke, and the rest of the SPLASH committee and volunteers. @splashcon 2021! https://t.co/N6rynVryuA 2021-06-15 22:13:49 RT @michael_w_hicks: Outgoing SIGPLAN Chair Jens Palsberg reviews the results of the survey taken back in March about SIGPLAN conference or… 2021-06-15 22:13:24 I ordered lunch but they ran out and sent me a gift card instead. The gift card doesn't taste very good and isn't very satisfying 2021-06-15 21:54:39 Static analysis is external typing Program transformations are external rewrites 2021-06-15 20:42:54 RT @mmitchell_ai: This looks like a dataset I spent over 2 years working on (with Susanna as well, who is great). https://t.co/UI7cgQnHH0 2021-06-15 20:40:10 @julianosiloto (In a way that doesn't just trust you to be right---it shouldn't let you lie or be wrong about the shape, unless you deliberately forfeit safety which Dex lets you do) 2021-06-15 20:37:05 @julianosiloto But basically, without deliberately using unsafe features, the compiler should not let you lie or be wrong about the shape at any point 2021-06-15 20:36:09 @julianosiloto I think the difference is that there you're just trusted, with dependent types you really have to choose the shape that matches the dynamic shape, though Dex still gives you an "unsafe" way to say "I promise it's really this shape" for when you find that too hard to do yourself 2021-06-15 20:17:36 @LauraViglioni @julianosiloto (yes) 2021-06-15 20:17:18 @julianosiloto https://t.co/fdHMAcQCDp 2021-06-15 20:16:37 @julianosiloto For machine learning people, the most immediate application is that a number of languages like Dex are using limited forms of these dependent types to catch shape errors for you before you even start training. You can statically type tensor shape with dependent types 2021-06-15 20:13:20 @LauraViglioni @julianosiloto https://t.co/70TwW5g9nP 2021-06-15 20:11:45 @julianosiloto "Polymorphism" really is types that dependent on types. "Dependent types" are types that depend on terms. So you can type things like lengths of lists, shapes of arrays, sizes of trees, and even way, way fancier things. Then you can rule out lots and lots of bugs at compile time 2021-06-15 20:09:21 @julianosiloto Or you could restrict the lengths of the input lists, so you could write a function "head" that returns the first element of a nonempty list ... but actually make it _impossible_ to even call that function at all if your list is empty. The compiler can stop you 2021-06-15 20:08:00 @julianosiloto What if you could write, for example: List< where n is the length of the list? Then you could write functions, like list reverse, that take a list of length n and give you back a list of length n 2021-06-15 20:06:33 @julianosiloto Ah OK. So you know how a lot of languages of polymorphic lists? Like list< This is nice, but it's restricted: T has to be a type. What if T could be any term? 2021-06-15 20:02:37 @julianosiloto This is a preliminary question that changes how I answer 2021-06-15 20:01:57 @julianosiloto Are you familiar with types in programming? Especially static types, when your compiler yells at you for, say, using the integer 3 like it's a string when it's not? 2021-06-15 19:57:50 but like actually 2021-06-15 19:57:21 Current hobbies: - trick machine learning people into using dependent types - trick dependent types people into using machine learning 2021-06-15 19:50:46 @nairbv Probably abstracting away search concepts like beam width so that hardcore PL/DTT/proof automation people don't need to go learn anything new, which they'll resist 2021-06-15 19:48:21 @nairbv I don't want people to have to think about Python bindings, so we should probably hide those entirely. The models all predict proof tactics, probably we want a way to serve those by their compiler internal representation and not only as strings 2021-06-15 19:46:24 @nairbv Nobody in this domain would ever use Python, so part of it is going to be serializing, calling out to the models, and so on from within a language people will use (OCaml). Scale is less likely to matter 2021-06-15 19:18:45 @nairbv So I'm hoping to see some examples and find some things I like, and see if any of those ideas carry over to my domain (proof automation in interactive theorem provers) 2021-06-15 19:17:21 @nairbv Yeah, I just want examples, a student I'm advising is putting together an API for my domain and I want to have ideas to help her out 2021-06-15 19:07:42 @DrEigenbastard @codydroux @CoqLang So what we're looking at in this particular case is a way for plugin developers, so people extending the proof assistant in some way (usually with new automation), to call out to learned models in addition to the APIs we already use (type checking & 2021-06-15 19:03:12 @_joaogui1 @CohereAI Looks like also NLP? NLP world seems on top of this. 2021-06-15 19:02:27 @_joaogui1 @CohereAI What domain? 2021-06-15 18:50:49 If you're in a specific domain in which there are really, really nice APIs built around existing learned models that people can call out to without having to know anything about machine learning at all, swapping between different models as needed, I'd enjoy seeing those APIs 2021-06-15 18:30:55 Here's the branch of coq-plugin-lib we will add the model to: https://t.co/vHdqN9sWh5 Hoping this way you should be able to easily write tactics and commands that call out to both neural networks and the standard Coq compiler APIs like unification, type checking, etc 2021-06-15 18:28:27 We will separate out the APIs needed during development for our use case but don't want it to be too specialized 2021-06-15 18:23:09 Was on the fence about this, but I think it'd be cool to be able to just freely call out to a model from within a Coq plugin without having to know anything about machine learning 2021-06-15 18:19:44 We're just going to stick a learned proof tactic model into coq-plugin-lib. Any APIs that would be useful for interfacing with a model from within a Coq plugin? @CoqLang 2021-06-15 17:57:49 These are really good https://t.co/R1ZGPpGpih 2021-06-15 17:55:16 @basus Probably 2021-06-15 17:53:25 @basus Austin Powers: The Spy Who Shagged Me? 2021-06-15 17:43:09 @andrejbauer Hahaha I'll clarify 2021-06-15 17:41:44 (Up to unification etc, the problem is undecidable) 2021-06-15 17:40:37 @TobyMeadows In the loosest sense, I mean that the content of proofs matters 2021-06-15 17:39:30 Benchmark suite and training data should capture this, would be really fascinating. (I know how to implement this symbolically as a proof term transformation that works whenever there is a symmetry property to exploit) 2021-06-15 17:38:12 While it is on my mind, @ChrSzegedy, one thing I think will be really interesting about a natural language to formal proof AI system will be whether it can infer and execute things like "similarly in the opposite direction," which operate over existing proof content 2021-06-15 17:24:40 Natural language proof in most mathematical fields is a proof relevant metalanguage 2021-06-15 17:23:48 And proof irrelevance oh my gosh. So many mathematicians are taught proofs are irrelevant, and then go on reasoning about proofs in a relevant way themselves: "this proof is beautiful but this one isn't" or "similarly in the opposite direction" or "this proof is insightful" 2021-06-15 17:21:03 So many of you lived your entire lives thinking of the law of the excluded middle as some inherent truth about the world and I blame your early undergrad professors https://t.co/eOSOU3dGar 2021-06-15 17:18:06 @TobyMeadows I'd say any kind of derivation from rules in my mind is equivalent to a form of reasoning, though perhaps that doesn't mean it *is* reasoning 2021-06-15 17:17:22 @TobyMeadows Fair, I just think limiting to a single logical system is a huge mistake, entire swaths of computer scientists and mathematicians live their lives thinking the axioms of a particular logical system are just inherent truths, and never learning other ways of thinking 2021-06-15 17:03:00 @TobyMeadows Seems like reasoning to me 2021-06-15 17:02:29 @TobyMeadows I think your notion of reasoning might mean something specific? Can't any system of rules be used to reason, even if not always in a useful way? Reasoning in my mind is even broader, since it can't always be captured by systems of rules 2021-06-15 17:00:08 @TobyMeadows I'd be OK with "systems of rules from which other things can be derived" though at that point that just seems like "reasoning" 2021-06-15 16:57:03 @TobyMeadows How are sets and strings already primitive 2021-06-15 16:52:51 @TobyMeadows What would you propose? I haven't seen anything more liberal than reasoning or arguments 2021-06-15 16:52:10 Imagine if we taught students "programming" in Java but never even bothered telling them other languages exist 2021-06-15 16:51:45 Easy segues to Gödel, metatheory, properties of different logical systems, and so on. Doesn't need to dominate the class, but can be simplified and explained briefly to expand students' worldviews 2021-06-15 16:49:23 Then later you can be like, "by the way, we made these assumptions. Let's look at what kind of reasoning we can and can't do when we remove some of these assumptions, or add new assumptions" 2021-06-15 16:48:05 Lowest hanging fruit is just saying "a logic is a system of reasoning 2021-06-15 16:45:56 @andrejbauer @Lavoisierbug It didn't occur to me until I learned Curry Howard my senior year of college that there was no "logic," but rather there were logics 2021-06-15 16:45:17 @andrejbauer @Lavoisierbug I think the thing here is mostly that this is often called "logic" and not "here is a very particular logical system, or group of logical systems" 2021-06-15 16:40:10 (Since I have recent posts about depression, I should clarify this is a joke and I'm feeling pretty reasonable right now) 2021-06-15 16:36:31 A strong contributor to improving the community and a creative researcher unafraid of hard problems. Further inspection reveals a large collection of unwashed dishes in Talia's bedroom that should be addressed before the heaven-ready. Weak accept. https://t.co/CIIHkt35Rl 2021-06-15 16:32:27 RT @MIT_CSAIL: Time spent on mobile devices per day for US adults: 2021: 4.4 hrs 2020: 4.3 hrs 2019: 3.7 hrs 2018: 3.6 hrs 2017: 3.4 hrs 2… 2021-06-15 16:25:01 RT @nameshiv: one time my sister bit into a guava and and started screaming that it had a worm in it and then my mom was like "at least it… 2021-06-15 16:20:16 RT @ResistenciaNW: Hunger strike begins at NWDC while nine asylum seekers recently xferred test positive for COVID-19. ICE and GEO contin… 2021-06-15 16:19:37 RT @ResistenciaNW: Call, email & (360) 902-4111 email: https://… 2021-06-15 16:19:12 RT @ResistenciaNW: CALL TO ACTION Call & C:… 2021-06-15 16:18:59 RT @ResistenciaNW: HUNGER STRIKER DEMANDS 1. DOH must have access to the facility 2. COVID-19 TESTS MUST BE DONE BEFORE TRANSFERS BETWEEN… 2021-06-15 16:17:20 RT @kibblesmith: Where is this Pulitzer https://t.co/tRpJV9enhB 2021-06-15 16:16:50 RT @kibblesmith: This very plausible scenario brought to you by my best man recruiting a friend to crash our wedding as Spider-Man and roas… 2021-06-15 16:09:40 Parsing https://t.co/l900OBXR9Z 2021-06-15 16:07:29 Fudge for breakfast though 2021-06-15 16:00:04 This is from the @_KarenHao article yesterday 2021-06-15 15:39:28 This is my personal favorite: https://t.co/LBYUMLA5wu 2021-06-15 15:34:02 Really amazing digital photography projects by @uwcse students. Also, I want waffles now https://t.co/zjclXAIttw 2021-06-15 15:17:40 @HadasKressGazit @DrCABerry Congrats!! 2021-06-15 15:17:10 RT @HadasKressGazit: In exciting news :) 2021-06-15 15:15:24 RT @Reductress: Man Realizes Personal Demons Actually Consequences of His Actions: https://t.co/SGQsMZnV2P https://t.co/GM4kCWO3qb 2021-06-15 15:08:36 Now do the US https://t.co/vZWWajhNUg 2021-06-15 15:07:59 RT @hourlyFox: https://t.co/uBpujjf5Wa 2021-06-15 15:00:51 So many of you and I love it https://t.co/Os00vhdaVj 2021-06-15 14:58:36 @yisongyue Good 2021-06-15 14:48:07 RT @xkcd: Base Rate https://t.co/bVZ0cerahj https://t.co/1Q58nPwhay 2021-06-15 14:46:58 RT @RonWyden: Everyone should be paying attention to the fact that the Western United States is a tinderbox ready to go up in flames. This… 2021-06-15 14:42:01 Working on a postdoc in sleeping and eating vegetables 2021-06-15 13:39:23 RT @PTOOP: @jer_gib @TaliaRinger Indeed! But a key extra @jer_gib ingredient is Day tensor (which is to Applicative what composition is to… 2021-06-15 05:47:12 RT @Sparky15756: @AdhdAngsty @Motscoud_win https://t.co/0GUE1rmDNR 2021-06-15 05:46:57 RT @NekoMasterJuly: @Socialist_Lady @AdhdAngsty most people get a slider i got a button 2021-06-15 05:45:57 RT @AdhdAngsty: This was forwarded onto me so I'm not sure where it's from, but how good is this meme https://t.co/sBF8pGTp8v 2021-06-15 05:35:28 RT @isammioficia1: NALA. https://t.co/VR3D5WnauU 2021-06-15 04:57:44 https://t.co/KhTJdXTJtt https://t.co/hykkbr4WtB 2021-06-15 04:47:30 RT @blaireerskine: ERCOT spokesperson https://t.co/sQzrMUVMWG 2021-06-15 04:38:00 but what was Fermat's penultimate theorem? 2021-06-15 04:32:15 RT @QuinnyPig: After the apology, there are always going to be a few folks for whom no apology will ever be sufficient. After the initial (… 2021-06-15 04:31:27 RT @QuinnyPig: An apology costs you nothing. You haven't "lost." You haven't somehow made yourself less relevant to the things you normally… 2021-06-15 04:31:00 RT @QuinnyPig: I've taken a look into "cancel culture" and the folks who've supposedly been canceled for Bad Tweets. Every one of them when… 2021-06-15 04:28:28 Good advice but also almost impossible to implement for most ADHD people. I've responded before I've even had the thought to monitor my feelings lol https://t.co/pOsvY4QOVS 2021-06-15 04:27:41 RT @QuinnyPig: So You've Been Called Out On Twitter: A ShitPoster's Guide On How To Proceed https://t.co/dVDP4PxCJa 2021-06-15 04:04:56 RT @blaireerskine: Why I love working for the CIA. https://t.co/Frs3OnNgP3 2021-06-15 04:01:46 RT @blaireerskine: anti-vaxx doctor exposes vaccine side effects https://t.co/A0kqjQQv6U 2021-06-15 03:56:03 Oh gosh https://t.co/wxYFCgHWLd 2021-06-15 03:41:22 @evijitghosh Still slowly climbing out but mostly spent a whole day picking random fights about how hopeless the future was until 4AM haha 2021-06-15 03:40:54 @evijitghosh Saw my whole lab for the first time since pre-pandemic yesterday and the depressive crash after was wild 2021-06-15 03:36:31 @hillelogram @chrisamaphone I'm amazed you manage to keep a thing and read the same copy twice 2021-06-15 03:34:59 @chrisamaphone That corner has a couch and a table and I get a hot beverage or some boba and relax and just let myself try to enjoy it and get into the coffeeshop mood, but without the coffeeshop sensory overload and distractions 2021-06-15 03:33:57 @chrisamaphone Oh one I forgot. In my shared office at UW I never read papers in the same place I write code or papers. I have a reading corner, and when I go there, I know it's goddamn reading time and I'm not going to pick up any technology unless I need to Tweet questions to the world 2021-06-15 03:30:21 @hillelogram @chrisamaphone I love how you corrected yourself hahaha 2021-06-15 03:23:24 @chrisamaphone You: how do ADHD people read papers? Every ADHD person: preprocess and then converse 2021-06-15 03:16:45 Walked or ran by this magical place every week for a whole year and didn't know it existed until last week https://t.co/u8C5So92Wf 2021-06-15 03:03:22 @znmeb Yeah it was all symbolic in extremely limiting logical systems IMO 2021-06-15 02:48:12 Cubical lets you reason symbolically and computationally about the real number interval and about higher-dimensional paths so that's pretty freaking cool 2021-06-15 02:46:44 Oh I also think dependent types will just be absolutely necessary to express certain kinds of symbolic knowledge in most important domains, because real life isn't simple enough to just throw away reasoning about terms 2021-06-15 02:43:20 @evijitghosh Depressive crash lol 2021-06-15 02:41:28 So I hope you don't all stick to classical logics just because it's what you know! Constructive logics can carry us to a world of fascinating and wonderful and useful and efficient automation 2021-06-15 02:40:33 With good neurosymbolic techniques I think that structure will get you programs and proofs even in traditional synthesis tasks in way fewer examples than you would otherwise need, with much more certainty 2021-06-15 02:39:15 I like constructive logic because I think it has the most potential for the kinds of automation that I think will matter the most in the future, given enough work on them---reuse and repair 2021-06-15 02:38:17 Missing out on beautiful information just sitting there waiting to be used---existing programs and proofs. And in constructive logics, these examples are just so beautifully fantastically structured 2021-06-15 02:37:16 The problem is that people became so enamored with classical logics that they designed proof and program automation in a particular way, thinking only about the starting point and the goal 2021-06-15 02:36:24 Plus the more axiomatic freedom you have, the more external automation you can build with well-defined behaviors. Logics on top of logics on top of logics 2021-06-15 02:35:39 FOL and classical HOL have a lot of automation built into them already, but I strongly believe this gap is an artifact of history. For everything you lose by omitting classical axioms, you gain in extremely rich structure just waiting to help you out 2021-06-15 02:33:51 I think one thing people get confused about a lot is that they think I'm into DTT and constructive logic and cubical and whatnot because they're principled and so on. But actually my favorite thing about them is how conducive they are to automation https://t.co/lHeWOW3loZ 2021-06-15 02:26:55 @mmitchell_ai @_KarenHao @timnitGebru @willie_agnew Truth 2021-06-15 02:23:21 RT @seL4Foundation: We’re pleased to announce the release of #seL4 12.1.0, and consistent releases of CAmkES 3.10.0 and CapDL 0.2.1. Detail… 2021-06-15 02:22:49 Like guilty, like I'm taking up space I shouldn't take up or something 2021-06-15 02:22:13 Sometimes I just kind of feel bad for existing 2021-06-15 02:20:45 @natefoster Any research funding for this is great though 2021-06-15 02:19:04 @natefoster Nice. I wish more type theory folks participated because I worry we will get stuck on just one logical system with a lot of poor properties for taking advantage of structure and information 2021-06-15 02:16:46 @natefoster How big are those grants? 2021-06-15 02:16:08 @natefoster 2021-06-15 02:12:35 @dlowd Reasonable work in this space is new and relatively unexplored 2021-06-15 02:11:53 @dlowd The neural nets were bad and the logics were the least interesting and useful ones, and the approaches were trivial 2021-06-15 02:05:30 How have I lost both my car key and my credit card over just two days 2021-06-15 01:53:54 @AriKatz20 I'm advising two students on a NS decompiler from proof terms to proof scripts in Coq, kinda excited to see what happens 2021-06-15 01:51:46 Oh and domain experts obviously 2021-06-15 01:50:09 If I were a government agency with 20 million to throw at a problem right now I'd probably pick neurosymbolic reasoning bringing together type theorists, ITP & 2021-06-15 01:46:01 @jost_george It's fantastic it should have happened like ten years ago tbh 2021-06-15 01:42:11 I don't think this problem is fundamental to big tech though. Just the NS work is fairly new and needs to show really good results, it will happen soon 2021-06-15 01:40:52 Symbolic reasoning is very alive in programming languages research though lol 2021-06-15 01:40:19 Yessssssss let's fund neurosymbolic reasoning https://t.co/B1a4i90JMI 2021-06-15 01:23:39 RT @reduct_rs: ‘Tech is A Merictocracy,’ Says Man Who Thinks You Can Write Safe C if You ‘Just Don't Write Bad Code’ 2021-06-15 01:23:20 RT @chrisamaphone: @TaliaRinger lol yes, it's fascinating how many ways people have been citing to "talk back" to the paper as we read. hig… 2021-06-15 01:19:35 @chrisamaphone Except when reviewing, then I really make sure I understand everything. In that case I do two reads, one skimming but not stopping to understand anything, then I wait a day, then I go back and actually read. Some preprocessing helps for some reason 2021-06-15 01:18:22 @chrisamaphone YMMV but I always have to print them out, have to write all over them, and accept that my brain will naturally think of 50 cool ideas but maybe miss out on some details, while other people may catch every detail but not have those ideas 2021-06-15 01:17:04 RT @chrisamaphone: ADHDers, do you have any tips for reading academic papers? A few mentees have asked me this recently, and I shared m… 2021-06-15 01:09:13 What do people think will follow from all of this? Real question https://t.co/sWLtXLyey4 2021-06-15 00:48:58 RT @michael_w_hicks: On PL Perspectives, Kesha Hietala and @certifiablyrand describe how techniques from traditional compiler verification… 2021-06-15 00:41:19 @_dcrc_ Awesome. I'd like a way to stop getting quote Tweeted similarly, those often lead to dogpiling 2021-06-15 00:40:34 RT @_dcrc_: Getting a little TOO much attention? I want to make it easier to address that in 3 ways. 1⃣ Get notified when you’re getting… 2021-06-15 00:39:57 RT @_dcrc_: Unmention yourself I want to make it easier to untag yourself from a Tweet or conversation you don't want to be involved in.… 2021-06-15 00:39:53 RT @_dcrc_: Sometimes you want to talk, and sometimes you just ... don't. Check out these early concepts that could help control unwanted… 2021-06-15 00:08:40 @jost_george @brendanzab @RasmusKallqvist Yeah "PL" usually refers to programming languages research specifically 2021-06-15 00:07:36 @jost_george @brendanzab @RasmusKallqvist Yeah it's programming languages, which these days also includes proof automation and so on 2021-06-14 23:05:47 @sarah_zrf Fair 2021-06-14 23:04:57 @sarah_zrf That seems more valid when it is about the type system and not about a particular construct that can be implemented in many different type systems 2021-06-14 23:04:19 @sarah_zrf I also got annoyed in college having to learn things in 2 or 3 dimensions before infinitely many, though, so I'm weird 2021-06-14 23:03:32 @sarah_zrf Lol. Yeah I just feel like in my mind it's weird to arbitrarily restrict a concept to not include dependent types 2021-06-14 22:58:11 @sarah_zrf That seems silly though, that's a property of the language expressing it, the natural generalization would be dependent 2021-06-14 22:56:26 @sarah_zrf Is there a type equivalence between every ADT and a polynomial like this? 2021-06-14 22:54:18 @sarah_zrf Are there any ADTs that cannot be defined as inductive types? I'd guess not? If you make the sums and products dependent, do you get something as expressive as inductive types & 2021-06-14 22:52:40 @sarah_zrf Literally with sum and product types? 2021-06-14 22:51:19 @sarah_zrf What do you mean by polynomial? 2021-06-14 22:45:34 /s/Codes/Tweets https://t.co/RdWzupbxVC 2021-06-14 22:41:17 @vinegarbin They got mad at me for making the title green instead of black last week, so I had to resubmit and wait haha 2021-06-14 22:34:59 especially as they interact with parallelization 2021-06-14 22:33:47 RT @brendanzab: So… the early work on ADTs (Algebraic Data Types) was originally based on ADTs (Abstract Data Types)! This seems less surpr… 2021-06-14 22:33:38 @natefoster @JAldrichPL Now I'm more confused 2021-06-14 22:32:25 @elazar_g I think they want "it only lies if you don't lie to it yourself, and if you don't do convoluted stuff" 2021-06-14 21:32:32 RT @samth: @TaliaRinger @JAldrichPL A PhD is a qualification to be confused about a vast range of subjects. 2021-06-14 21:26:35 Oh even worse we just overload acronyms because we hate happiness and fun 2021-06-14 21:25:27 @samth @JAldrichPL To be fair I think I was confused about this as recently as 10 minutes ago, despite having a PhD and stuff now 2021-06-14 21:19:14 @brendanzab ugh why do we do this to ourselves 2021-06-14 21:18:20 (@roeschinc sent me this blog post when I asked, but @MarisaVeryMoe said it's probably better to just optimize the recursors directly 2021-06-14 21:17:14 What are some optimizations over recursors/eliminators/induction principles? I've found something on defunctionalization, but I'm curious what optimizations exist without defunctionalization https://t.co/57073wTagW 2021-06-14 21:09:41 guys guys guys I just realized Dex is literally ML for ML the acronym joke we've been waiting for https://t.co/JnT2dsBX84 2021-06-14 19:59:23 RT @SigplanM: We are awaiting the latest registration data for PLDI this week, so haven't sent out invites for mentors and mentees to regis… 2021-06-14 19:14:30 Yessss my thesis was accepted 2021-06-14 19:11:39 Can you really get 99 packets of honey for free? 2021-06-14 19:11:07 Please stop tempting me to fuzz this https://t.co/5eZV0vR2FY 2021-06-14 19:03:46 @porglezomp Recursion in general can be nonterminating, I don't know of any concepts of induction that support arbitrary nontermination 2021-06-14 19:03:09 @jer_gib @PLDI I already registered for PLDI, is this included? 2021-06-14 19:02:21 @porglezomp Induction is equivalent to recursion with a syntactic guard condition, I know that much 2021-06-14 18:54:23 @ericlippert @dwaynereeves Then for your power users, you can build soundness on top of that, by way of a plugin or something 2021-06-14 18:53:43 @ericlippert @dwaynereeves It seems like a fundamental need is the ability to say "I promise," much like Axioms in proof assistants, but with that being socially acceptable and with strong support, maybe even the default 2021-06-14 18:48:01 I mean ADTs in the Haskell sense, the term seems to be broader in PL, making things even more confusing 2021-06-14 18:42:58 I conclude that ADTs are inductive types in other type systems but the communities gave them different names for weird reasons. ADTs in practice come with limited recursors but that also isn't fundamental 2021-06-14 18:41:51 StackOverflow has a bunch of nonsense about inductive types allowing for dependency while ADTs do not, but that's not a property of inductive types, that's a property of the particular languages they exist within 2021-06-14 18:40:46 And inductive types, someone asked me if these had ever been implemented outside of a proof assistant, and how they differ from ADTs fundamentally (not just in terms of implementation, which I know), and I didn't really know the answer 2021-06-14 18:38:26 As an example, it seems like most people who want dependent types for practical programming actually don't care about soundness too much, they just want a helpful and very natural annotation system, basically 2021-06-14 18:37:38 One observation from talking to industrial teams: PL has a lot of cool concepts that get lumped together into the same language or community, and industrial language designers then need to decouple them from features they don't want 2021-06-14 18:36:31 @astahfrom Unacceptable 2021-06-14 16:25:24 @MelissaVasi @OpenAcademics @PhDVoice @hapyresearchers @AcademicChatter I enjoyed mine and got lots of cool questions from friends attending remotely around the world 2021-06-14 16:21:57 @astahfrom You mean "hey HTalulualalTaTaHT!" 2021-06-14 16:18:36 @astahfrom That will be Dr. Professor HTalulualalTaTaHT to you 2021-06-14 16:05:41 @_pdarragh Mostly wherever there is an intersection it's prudent PL people are in the loop 2021-06-14 16:05:10 @_pdarragh It's not the way, but it's a way that will produce convincing results on a surface level that cause hype cycles of "PL is dead ML is everything" thereby running researchers out of funding. ML wasn’t applicable to all of linguistics, but still devastated areas it didn't help 2021-06-14 15:46:12 @_pdarragh Learn useful natural language explanations of code and use that to communicate back useful error messages to users 2021-06-14 15:45:05 @_pdarragh Create new abstractions by tracking changes over time and synthesizing abstractions that would have made the code robust to change, then helping programmers redesign their programs 2021-06-14 15:41:35 What are you doing phone that is not my name https://t.co/JwuyB6d9jK 2021-06-14 15:39:09 @_pdarragh You'd be surprised 2021-06-14 15:33:18 RT @samth: @wenkokke @TaliaRinger @edwinbrady To enumerate the differences a little more: In GHC, you can compute with types (in restricted… 2021-06-14 15:29:53 RT @haskellhutt: @wenkokke @TaliaRinger @samth @edwinbrady I also like Edwin's suggestion. Dependent types = first-class types = types as… 2021-06-14 15:29:26 RT @wenkokke: @TaliaRinger @samth I think @edwinbrady's suggestion of saying first-class types when you mean full dependent types would be… 2021-06-14 15:24:13 @algeom @Dualprince I'd avoid too much responding in a thread on his post like this, can have a dogpiling effect rather quickly 2021-06-14 15:23:00 @tomyudi @C4pn5 Programming languages research 2021-06-14 15:18:29 @Dualprince @algeom I'll expand acronyms I don't expect the target audience to get in the future. But for a lot of my posts, the target audience isn't the general public, it's a subset of people who follow me 2021-06-14 15:15:38 @BorisAlmonacid Programming languages research 2021-06-14 15:14:51 @algeom @Dualprince It's fine not to know what PL is, I just don't know why I don't have to elaborate ML but I do for PL when I'm addressing researchers in my own field 2021-06-14 15:12:37 @Dualprince Programming languages, and I'm addressing programming languages researchers in this thread, not anyone else 2021-06-14 15:02:21 RT @derKha: What to call a tactic that is not quite `crush`? https://t.co/qcGApHtgrt 2021-06-14 15:01:31 RT @sciam: We are entering a new era in sending people beyond Earth. https://t.co/0MgGMglXK9 2021-06-14 15:00:05 RT @RutiRegan: I do not envy American politicians who have to make polite comments about the recent change of government in Israel, but who… 2021-06-14 14:35:32 RT @asayeed: @TaliaRinger @va2lam yeah, that's precisely what has happened, and now that the limitations of ML for human language tech have… 2021-06-14 10:56:21 Spiral of total depression and disillusionment and anxiety until the middle of the night cool cool 2021-06-14 09:45:19 Can I make my whole Twitter account GPL 2021-06-14 09:44:42 So how much of Twitter popularity of academic researchers is just people following us to capitalize off of research communities without contributing back to them or adequately crediting them 2021-06-14 09:30:10 RT @dlowd: Unpopular opinion: the best thing young people can do early in their careers is to know the rules, and so do I — a full commitme… 2021-06-14 09:29:49 @jaycech3n I think the better answer is overhauling our economic system to free ideas can we do that 2021-06-14 09:25:15 @jaycech3n But I forgot we built an economic system that rewards psychopathic behavior and punishes kindness 2021-06-14 09:24:51 @jaycech3n And I believed enough in human nature to think people would just actually share back with communities when they do those kinds of things 2021-06-14 09:24:12 @jaycech3n I think I just never realized how terrible the world is where people just take whole ass ideas people dedicate their entire lives to advancing for societal benefit and turn them into profit without contributing back to the community. That is theft 2021-06-14 09:19:08 I don't care if people profit off of things I do or not but nobody ever should be able to profit off of secrets. Secrets are for keeping people safe 2021-06-14 09:17:46 Why did we build a whole economic system that incentivizes psychopathic behavior I hate it 2021-06-14 09:16:31 You just shouldn't be able to profit off of information hoarding and secrecy, period. Profit by building better products that people want to buy, but don't keep the information secret. Competition and collaboration should work hand in hand and be constant and neverending 2021-06-14 09:14:48 I think I'd rather just all corporate information be public, hoarding information is unethical and acts against rather than for human progress, the economic justification relying on faulty assumptions about human nature 2021-06-14 09:13:28 Starting to understand GPL honestly 2021-06-14 08:30:28 @PTOOP I'll talk to you after tomorrow 2021-06-14 08:12:33 @Natanael_L @UlrichJunker Woo-hoo monopoly 2021-06-14 08:00:05 @MarisaVeryMoe It's X, there's something to worry about 2021-06-14 07:47:26 @Natanael_L @UlrichJunker Why can't they figure out academics aren't looking to get rich if we wanted that we would quit and start companies 2021-06-14 07:44:41 The idea of secret projects in PL when lives don't depend on the secrecy honestly makes me angry 2021-06-14 07:37:49 @UlrichJunker Can they just not 2021-06-14 07:35:38 @alcides I don't think those are quite PL theory 2021-06-14 07:20:30 @PencilearsArt What is machine learning for programming language theory though lol 2021-06-14 07:18:16 Who runs this lab and can I just explain to them how this field works 2021-06-14 07:15:18 There are exactly zero people in the world with experience in machine learning for programming language theory https://t.co/MMEWu4GcMf 2021-06-14 07:10:26 PL/SE is a field that strongly values openness and collaboration, this is a depressing development 2021-06-14 07:04:02 @_joaogui1 I love the idea of ML experience for PL theory, something literally nobody has 2021-06-14 07:03:37 RT @secretasianman: @TaliaRinger they're not saying "10x" ironically. fascinating 2021-06-14 06:59:59 "But we need to start the company first because $$$$" all of us have ideas that would make good companies we just don't want to do that, you won't lose anything by just sharing what you're doing, you will gain a lot though, and damage the field less 2021-06-14 06:55:15 Just talk to experts and exchange information openly and innovate together 2021-06-14 06:54:51 Secret PL projects yikes 2021-06-14 06:51:20 This creeps me out so much in ways I can't even begin to describe https://t.co/vEVgYc8ZDq 2021-06-14 06:31:57 RT @luna_lun_a: RT this out of context 2021-06-14 06:28:44 RT @thegautamkamath: Given the popularity of this unpopular opinion, here's a question for the academic **faculty** out there. Not making a… 2021-06-14 06:20:46 We should disincentivize this kind of behavior because it is net harmful for the field 2021-06-14 06:18:45 It shouldn't be about "winning." It should be about helping each other. Everyone has complementary knowledge, and when we bring it together the field advances dramatically 2021-06-14 06:16:13 This is nice if you want to be a selfish jerk, but the right thing to do for the sake of the field is collaborate https://t.co/PWdf9zRId5 2021-06-14 06:12:51 RT @tiffanycli: Aggressive hustle culture is annoying, but so is aggressive wellness culture. It’s great to tell folks to stop working on t… 2021-06-14 06:04:31 @MonniauxD 2 for sure. 1 not on the project directly, except for the student who knows both areas, but I have gotten lots and lots of useful advice from ML folks 2021-06-14 06:03:09 @MonniauxD Some ML people who care about logic/PL: @yisongyue, @ChrSzegedy. Both willing to collaborate with PL people when interests overlap and the case is convincing 2021-06-14 06:01:03 @MonniauxD I have two projects right now at the intersection: 1. a neurosymbolic decompiler from Gallina to Ltac (PL + ML for PL) 2. (hopefully starting tomorrow) automation for dependently typed learning frameworks (PL for ML) 2021-06-14 05:52:19 @MonniauxD ML and PL are so beautifully complementary that most things that are hard in ML are solvable by good PL techniques with enough innovation, and similarly in the opposite direction 2021-06-14 05:51:38 @MonniauxD Yes. Sometimes you don't want to call it PL, you want to listen to them about what's hard for them, think about techniques you know that could help, and then offer to help them build in those technique 2021-06-14 05:44:25 In summary, ML has more power, so it's our job to show them we're useful to them. And we are, but we need to convince them that's true 2021-06-14 05:38:10 @gmravi2003 No, I'm saying we care about our life's work enough to bitterly relate to the harsh language 2021-06-14 05:36:17 @gmravi2003 But the shortcomings of ML without that domain knowledge often should have been obvious 2021-06-14 05:34:10 @gmravi2003 I don't think it's reasonable to feel bad about how I phrase the way that the most well-funded research area in any field destroys careers of those who do not cooperate, then later reintegrates their knowledge. I'm targeting PL folks and I think the harsh language is effective 2021-06-14 05:29:04 @gmravi2003 I'm being harsh because people have lost their entire careers over this, despite their insights later being useful enough to be reintegrated. The reality is harsh 2021-06-14 05:26:56 @va2lam @asayeed Yeah and it's awful if people lose jobs and funding in the interim, so we need to make the limitations clear immediately by working with them 2021-06-14 05:25:10 @gmravi2003 Sure, in one of two ways: 1) in fields that collaborated, by birthing new areas that bring together knowledge from each gracefully 2) in fields that didn't, by putting huge swaths of human beings out of work, then later reintegrating their ideas 2021-06-14 05:21:10 The only reasonable solution to this is to work with them to build better benchmark suites and evaluation methods 2021-06-14 05:20:25 Oh another thing: left to their own devices, ML folks will design benchmark suites for your domain that don't capture lots of important challenges and tasks in your domain, then show supremacy on those ill-informed benchmark suites 2021-06-14 05:05:27 @yisongyue That's accurate. Complete with appropriation after ravaging the native population 2021-06-14 05:03:12 @PLDI If I initially didn't register for this, then got weirdly excited about some work in this space, can I still register and attend? 2021-06-14 05:02:42 RT @PLDI: Co-located with PLDI: For those who can't decide what "ML" stands for, the fifth Annual Symposium on Machine Programming (MAPS, f… 2021-06-14 05:02:37 @dataghees @awsTO There's also this: https://t.co/NByqwuIoaj 2021-06-14 05:01:24 @dataghees For ML + PL? It's scattered right now, there are some surveys coming out though (not mine), I will post them when they are public. @awsTO just posted a book on verification of neural nets, should be in his Twitter feed, that could be one cool place to start 2021-06-14 04:55:31 There is a lot more but like, these are two beautifully complementary fields. The ML people have more funding and more influence, so we need to convince them that's true, not ignore them and do our own thing 2021-06-14 04:54:08 - PL theory probably has a lot to contribute to foundations of ML 2021-06-14 04:52:34 - PL is great at helping you build correct things and showing they're correct with strong guarantees 2021-06-14 04:50:48 - nobody knows what the hell neural networks are doing, but literally half of PL is about explaining programs 2021-06-14 04:49:57 @yisongyue Haha sorry, though to be fair I'd place you squarely in the small group of people doing exactly the right thing 2021-06-14 04:48:58 - ML programmers are programmers who could benefit from a wide range of elegant PL techniques, including dependent types 2021-06-14 04:48:13 Things PL can help with: - it's often absurd to need more than just a few examples in really structured domains 2021-06-14 04:42:58 @techpractical See thread 2021-06-14 04:40:40 So don't resist. Work with them and build something better than they could ever imagine 2021-06-14 04:40:09 This is why I think if you don't want ML to replace PL, it's actually prudent to collaborate with ML people and build a future of that harmony. ML often takes over fields like a cancer, then later steals their domain knowledge after those researchers have already lost funding 2021-06-14 04:38:42 2. If PL people collaborate with ML people, they will help us with things that are hard for us, and we will help them with things that are hard for them. The two fields will continue to exist and work together in beautiful harmony 2021-06-14 04:37:33 PL funding will suffer, and it will look like ML has replaced PL. 10 years later, ML people will accept the non-obvious limitations, and will start reintegrating traditional PL techniques. But it will look like their own contributions 2021-06-14 04:35:52 1. If PL people resist collaboration with ML people, ML will produce results that look much better than PL results alone, even though complementary, in many areas of PL. These will have limitations that could be addressed by traditional PL, but those will not be obvious 2021-06-14 04:34:50 I don't think ML will replace PL. I think there are two possible futures, and we have control over which one we choose: 2021-06-14 04:21:42 Watching all of the PLSE folks go on to do such amazing things makes me feel so fortunate to have been part of such an amazing lab 2021-06-14 04:00:25 RT @kirby_cho: @hillelogram @slowagon @TaliaRinger @sheilaRoper11 @VP Yes! Only take away voting rights if you are convicted of a voting of… 2021-06-14 03:59:07 RT @hillelogram: @slowagon @TaliaRinger @sheilaRoper11 @VP Minor disagree 2021-06-14 03:58:43 @mukesh_tiwari @tobycmurray Complaining can lead to innovation! Or compassion, or bonding 2021-06-14 03:54:04 https://t.co/iLgJjR70JQ 2021-06-14 03:53:36 I was just looking for some diagrams for tattoos 2021-06-14 03:53:24 Stay classy nlab https://t.co/YSc4FEjgrm 2021-06-14 03:14:56 https://t.co/MVZ1WBcBXq 2021-06-14 03:13:57 https://t.co/kWDHLI8LrD 2021-06-14 03:13:10 https://t.co/F7wJK1n4Uk 2021-06-14 03:12:10 https://t.co/TZvFPagm3x 2021-06-14 03:10:27 Some photos from the PLSE & 2021-06-14 03:07:38 RT @HamasakiLaw: https://t.co/bPhIaeug8y 2021-06-14 02:29:10 @slowagon @sheilaRoper11 @VP Yep and when the justice system is racist, it will disproportionately disenfranchise Black Americans 2021-06-14 02:28:20 RT @slowagon: @sheilaRoper11 @TaliaRinger @VP If you serve your sentence ALL your rights should be returned. Otherwise Felony convictions w… 2021-06-14 02:23:40 But if you all start calling these "tensor types" or "deep types" or some nonsense I'll make fun of you every day forever 2021-06-14 02:21:56 @samth @laughinghan Yeah, then there's separation of concerns: that proof can sometimes be dispatched by a tool, and sometimes by a human 2021-06-14 02:17:17 I think these few months I'm going to spend some time trying to help with automation for industrial adoption of dependent types. I just think the theory and interest are there so might as well try to help with implementation. Worst case I'll discover where the theory is behind 2021-06-14 02:13:16 @norootcause @RaghavMalik15 https://t.co/8Qr5jvbLeq 2021-06-14 02:08:11 Hey @PTOOP do you want to help trick some machine learning people into writing code in a dependently typed language 2021-06-14 01:52:37 @RaghavMalik15 It's tongue-in-cheek, I just mean people are often scared away by undecidable problems, when they can be gracefully approximated and augmented with some human effort very very very nicely 2021-06-14 01:50:05 @laughinghan @samth It seemed to me on first read there is only definitional and not propositional equality, so that might be the biggest barrier right now. I'll see 2021-06-14 01:49:33 @laughinghan @samth https://t.co/PlJz4UuML7 2021-06-14 01:48:15 @laughinghan Yeah, I'm not sure how common the hard cases will be, I'll need to interact with a bunch of ML folks and play with their tensors I guess 2021-06-14 01:45:10 @laughinghan @samth But sometimes that automation can fail, and the art is going to be in gracefully communicating that, and prompting the user for just the info needed for it to succeed. I'm talking to the Dex people tomorrow I'll need to see what's possible / useful for them 2021-06-14 01:42:32 @laughinghan @samth Either the language or some tooling external to the language needs to have a notion of propositional equality / equalities that can be proven, but do not reduce to be equal to one another. That info can come from annotations or from more automation 2021-06-14 01:40:45 @laughinghan @samth Yeah, the fundamental problem of intensional type theory 2021-06-14 01:39:03 Undecidable problems are often not that hard 2021-06-14 01:38:29 And also communicate back errors in a reasonable way, and guide the programmer to either fix something specific or supply just the right missing information 2021-06-14 01:36:07 (This step would be conservative though, sometimes missing things that are obviously, to you, OK, leaving you or another tool with a proof burden about numbers) 2021-06-14 01:28:26 I'll have better thoughts on all of this after tomorrow, but I would be moderately surprised at this point if there is not a way to make static shape error checking almost entirely painless even for existing tensor programs and libraries 2021-06-14 01:25:09 @catnaroek I think this is already handled by the language via some monadic effects system 2021-06-14 01:23:31 (And some classes of those problems are automatable of course) 2021-06-14 01:22:55 Like you should sometimes need to say "hi compiler, that ugly symbolic number you've inferred is in fact equal to this pretty one I assert 2021-06-14 01:21:23 So the only human effort involved should I think reduce to proving properties about numbers, but I'm not positive 2021-06-14 01:19:11 @samth I'm thinking out loud right now, by the way, so I don't think this will be perfectly comprehensible until I implement something 2021-06-14 01:18:15 @samth Those questions might be easier to ask programmers 2021-06-14 01:17:55 @samth You can always infer ugly indices, but the undecidability is in choosing *your own* indices, and knowing for sure they're OK. I think it is reduced to a problem comparing only numbers 2021-06-14 01:14:29 Those boundaries will not always be easy for humans to understand, though, making the errors confusing, so then there may be an extra step of "actually, I wanted these shapes" and some interaction with a tool 2021-06-14 01:13:39 Practically I think this means you can take any arbitrary program written in your favorite tensor framework and transform it to a program that behaves the same way, but has statically checked boundaries on tensor shapes 2021-06-14 01:06:44 Also you probably need funext 2021-06-14 01:06:10 The sizes may be kind of ugly or confusing by default, and the user may want nicer looking ones, at which point annotations would be needed 2021-06-14 01:05:32 I think in theory given an arbitrary program about arrays, there should be a way to automatically transform it into a dependently typed it function about arrays with sizes checked at compile time, with no human effort, though one caveat 2021-06-14 00:45:24 So much automation is possible 2021-06-14 00:44:08 Moreover the length of a list is fully determined by a fold over the list 2021-06-14 00:43:41 When dependently typed arrays are viewed functions from fin sets to T, this much I know: - functions from fin sets to T are equivalent to dependently typed vectors up to funext, and - lists at a given length are equivalent to vectors of that length 2021-06-13 23:10:38 @RezaRob @JeffDean @octoml @roeschinc I don't know the details of their performance claims, and so won't speak for them. I do know they do impressive work that lines to with many of Jeff's listed interests, that is both principled and useful, and has seen a lot of practical use already 2021-06-13 23:04:20 @VP Agreed, which is why we need convicted felons to also have voting rights 2021-06-13 22:49:45 RT @awesomebrandi: A huge part of why girls have gone disproportionately undiagnosed with neurodivergent disability, is because society bas… 2021-06-13 22:45:27 Sub Pop is a local label that Nirvana was under. So our album is definitely grunge 2021-06-13 22:41:20 - boba - bicycle - backpack - SubPop sticker - graffiti - rain - homebrewed ginger beer - Gas Works 2021-06-13 22:40:14 This photo has so much Seattle in it I'm not sure where to start 2021-06-13 22:34:29 PLSE album cover https://t.co/pKei12EAtY 2021-06-13 21:43:04 https://t.co/dcd9CpQUJH 2021-06-13 21:41:31 It's raining so much but the lab hasn't seen each other for so long that everyone is just standing outside in the rain in Gas Works like it's NBD 2021-06-13 21:01:30 Calling me out https://t.co/2tpvJ2PzBa 2021-06-13 19:01:05 PLSE end of year picnic haul https://t.co/VYQIHaL1VL 2021-06-13 18:48:41 https://t.co/mvlZfq9wjq 2021-06-13 18:47:35 https://t.co/eANQdfdE9a 2021-06-13 18:46:45 (The name of a very good boba chain) 2021-06-13 18:45:57 Happy Lemon https://t.co/qX2tdOCNod 2021-06-13 18:37:12 What's the power dynamic on following back cool undergrads at your institution once you're faculty 2021-06-13 18:27:52 RT @BarakRavid: BREAKING: Biden congratulates Naftali Bennett and Yair Lapid and says he expects working with the new Israeli government ht… 2021-06-13 18:27:42 RT @NTarnopolsky: #Breaking: Naftali Bennett's swearing-in https://t.co/e8DwxExZqE 2021-06-13 18:21:10 RT @Istanbultelaviv: Tel Aviv is overflowing with joy. https://t.co/QrNMr7vDG0 2021-06-13 18:18:30 Americans: let's agree to disagree Israelis: let's disagree for the sake of disagreeing 2021-06-13 18:06:11 RT @AP: BREAKING: Israel’s parliament approves new coalition government, ending Prime Minister Benjamin Netanyahu’s historic 12-year rule.… 2021-06-13 18:05:54 RT @axios: BREAKING: Netanyahu is out as Israel's prime minister after new government survives confidence vote https://t.co/uLaeCIPo1A 2021-06-13 18:05:50 RT @Yair_Rosenberg: The vote has passed. Netanyahu has been replaced. Israel has a new government and a new prime minister. 2021-06-13 17:50:51 The research community is this giant social knowledge graph of people who each care about and know different things, and in every potential edge is something new and beautiful 2021-06-13 17:48:54 One of the things I'm most excited about as a professor is building connections for my students, and teaching them how to build their own. No advisor is ever going to be able to answer every question 2021-06-13 17:46:38 RT @yairlapid: https://t.co/fOcUTpRKyF 2021-06-13 17:46:15 RT @AnshelPfeffer: Netanyahu has no good wishes or words of encouragement for the new government of Israel. Instead he ends his last speech… 2021-06-13 17:44:55 RT @AviMayer: Outgoing Speaker of the Knesset Levin has handed over the gavel to incoming Speaker Levy, who will now preside over the vote… 2021-06-13 17:43:43 RT @AviMayer: BREAKING: @MKMickeyLevy has been elected Speaker of the Knesset. 2021-06-13 17:32:15 RT @Iconawrites: Unpopular opinion: the best thing young people can do early in their career is become the nurse of a millionaire mystery n… 2021-06-13 17:25:02 The best thing young people can do early in their careers is whatever they feel like doing https://t.co/qxSUNaqr06 2021-06-13 17:21:02 RT @anybodyclarex: CPR It takes 5 minutes to learn and could save a life. https://t.co/nEGqdFotfP 2021-06-13 16:59:48 RT @NTarnopolsky: Sunset & 2021-06-13 16:52:32 RT @BarakRavid: דיפלומט ישראלי על נאום נתניהו: נראה שבהתקפותיו על הממשל האמריקני נתניהו החליט לשרוף את המועדון ואחריו המבול. נתניהו מוכן לח… 2021-06-13 16:46:52 RT @jacobkornbluh: “He decided to damage the U.S.-Israel relationship for his own personal interests and is trying to leave scorched earth… 2021-06-13 16:38:37 RT @BarakRavid: Netanyahu: The Biden administration already demands Israel to freeze building in the settlements and in Jerusalem. I reject… 2021-06-13 07:05:19 @techreview The "massive amounts of data" thing isn't even fundamental in ML let alone in AI more broadly https://t.co/Jj73Of2uCl 2021-06-13 06:36:34 RT @haltakov: Biased datasets Some papers used a dataset that contained non-COVID images from children and COVID images from adults.… 2021-06-13 06:36:11 RT @haltakov: Can you detect COVID-19 using Machine Learning? You have an X-ray or CT scan and the task is to detect if the patient has… 2021-06-13 06:22:01 RT @AriKatz20: @TaliaRinger That's exactly the approach of https://t.co/Hc4IsYc7yi in #julialang: Equation and mechanistic model discovery… 2021-06-13 06:21:39 @AriKatz20 @ChrisRackauckas Cool! 2021-06-13 06:14:03 RT @schemeprincess: I married this man. 2021-06-13 05:49:57 Best grad school memory https://t.co/Ynfr0iJDSZ 2021-06-13 05:47:49 @ev_maus Ah, yes https://t.co/Cv0xPUGiIN 2021-06-13 05:44:53 @ev_maus October 2015, we put one on the skull that for some reason we also had. But there were photos at some point. I'll see if I can find them 2021-06-13 05:38:02 RT @awsTO: I've written a book: Introduction to Neural Network Verification. You can find a pdf here: https://t.co/QMyFVAGha0 The boo… 2021-06-13 05:33:04 @JeffDean @RezaRob @octoml @roeschinc @gushfsmith Honestly also @MarisaVeryMoe, Steven Lybomirsky, a lot of @ztatlock's students these days, the entire @ai_sampl group 2021-06-13 05:26:26 @JeffDean @RezaRob A fundamental strain that I see right now is how we balance things that we love in PL---understanding and correctness---with the drive to integrate ML into more PL tools. Done right, I hope we can get the best of both worlds, and have them feed into each other, but it's worrying 2021-06-13 05:23:07 @JeffDean @RezaRob @octoml @roeschinc Also @gushfsmith. Both at or just graduated from UW. Just a feeling. (I have a serious instinct to set up connections, hence the whole mentoring program thing) 2021-06-13 05:18:40 @JeffDean @RezaRob You'll probably like @octoml and the work by @roeschinc, if you are not already aware of it 2021-06-13 05:09:00 I bought these for the lab my first year of my PhD 2021-06-13 05:04:25 @jmschreiber91 @KevinKaichuang Cool, thanks for the recommendations 2021-06-13 02:12:13 Findings in the @uwplse lab https://t.co/uES4OHrRCr 2021-06-13 01:36:36 It's so nice to really be able to dedicate time to this program again right now. I've put in like 10 hours the last few days and I hope to put in many, many more this next week. 2021-06-13 01:34:56 I hope to get the updated PLDI registration data on Monday so that I can contact folks I haven't contacted yet with the information to sign up! If you haven't received it in the meantime, feel free to ask me for it directly. 2021-06-13 01:34:03 6 SIGPLAN-M mentor-mentee matches done for the PLDI wave. 4 emailed, 2 I still need to email. Other committee members will take over the next PLDI matches, but in parallel I'll work on outstanding rematches, and try to recruit mentors for some waitlisted mentees. 2021-06-12 23:54:24 @AlbertQJiang Yeah, this thought process all stemmed out of a conversation I had with @Skiminok earlier on "is it with enough creativity possible to automatically split existing neural nets into neural and symbolic parts?" 2021-06-12 23:48:29 @KevinKaichuang Biology is very young 2021-06-12 23:47:56 @KevinKaichuang And for years it sounds like people thought these were not just not beautiful, but actually kind of useless? 2021-06-12 23:47:26 @KevinKaichuang Whatever is going on here is absolutely stunningly beautiful, the process of creating metaprograms and higher-order programs in our bodies over time, even if most are useless. I'd love to understand each one of the useful programs that arise https://t.co/SOG6K1f1yt 2021-06-12 23:45:20 @KevinKaichuang Haha yeah but to like, ML + PL/FM, bio is a stretch! 2021-06-12 23:44:43 @KevinKaichuang As in, I don't know the space of prior work. I just don't think "we haven't found beauty" is a reasonable reason to expect there to be no beauty, especially if you aren't looking for the beauty to fully explain the phenomenon 2021-06-12 23:42:53 @KevinKaichuang Not my area of expertise, unfortunately 2021-06-12 23:41:31 @KevinKaichuang I don't think every part would be beautiful, but I think you'd end up with something 90% beautiful and 10% gross 2021-06-12 23:40:27 @KevinKaichuang Oh gosh I doubt it, math has been around for thousands of years and finds new beauty basically every day 2021-06-12 23:39:59 @KevinKaichuang I hope at some point neural nets are a part of the process, but the learned model is just some intermediate step toward extracting something more beautiful and insightful, even if parts of it are still approximate and fuzzy 2021-06-12 23:38:13 @KevinKaichuang That's fair, but wouldn't it be even nicer if you could in the end learn some really deep, clean, beautiful insights from inside of that mess? For sure not everything can be explained beautifully, but I believe beauty typically does arise from nature if only we can see it 2021-06-12 23:33:08 It is even possible to automate that process. If neural networks take over more of CS, I hope that we learn much better ways of extracting beauty and insights from them 2021-06-12 23:32:09 In Social Processes, similar arguments were made against computer-checked proofs. But these proofs actually are beautiful if you understand the structure, and it is possible to extract insights from them, truly generalize those insights, and communicate them back to people 2021-06-12 22:57:39 Nonetheless a neural net is an efficient way to do this when there isn't a known function. So I want to see neural nets split into two parts, not just translated directly to a rich language with real numbers and infinitely many dimensions. 2021-06-12 22:56:33 Of course, there will be things you cannot specify at all except in terms of all data you have already seen, which would be inefficient in a rich logical system (though I gather it is what neural nets are mostly doing if you look deep into them, which is why they make me sad). 2021-06-12 22:54:20 I actually think the "approximately" part is a limitation of the particular representations, not their symbolic nature. You can represent infinite dimensions and real numbers and extremely complex dependencies in a rich enough logical system. 2021-06-12 22:52:30 I like work that (approximately) "refactors" parts of neural nets into symbolic representations for this reason. 2021-06-12 22:51:22 In math, beauty often corresponds to reusable insights. I think these insights are approximated for neural nets, but I need to be able to pull them out and manipulate them and use them to do new things myself, or else the world has lost its beauty and its truth to some degree. 2021-06-12 22:49:42 If we can later extract the beauty from the neural net or whatever, OK, I can live with that. But my understanding is that what is happening under the hood actually isn't beautiful (the technique may be beautiful, but the end result is not). 2021-06-12 22:46:59 Sometimes training picks up on most of the beauty, but we don't get a chance to see the beauty ourselves, at least not yet. And I know I'm weird, but I really think that beauty is essential, it's a form of truth. 2021-06-12 22:45:39 I like machine learning as a useful set of techniques, but I honestly don't want it to replace too much because part of what we often lose in that process is (in my opinion) beauty. 2021-06-12 22:42:21 @yuvalmarton @CheeToS_ Yes, it is wonderful 2021-06-12 22:39:52 @SimulatedPhD It makes me so angry that we need a prescription to get glasses, when for most people it would make a huge difference to be able to just try them on and pick a pair 2021-06-12 22:38:09 @yuvalmarton @CheeToS_ Yes, I sat outside :) 2021-06-12 22:26:20 @CheeToS_ Salt & 2021-06-12 20:33:16 God bless America 2021-06-12 20:32:23 Part of a healthy breakfast https://t.co/o8G4j4DISt 2021-06-12 19:23:33 RT @dinoman_j: https://t.co/rImXuIDw7v 2021-06-12 19:21:22 RT @dinoman_j: https://t.co/aQvNDGWXmD 2021-06-12 19:18:36 RT @twisteddoodles: Modern ghosts https://t.co/SICkLlREWe 2021-06-12 18:35:41 RT @SNeurotypicals: just in case you came here to find out more about the 11-year solar storm cycle and how ~it's happening~ https://t.co/u… 2021-06-12 18:35:34 RT @SNeurotypicals: Petition to add "space weather" to the invisible list of acceptable small talk topics 2021-06-12 18:35:21 RT @SNeurotypicals: Neurotypical small talk guides: Try discussing the weather! This is a nice neutral subject Me to my NT colleague: *a… 2021-06-12 18:34:49 Sounds familiar https://t.co/YlePjfOKOH 2021-06-12 18:32:16 RT @manisha72617183: A slice of pie costs $2.75 in Jamaica. It's $3.25 in Barbados. And $3.75 in Cuba. These are the pie rates of the C… 2021-06-12 18:31:52 RT @joepolitz: @natefoster @TaliaRinger @UlfarEr Yup the main idea is hash-chaining on (potentially transparent-to-user, potentially encryp… 2021-06-12 17:42:42 RT @natefoster: @TaliaRinger See @UlfarEr et al.’s Macaroons [NDSS ‘14] https://t.co/5aRdxvN4p3 2021-06-12 17:40:56 Doesn't appear to be super gendered, or there are too many confounding variables. (Surely base level of respect is gendered, but the jump without controlling for other things does not appear to be based on my unrepresentative sample of Twitter followers) https://t.co/0oOqT1Swlp 2021-06-12 14:32:05 Are they good? https://t.co/CPM8XcNfGR 2021-06-12 14:30:03 Reading now that 10% of the world's population can't access glasses but need them? Is there a good charity for this 2021-06-12 14:28:04 My family has good vision, the doctor growing up just said my eyes grew wrong during puberty lol 2021-06-12 14:26:26 Glasses and contact lenses are really amazing innovations, honestly. My contact prescription is like -6.00 and without them I think life would be much harder 2021-06-12 14:19:32 RT @blackinxnetwork: Free Online Everyone Welcome Register and see agenda here https://t.co/Zmo5ENPa4h https://t.co/5CN9yspz7l 2021-06-12 14:19:19 RT @blackinxnetwork: We're less than three weeks away from the first ever #BlackInXConference Get into it!! Check us out and learn how… 2021-06-12 14:13:08 RT @jasonintrator: Ever since I published the Economist piece explaining Critical Race Theory, I have received the worst amount of antisemi… 2021-06-12 13:57:19 RT @AnnaLovesVoting: Twitter it's been two years and I have never once cared about the two MIT grads who solved wine. Please leave me alone. 2021-06-12 13:27:55 There is some super cool metaprogramming going on in our bodies over time https://t.co/0O2kyLn25j 2021-06-12 06:07:38 @axelsimon @1HommeAzerty https://t.co/ZF3tcNscm4 2021-06-12 06:07:10 RT @axelsimon: This is too good. 2021-06-12 06:06:12 @catnaroek It's satire 2021-06-12 05:49:57 Lmao https://t.co/AvBUSJCaeZ 2021-06-12 05:46:00 RT @MathAndCobb: NEVER. FORGET. "Turing was prosecuted in 1952 for homosexual acts 2021-06-12 05:17:47 @LeifAndersen Even then, though, how? Like nobody in mainstream politics at the national scale seems to care about this 2021-06-12 05:16:48 We literally paid with our own lives to keep people imprisoned, disenfranchised, and enslaved. Why? What is wrong with us? 2021-06-12 05:15:40 @LeifAndersen Yeah I agree at this point 2021-06-12 05:15:21 We also have the highest incarceration rate. We could have saved so many lives by releasing nonviolent offenders early in the pandemic, even lives of people who aren't incarcerated, but in most counties we chose not to 2021-06-12 05:12:42 I usually like to fight to change systems, but here I don't even know where to start 2021-06-12 05:11:24 Sometimes I just feel sad about this 2021-06-12 05:10:53 In this country, convicted felons can be legally punished by slavery, and then lose the right to vote when released. The death penalty is still legal. All this while our justice system is racist. It's just so regressive 2021-06-12 05:01:35 @skinnylatte @sabcatsilver https://t.co/syaNWCW6yT 2021-06-12 05:01:19 RT @skinnylatte: For Pride Month I want to celebrate how awesome it is to be married to a bisexual person. @sabcatsilver is a wonderful hum… 2021-06-12 04:58:00 RT @edyong209: !!!!!!!! 2021-06-12 04:19:06 The distinction between cookies and tracking cookies is a start, but not sufficiently granular. 2021-06-12 04:15:44 Some uses of cookies are fine and make my web experience better, and some are profoundly disturbing. I'd like to know what the cookie is being used for, where it is flowing, where it will be checked, so that I know whether to accept or reject it 2021-06-12 04:14:29 It strikes me as weird that we just have the choice to accept or reject cookies in our browser, but no granularity about the kind of cookies and how they are used. Cookies should come with granular permissions 2021-06-12 03:14:14 RT @HardSciFiMovies: I’m still waiting for an explanation for this one https://t.co/IWKbxPiFBB 2021-06-12 02:56:07 @HardSciFiMovies @BrianTRice Is Minnesota OK 2021-06-12 02:49:05 RT @ravenscimaven: Bernie Sanders is six feet tall 2021-06-12 02:43:55 RT @TaliaRinger: Poll for CS faculty: I am a man (yes/no) and when I got a faculty position, I noticed a gigantic jump in how much people w… 2021-06-12 02:27:30 RT @JeffDean: Mentor signups are here: https://t.co/Lxi1pnHDSa 2021-06-12 02:27:28 RT @JeffDean: If you have a background in programming languages or compilers, this is a great opportunity to mentor others in the field! ht… 2021-06-12 02:11:47 RT @uwcse: Congrats to the #UWAllen #Classof2021! You persevered in the face of enormous challenges while exemplifying a commitment to exce… 2021-06-12 02:08:38 @PerryWagle Still preferable to being manipulated TBH 2021-06-12 02:00:11 Turning off ad personalization has its costs https://t.co/aK2XCRUCDr 2021-06-12 01:58:59 RT @nywolforg: A reminder from wolves: love is what really matters https://t.co/vJ0564BQhc 2021-06-12 01:45:12 @rob_rix \gamma \vdash ACAB 2021-06-12 01:44:33 I'm so happy I finally had time for this. I know I've been slow to work on things like rematching and the waitlist recently, and to respond to your emails. But I'm getting to it now that I've finally defended https://t.co/XQi2rS2l2F 2021-06-12 01:38:43 RT @SigplanM: As of today, mentor-mentee matching for @PLDI is underway! The first batch of matching emails will go out tomorrow. Looking f… 2021-06-12 00:13:56 Please, please distribute this within your university, your industrial researcher group, anywhere you know PL researchers. We'd love your help! 2021-06-11 23:56:38 "But am I really good enough to be a mentor?" yes 2021-06-11 23:56:25 Use this form to register as a mentor: https://t.co/RRdOQYHvCe 2021-06-11 23:55:56 I am super looking for programming languages researchers to sign up as mentors for SIGPLAN-M. We have a lot of interest from PLDI, and I'm worried we'll run out of slots. Please sign up to mentor! More info here: https://t.co/qX1mS76Iw0 2021-06-11 23:50:00 RT @SigplanM: We are always looking for new mentors! Programming languages researchers of all experience levels (graduate students, industr… 2021-06-11 23:19:57 Sorry to all of my followers who are now aware of every blink 2021-06-11 23:15:38 Eye doctor, every time I go see her: you need to remember to blink 2021-06-11 22:41:52 @palvaro I don't think people against this want racism to go away, I think they want to avoid any facts that cause them any discomfort, or imply anything about their accomplishments not just being purely from merit, but also from racist systems helping them at the cost of others 2021-06-11 22:34:05 @palvaro I'm having trouble understanding how people can so easily just deny history and reality 2021-06-11 22:32:02 Imagine being this afraid of the truth 2021-06-11 22:31:50 Like it's the entire legacy of this country 2021-06-11 22:31:34 "The amendment ... prohibits "the teaching of Critical Race Theory, meaning the theory that racism is not merely the product of prejudice, but that racism is embedded in American society and its legal systems in order to uphold the supremacy of white persons." That's true though 2021-06-11 22:30:03 Imagine being this afraid of an academic theory JFC https://t.co/QI9eIf5Gkd 2021-06-11 22:28:45 @lianghci I've cleared out my stuff haha 2021-06-11 22:08:10 @jjcarett2 Ah it's fine, no worries! 2021-06-11 22:02:53 @jjcarett2 When evaluating faculty job offers, I told all of them I'd need extra admin support for emails and planning because of ADHD, and either got them to agree to that or negotiated extra startup to hire my own part-time admin 2021-06-11 21:55:06 Same goes for DMs, I've mostly been reacting with emoji reactions unless I had strong reason to communicate with words 2021-06-11 21:54:15 I'm going through them, just slowly. My inbox is down to 28 emails now. It's hard for me to read long things these days, so you might be better off giving just enough context to show that you know who I am and what I do, and then ask for a 30 minute meeting and give me some times 2021-06-11 21:53:02 If you've emailed me in the last n days for some large n, sorry that I probably haven't gotten to it. I currently have the email load of junior faculty, but without the admin support I'm going to have when I actually start my job 2021-06-11 21:46:13 RT @reduct_rs: REPORT: It’s Not Your Team Lead and Product Manager's Fault, It’s Society’s Fault, but You Can’t Fight With Society in The B… 2021-06-11 21:45:31 RT @reduct_rs: 3 Signs From Childhood That You Were Trans If You Need More Validation Than Just Writing Rust Now 2021-06-11 21:14:30 @samth What about community members? Computer scientists from other universities? 2021-06-11 21:04:47 Year, actually 2021-06-11 21:03:15 I was alone in the office all summer and effectively annexed the communal areas by placing facts on the tables https://t.co/S0mnsufa9l 2021-06-11 20:55:51 Haircut https://t.co/7EmzuX0nDC 2021-06-11 20:34:31 @yoavgo @jekbradbury Thank you! 2021-06-11 20:29:38 @apaszke @jekbradbury Oh, that's awesome! I was hoping for something like this 2021-06-11 20:09:48 @nicoleradziwill True 2021-06-11 20:09:19 @GlennF Fully vaccinated in Seattle, first dose in SF 2021-06-11 20:07:57 Feeling strangely sad, but without an easy reason to point to today 2021-06-11 19:54:41 @jost_george @seL4Foundation Glad! 2021-06-11 19:02:13 Same https://t.co/oV7vuEjOfN 2021-06-11 19:01:17 @jost_george @seL4Foundation Also the CompCert verified C compiler. My website has a survey paper of Proof Engineering, I recommend checking it out 2021-06-11 19:00:23 @jost_george @seL4Foundation is the classic one. Verified operating system microkernel 2021-06-11 18:48:08 @jost_george I'm talking proving crticial properties of large software systems 2021-06-11 18:43:21 @jost_george It refers specifically to the discipline of verification using proof assistants at scale 2021-06-11 18:18:04 @jad_kabbara They're trolling but @mathshistory was not when they did the same thing 2021-06-11 18:13:03 I've also updated my CV: https://t.co/D7pRFjMtQP 2021-06-11 18:11:29 I updated my website to reflect the whole incoming professor thing, and also to talk about some cool work I'm doing this summer. Check it out: https://t.co/kuzjIPXr1M 2021-06-11 17:51:03 I'll make follow-up questions based on some of the anecdotes if they point out interesting things I hadn't considered 2021-06-11 17:47:43 As always, anecdotes as supplementary data are welcome 2021-06-11 17:44:23 I don't have a good hypothesis, just want to see what people have experienced 2021-06-11 17:43:50 I'm curious because I've always felt like I've had to fight for respect, whereas now it is the default even when I'm clueless. I don't know how gendered this is---is what I am experiencing comparable to what men in CS experienced as PhD students? 2021-06-11 17:41:12 @hipsterelectron It is annoying when people try to assert they know more than me about my area of expertise, but when other people actually do know a lot more than me, I want to hear lol 2021-06-11 17:40:19 @hipsterelectron Me: neurosymbolic is the future of both ML and PL!! Everyone: neurosymbolic is the future! Me: wait no I need ML people to argue with me I'm not an ML person I just started reading ML papers 6 months ago 2021-06-11 17:37:42 Poll for CS faculty: I am a man (yes/no) and when I got a faculty position, I noticed a gigantic jump in how much people who knew my position respected me by default (yes/no) 2021-06-11 17:32:12 @mariaruizv The first 2021-06-11 17:29:45 I've realized now I have to start qualifying a lot of my opinions with "in my opinion" or "I think" so people who respect me a lot don't interpret them as objective truth. That's cool I'll do that more 2021-06-11 17:28:09 When I got the job at Illinois people started taking my impulsive opinions very seriously, so it's perhaps important to remind people that my expertise is in verification, dependent types, proof automation, proof engineering, and so on, but I'm still quite new to some fields 2021-06-11 17:20:28 Lol this account https://t.co/Vg01YmsO2b 2021-06-11 17:17:21 @lindsey @wilton_quinn Turns out extra admin support isn't considered a "reasonable accommodation" though, even though equally labor-intensive accommodations for other disabilities are considered "reasonable" 2021-06-11 17:16:24 @lindsey @wilton_quinn ADHD accommodations for adults really need to get up to speed with our understanding of what ADHD actually is. Every place I had a second visit, I made it clear my admin needs would be high and asked for extra startup to hire admin if that wasn’t included 2021-06-11 17:03:16 RT @AmyChanta: For those asking, you can start the work by engaging with experts in the DEI space, AND compensating them for their time. E… 2021-06-11 17:03:02 RT @AmyChanta: And before anyone wants to jump in here and wave their "diversity hiring" metrics at me, please also be sure to include your… 2021-06-11 17:02:49 RT @AmyChanta: If I can spend 5 minutes researching your org and turn up half a dozen yellow flags, I'm not going to feel safe enough to ap… 2021-06-11 17:02:41 RT @AmyChanta: If you're a white leader at a company, lamenting about why -- despite your various "cool place to work" awards (which are us… 2021-06-11 17:02:14 RT @AmyChanta: When I find that diversity is severely lacking in upper management, that puts me on alert. That means my next stop is Glassd… 2021-06-11 17:01:51 RT @AmyChanta: For anyone following along, I've decided to document parts of my job search. Why? Because even early in my process I've seen… 2021-06-11 17:01:41 RT @KimCrayton1: Because knowledge is power and ignorance is no longer an excuse for causing harm AND increasingly, potentials employees EX… 2021-06-11 16:48:06 RT @carterforva: Every day. https://t.co/KEAnNB9Gy1 2021-06-11 16:40:07 @samth I'm OK with these lol 2021-06-11 16:39:23 The ADHD tax of ordering lunch at work because even though you went through all that effort to buy groceries and cook for once, you accidentally left your lunch at home https://t.co/Y6s84LWnrj 2021-06-11 16:36:26 @samth Where should we? 2021-06-11 16:33:16 It gives a value to how hard these tasks are quite well that I have at times forfeited hundreds of dollars rather than find receipts to send 2021-06-11 16:28:52 RT @erupton: @TaliaRinger I have literally *never* managed to get reimbursed for conference travel that is supposedly part of my compensati… 2021-06-11 16:25:41 RT @wynneliz1: @TaliaRinger This feels related to the tax of paying for clothes you don't like/don't fit because you can't return them in t… 2021-06-11 16:23:28 The ADHD tax of paying out of pocket because it is prohibitively difficult to submit reimbursements https://t.co/FNdUZXYVCd 2021-06-11 16:21:32 @samth Fair, I just think it's better for both communities in the long run to be explicit about the connection 2021-06-11 16:18:43 @samth "Typed Racket has a restricted class of dependent types called refinement types"? 2021-06-11 16:15:55 RT @samth: @jjcarett2 @alpha_convert @TaliaRinger One other more social point: being more rather than less inclusive about who is part of y… 2021-06-11 16:14:59 @samth I think that's a confusion about what dependent types are. We need to say "arbitrary dependent types" when that's what we mean IMO 2021-06-11 16:13:24 @RanjitJhala I'll argue against anyone in my community using "dependent type" only for Pi and nothing else lol 2021-06-11 16:09:36 @_joaogui1 Sure, I'm just less familiar with how those work 2021-06-11 16:07:55 @_joaogui1 Thanks! 2021-06-11 16:03:57 RT @RanjitJhala: @jjcarett2 @alpha_convert @samth @TaliaRinger Boolean refinements are just a convenience for a common case: they coexist q… 2021-06-11 16:02:10 @breandan @ChrSzegedy @geoffreyirving "And have implemented this for a similar problem for another language" 2021-06-11 16:01:42 @breandan @ChrSzegedy @geoffreyirving Oh, the problems I solve are always undecidable. What I mean by "easy" is "I know how to ask the user just the right questions to automate all of the annoying parts" 2021-06-11 15:59:28 RT @m_misamore: @TaliaRinger . We need to be able to prove nontrivial invariants without going nuts trying to convince the type checker of… 2021-06-11 15:57:42 RT @jonmsterling: @jjcarett2 @alpha_convert @samth @TaliaRinger But i admit that all(?) present formulations are. but to me, refinement is… 2021-06-11 08:34:31 @jekbradbury I'm mostly curious about runtime errors people encounter during development with tensor lengths and sizes, so if you have examples of where that has happened in the past that's even more interesting 2021-06-11 08:25:43 @jekbradbury Cool, thanks! 2021-06-11 08:25:11 @YieldsF @Azumanga Maybe that's why I have no issue with compilation times 2021-06-11 08:17:34 Can people just send me a number of examples of their F/OSS projects using TensorFlow and PyTorch? I want a good sense of how people program right now before I become complicit in trying to make that programming any easier 2021-06-11 07:46:39 RT @BowToChris: @TaliaRinger I actually can't remember the last time I wrote something that compiled and was wrong. Thanks, @CoqLang 2021-06-11 07:46:15 @Azumanga Ah I guess I rarely compile when coding since I need very little feedback 2021-06-11 07:43:55 This just in @RanjitJhala you're one of us now sorry 2021-06-11 07:41:23 If we can stop pretending different implementations of sigma make it somehow dependent or not then we get all the automation from both communities 2021-06-11 07:39:47 Like sure, sometimes the predicate that depends on the term isn't technically a type in itself, but if the whole expression is a type, then I say it's still dependent, you've just restricted the scope of where you can type predicates 2021-06-11 07:37:50 I don't think there's a reasonable argument against refinement types being a kind of dependent type. Arguments against this usually just come down to implementation details and people not wanting to associate themselves with us nerdy DTT people 2021-06-11 07:28:34 @Azumanga I guess if you trust tests as much as static checks, that view makes sense, but the guarantees you get from types are stronger 2021-06-11 07:27:31 @Azumanga I don't think this is fundamental, just not enough compiler optimizations for dependent types. I'm also surprised, tough, since I've never run into compile times being an issue in any language 2021-06-11 05:40:53 @ChrSzegedy Says the guy looking to solve artificial general intelligence lol 2021-06-11 05:39:53 @natolambert Thank you 2021-06-11 05:24:19 RT @whatcatsdid: https://t.co/5klF946S6F 2021-06-11 05:21:47 One of my biggest fears about getting a faculty job was that I'd feel like I'd "made it" and stop having exciting goals. Hahahaha what a fear this is so false 2021-06-11 05:12:03 What are the most common bugs you run into in your day-to-day programming life that you wish the compiler could just rule out for you? 2021-06-11 05:03:43 @mHaGqnOACyFm0h5 That too 2021-06-11 05:01:59 @uwcse @uwplse @amyjko @UW_iSchool @uwdata Awesome 2021-06-11 05:01:38 @uwcse @uwdgh Congrats!!! 2021-06-11 04:59:05 RT @uwcse: #UWAllen #ICTD authors Naveena Karusala, David Seeh & 2021-06-11 04:58:50 RT @uwcse: #UWAllen @uwplse’s Chenglong Wang, Rastislav Bodik & 2021-06-11 04:39:06 @mmitchell_ai Yikes 2021-06-11 04:36:02 RT @jhemann: How am I just now seeing this? 2021-06-11 04:27:41 RT @BjornerNikolaj: @lorisdanto @EscardoMartin @jonmsterling Oh, I thought the trailer promised that Zach was going to run to my place with… 2021-06-11 04:26:11 @jonmsterling Some people enjoy the last minute adrenaline fueled push though. Zach does IIRC, and typically works with students who feel the same way, so it works out 2021-06-11 04:25:00 @jonmsterling I think it's funny, even as someone who freezes major changes to my paper several days out 2021-06-11 04:06:24 @DaTwinkDaddy Not familiar with ATS though 2021-06-11 04:05:55 @DaTwinkDaddy Dependent types require so much manual effort without good automation, but nobody I know of who is interested in automating DTT right now is a systems person 2021-06-11 04:02:56 I really just need some time to not teach, and I'm thankful for that. I'm very happy using this time to explore 2021-06-11 04:01:32 I think I managed to take a real break before my PhD because I hadn't learned so many interesting things. But honestly all I want to do right now is explore all of the amazing interesting things I've had no time to entertain this year 2021-06-11 03:57:14 @DaTwinkDaddy You need to replace a low level systems language with a better low level systems language, not a high level functional language 2021-06-11 03:56:31 @DaTwinkDaddy Rust has gotten really popular for exactly this kind of reason iirc 2021-06-11 03:54:39 How does @ChrSzegedy not have a Wikipedia article? Is anyone bored enough to start one? He's clearly notable lol https://t.co/9hX1JpCtaY 2021-06-11 03:46:51 @frankwyw7 It sounds fun 2021-06-11 03:46:34 And maybe get a Wikipedia article written about me one day. Maybe 2021-06-11 03:46:08 Preferably in a nontrivial way though 2021-06-11 03:45:34 I want to change the world 2021-06-11 03:43:38 Missing an "of" but you get the point 2021-06-11 03:41:01 @samth @vinegarbin @betainverse I uploaded a PDF of my dissertation to a web form on an airplane today, that was all I needed to do 2021-06-11 03:40:27 @samth @vinegarbin @betainverse Looool 2021-06-11 03:39:31 They just usually don't know what dependent types are 2021-06-11 03:39:16 People want dependent types with good UX and good automation 2021-06-11 03:39:00 Instead of "do you want dependent types" try asking "do you want the compiler to rule out this entire common class bugs" it works better and the answer is often yes 2021-06-11 03:36:18 @samth @vinegarbin What 2021-06-11 03:35:50 @woodwardjd So much work is open source you probably can contribute 2021-06-11 03:30:10 RT @vinegarbin: @TaliaRinger Still had to turn in paper. The grad school approved dissertations by measuring the margins on one page and th… 2021-06-11 03:30:07 @vinegarbin I wouldn't even be able to tell if you were lying at this point 2021-06-11 03:28:37 RT @fhuszar: Not sure who made this (via @getjonwithit) https://t.co/cotUL2YZIx 2021-06-11 03:15:43 RT @ThePlanetaryGuy: EVERYBODY!! The Chinese rover #Zhurong carried a small wireless camera that it placed on the ground TO TAKE A GROUP P… 2021-06-11 03:08:04 @mebassett I think they will help you out by asking you for minimal pieces of information, and figuring out all of the painful things automatically 2021-06-11 03:06:33 @Impredicativity https://t.co/VB74wHm1g3 2021-06-11 03:06:12 @Impredicativity Wow you made it on to Mathematical Mathematics Memes 2021-06-11 03:05:56 RT @Impredicativity: Let G be a group, but let it just be. Don't do anything about it. 2021-06-11 02:52:32 RT @samth: @lindsey @TaliaRinger No I think I'm going be complaining at the PLDI business meeting in 2040 that they should join PACMNeuroPL. 2021-06-11 02:49:46 RT @lindsey: @TaliaRinger reformat their POPL rejects in two-column format for PLDI 2021-06-11 02:49:44 @lindsey Hahahahahaha 2021-06-11 02:46:09 "We truly wish that we could celebrate in person" I mean you legally could though it's just the university refusing to adapt to the latest vaccination data and being extremely conservative in all respects 2021-06-11 02:42:01 Literally every time PL research is successful. It's our job not to be noticed https://t.co/GgtazJIwhn 2021-06-11 02:41:00 Ah yes, my sympathy scarf has arrived https://t.co/bcKhSDS1Uo 2021-06-11 02:26:43 @michaelasmith I think I thought that often when I was an engineer, but I didn't know what dependent types were, so mostly I found myself frustrated by things I couldn't express but didn't have the language to ask for 2021-06-11 02:23:41 @soldni It's cool, honestly I just want people using these, I'm happy to build automation so smooth people don't know they're using dependent types. But I will so, so make fun of people when they associate dependent types with machine learning in the future 2021-06-11 02:19:34 Y'all are going to forget their roots in PL and view them as some useful machine learning framework construct and I'll make fun of you forever 2021-06-11 02:17:24 Thank god though I will no longer have to hear "but nobody ever uses those" at some point in the next 5 years or so 2021-06-11 02:16:09 It turns out dependent types are REALLY USEFUL and programmers want them, we just need to make them easier to use, and we legit have the technology for that these days, the usability problems are not fundamental 2021-06-11 02:15:03 Honestly since my Python rant two days ago I've heard from two different teams developing two different dependently typed machine learning frameworks and who knew the industrial dependent type revolution I've been waiting for would start this way I'm shook 2021-06-11 02:08:07 @robs_cse Probably we will all be dying from climate change 2021-06-11 02:07:08 @woodwardjd The technology is being developed very quickly, but we need to prioritize HCI for synthesis and proof automation much more than we currently do 2021-06-11 01:26:38 Not tactics though that's not the right interface for that workflow 2021-06-11 01:25:38 I think programming will still exist, just it will be much more of an interactive process with automation, and a good user experience for that, and you won't need to write boilerplate or stress about details too much 2021-06-11 01:24:01 Finding code libraries and theorems ourselves. Porting ideas from one language to another. Maintaining legacy code Yeah I'm an optimist lol 2021-06-11 01:22:14 Bookmarking this Tweet in case Twitter still exists in 20 years lol 2021-06-11 01:21:51 I'm guessing a lot of the mundane tasks of programming and verification will be automated, but I would guess that 2021-06-11 01:17:58 What do you think people will look back on 20 years from now as some weirdly inefficient thing that current Ph.D. students had to do? 2021-06-11 01:16:38 @codydroux Ah, that's why editors are a thing! 2021-06-11 01:13:33 It's actually hilarious that search is coming full circle to conversation/Q& 2021-06-11 01:09:22 OK Jeeves, how do I get split column alignment for my bibliography? 2021-06-11 01:01:38 @RrrichardZach There's a book? 2021-06-11 00:50:45 @RaghavMalik15 I'd start at the relevant conferences/journals 2021-06-11 00:48:32 TBH even more than searching for articles, how did you figure out things like LaTeX formatting 2021-06-11 00:47:22 @RaghavMalik15 I assume citation chasing, starting from recent work. But this means reinforcing earlier omissions unfortunately 2021-06-11 00:46:30 RT @IgorBrigadir: Imagine requesting papers to read like this: sending an index card with a hand written note by post to a journal or autho… 2021-06-11 00:39:08 @InfraredEyes I don't even know what this means 2021-06-11 00:34:40 @palvaro Also did people have to write citation info themselves instead of copying and pasting preformatted bibtex info? 2021-06-11 00:31:32 I can't believe people did Ph.D.s before good search engines were a thing 2021-06-10 22:44:31 And here you can see the additions over time. The first peak is from copying and pasting in a bunch of papers and style files and so on :) https://t.co/QtSGKzIuun 2021-06-10 22:43:54 Here's a graph of my contributions over time, for example! https://t.co/hfV5PWePGn 2021-06-10 22:27:15 I didn't squash the Git history. I have strong feelings about the lack of availability of granular commit data, and how developers always rewrite history. I want this to be available in all of its glory :) 2021-06-10 22:26:03 Finally, I just made the repository for my thesis public, so that you can go back and look at the process of writing it. I hope this will help more junior graduate students understand what writing a thesis is like: https://t.co/XGdHkszPuU 2021-06-10 22:24:01 I've ordered some hard copies for me and my parents, but also a few extras. If you'd really love a fancy hard copy, let me know, I'd be happy to give away a couple of them to people who just generally like my work. 2021-06-10 22:23:24 As a reminder, the defense is on YouTube: https://t.co/TscPglVbmy 2021-06-10 22:22:56 Thank you so much to everyone who gave me feedback on my thesis! I've submitted the final version, which you can find on my website: https://t.co/XlBfGl6Kfz 2021-06-10 20:13:50 @elfprince13 Please send 2021-06-10 20:13:40 I really like Romeo of Darkness 2021-06-10 20:11:08 This but GPT-3? https://t.co/M5rVBIWF3c 2021-06-10 20:02:47 RT @hourlyFox: https://t.co/AepPDMaO7V 2021-06-10 20:02:06 RT @darth: can not get much more cyber secure than that tbh 2021-06-10 19:43:03 @ChrSzegedy Cool!! Thanks! 2021-06-10 19:37:50 - few example learning 2021-06-10 19:34:06 - biologically inspired computing 2021-06-10 19:31:44 - the hottest trends in cryptography - the hottest trends in topology & 2021-06-10 19:30:11 - symbolic methods for proof automation in higher-order logics (HOL, CIC, etc) that do not involve flattening to FOL - vector encodings of proofs in higher-order logics - metalearning & 2021-06-10 19:27:55 - quantum computing, verification, and synthesis 2021-06-10 19:27:01 - automating type theory - user interfaces for dependent types 2021-06-10 19:25:51 - benchmarks for program and proof synthesis and repair - usability of proof assistants 2021-06-10 19:22:58 Modal logic is always cool 2021-06-10 19:22:19 Some fun, very out there things: - anything that involves finding higher dimensional paths - categorical foundations of ML 2021-06-10 19:21:25 Generally, anything that involves automation to help humans write, modify, or understand programs or proofs 2021-06-10 19:19:58 Send me cool papers. Some current interests (though I'm open minded): - neurosymbolic learning - proof automation & - verification of NNs - cubical type theory - rewrite systems - differentiable programming - what's hot in dependent type theory https://t.co/uGnHVKTQwE 2021-06-10 19:16:33 @marileezafari @jessicamalonso @ginfueledbrat @shantilly_t Proud to be a source of chaos 2021-06-10 19:10:35 Eventual consistency 2021-06-10 19:10:24 Currently giving feedback on a draft of a neurosymbolic programming survey paper I promised I'd read months ago lol 2021-06-10 19:07:14 Weirdly unexpected, but right now I'm really loving finally having the chance to read all of the papers I hadn’t had a chance to read these past 6 months 2021-06-10 18:05:37 RT @math3ma: Part 2 of the “warm up” to enriched category theory is now up (!), where preordered sets and generalized metric spaces are sim… 2021-06-10 18:05:02 RT @thejeffreymarsh: Dear nonbinary teen I hope you know I respect you so much People want to dismiss us both by calling nonbinary a “yout… 2021-06-10 18:01:39 @mmitchell_ai (You're right though, I've never had anything good from Dunkin' Donuts) 2021-06-10 18:00:11 @mmitchell_ai Dunkin' on Dunkin' 2021-06-10 17:34:41 RT @UWVirology: Did you get a postcard or letter from @UWVirology about the WAVES Survey? Please respond! It will help our scientists and @… 2021-06-10 17:19:55 RT @NoContextBrits: https://t.co/4Uxl7xSsZ7 2021-06-10 17:18:08 RT @Maryamization: This is such a good question! So i'm gonna ask it for Atlanta (and in general The South™). What should I do in the next… 2021-06-10 16:29:52 RT @CliffordSosis: via trashcanpaul https://t.co/0mT8Z5so5I 2021-06-10 16:26:26 @johnregehr @krismicinski @thisisnotjuice @antumbral @krismicinski new life partner? 2021-06-10 16:23:53 @mmitchell_ai Hah thankfully I'm good at making bagels 2021-06-10 15:58:35 @johnregehr Have done that one thankfully :). Hoh Rainforest is probably my favorite forest so far 2021-06-10 15:52:14 I've never been to Mt. Rainier, if anyone wants to go hike there 2021-06-10 15:48:40 If you used to live in WA, but moved elsewhere: What are the things you miss most about WA? What should I do in the next month or two before I leave? 2021-06-10 15:34:57 RT @NoContextBrits: https://t.co/qTOy7rZezP 2021-06-10 15:29:59 RT @NoContextBrits: https://t.co/UCcvmRZiCn 2021-06-10 15:25:19 If you're looking to start a long-term cross-institutional mentoring program like @SigplanM in your research area, I'm happy to talk sometime about what you'd need to to do to set it up 2021-06-10 15:13:21 @neurocy @jeremysiek @k4rtik @certifiablyrand @purdue_pl @samth @lorisdanto @ravi_chugh It's good for curating though 2021-06-10 15:01:57 RT @inihelene: I am bisexual 2021-06-10 14:57:55 RT @NASASun: Thanks to everyone who watched with us online! If you missed the stream, you can see a few photos from our @NASA photographers… 2021-06-10 14:54:50 Strong bi constructivist vibes https://t.co/KH9NXEd6Fb 2021-06-10 14:53:59 @ferallove @inihelene Same, I'd be a lesbian if I could, no fun subjecting myself to awful gender dynamics dating men. Dating women is so much better. Though there are good men out there of course 2021-06-10 07:03:21 "them" here was ambiguous but meant to refer to "outpourings of gratitude" 2021-06-10 06:29:54 @KevinJLeach: Vanderbilt 2021-06-10 06:28:42 @joannacss: Notre Dame 2021-06-10 06:27:33 Michael D. Adams: Yale-NUS 2021-06-10 06:26:42 @PreethaChatterj: Drexel 2021-06-10 06:25:43 Fraser Brown: CMU 2021-06-10 06:06:31 RT @lorisdanto: Making this was a lot of fun! Thanks to the amazing cast (@alexandra8silva @ztatlock @rg9119) and co-creators (@polikarn @d… 2021-06-10 06:06:24 RT @poplconf: It is time to move past conference anthems. Watch the POPL2022 Movie Trailer here! 2021-06-10 06:05:09 @poplconf Is Zach OK 2021-06-10 06:04:51 RT @poplconf: https://t.co/TDiItsE7Ed 2021-06-10 06:04:38 Two of my favorite people @ztatlock and @alexandra8silva star in this amazing trailer. Featuring @ChandrakanaNaN's work https://t.co/G6eFXiPsz9 2021-06-10 05:53:57 @jaredweakly @MuzafferKal_ Truth 2021-06-10 05:49:40 kind mediocrity > 2021-06-10 05:45:09 Likewise, the actions you take that scare people who are not harmful away from the field are likely more damaging than the benefit of any work you could do yourself. I don't care if you have a Turing Award or whatever 2021-06-10 05:43:32 We need to change incentives to account for the fact that work that convinces even one person more productive than you stay in the field is as valuable to the field as your entire life's technical work https://t.co/ac5JhBsEjT 2021-06-10 05:10:26 My stomach hurts 2021-06-10 05:10:12 I don't like frog legs. I learned this by eating an entire container of fried frog legs 2021-06-10 05:06:08 RT @MSFTResearch: Congratulations to the 2021 Microsoft Research Faculty Fellows! This fellowship recognizes innovative, promising new facu… 2021-06-10 04:49:37 @mariyaivasileva Cute cute cute 2021-06-10 04:41:14 RT @mariyaivasileva: @TaliaRinger https://t.co/z5u81e9N8q 2021-06-10 04:41:11 @mariyaivasileva I want to know more about this adorable cutie 2021-06-10 04:40:47 @mariyaivasileva OMG 2021-06-10 04:29:53 RT @bisexualposibot: bisexuals are brilliant 2021-06-10 04:08:25 https://t.co/3UbIliMpSW 2021-06-10 04:07:50 https://t.co/WTpMD5KFkT 2021-06-10 04:07:38 https://t.co/JftD9aVh5Q 2021-06-10 04:06:52 https://t.co/iKklnEBCoi 2021-06-10 04:06:07 https://t.co/O50PPM1EH6 2021-06-05 00:23:54 לחיים https://t.co/kAAY4dpczE 2021-06-04 23:46:07 @dj_hci I think the general principle would apply, but the techniques would be different because of how natural language is structured. I will think about it, though! Especially if there isn't lit in that area yet. It took me a few hours to repair the transcript by hand 2021-06-04 23:44:34 RT @jwz: Caring For Your Geriatric Millennial. https://t.co/5jacTUqYMx 2021-06-04 23:31:36 @chowell2000 https://t.co/vj0TxNP1m8 2021-06-04 23:31:22 @MonniauxD https://t.co/vj0TxNP1m8 2021-06-04 23:31:11 Video: https://t.co/vj0TxNP1m8 2021-06-04 23:30:44 @johnregehr Thanks!!! 2021-06-04 23:12:25 @dj_hci you're the only one who works on this stuff who I know personally, so sorry if this means tagging you a lot for this kind of thing, but it's kind of a synthesis of my style of repair work with captioning 2021-06-04 23:11:02 Textual replacement isn't enough here, since how one word is interpreted impacts how the rest of the sentence is interpreted. But because of that, it should be possible for a tool to recalculate much better transcripts based on just a few early corrections that I give it, no? 2021-06-04 23:09:07 It would be nice to have some automation for repairing automatically generated transcripts for videos. Like when I correct the first occurrence of "fruit" to "proof" and "lemon" to "lemma," it would be really nice for automation to infer how to fix the rest of the transcript. 2021-06-04 22:53:10 I recorded the whole thing by pressing a button, then uploaded it to YouTube. After tweaking the generated transcript from another tool, I attached it and just asked YouTube to figure out how to sync it for me. I'll see how it does. 2021-06-04 22:52:13 Honestly, technology is cool. At the tail end of a pandemic, I just defended my thesis remotely to an audience that included friends and family from around the world. I automatically generated captions using technology that one of my friends who was watching worked on! 2021-06-04 22:38:47 I added them but they're processing, DM me for transcript in the meantime 2021-06-04 20:49:03 I'm currently working on fixing up the automatically generated captions from Zoom so that I can add those for accessibility. I expect to have those in about one Boba's worth of time 2021-06-04 20:48:40 Hi friends! Here's a video of my PhD thesis defense. Enjoy! https://t.co/RNl9zUpnke 2021-06-04 20:21:34 @adamberaha I'll honestly be celebrating over and over again for like two weeks with different people because the pandemic makes large gatherings too hard . Thank you! 2021-06-04 20:19:30 @JeffDean @djg98115 We have an entire animated emoji reaction in our lab Slack channel for whenever Dan makes a dad joke 2021-06-04 20:17:05 @wirgien @JeffDean *fête 2021-06-04 20:16:25 @hillelogram I aspire to this level of cool 2021-06-04 20:14:56 @JeffDean Clever. You give @djg98115 a run for his money 2021-06-04 20:08:26 What kind of boba should I get to celebrate right now 2021-06-04 19:55:59 @wrengr That sounds awful 2021-06-04 19:25:47 Sometimes the right thing to do when someone tries to quit is to talk them into waiting and reevaluating in a better mental state 2021-06-04 19:24:08 I really almost dropped out of grad school my fourth year. I tried to drop out over and over again and was very lucky to have an advisor who recognized it as a trauma response (even if he never used those words). It's really cool to finish and become a professor 2021-06-04 19:20:01 @samkaufman Possibly overclaiming haha 2021-06-04 19:18:17 RT @samkaufman: @TaliaRinger Congratulations! My favorite bit: “Humans and machines are not that different.” 2021-06-04 18:34:15 I'm keeping the title on Twitter for a tiny bit just to celebrate but I usually don't like titles lol, so it will go soon after 2021-06-04 18:30:26 @karger @linguangst @CantlonLab @wendyju @UMichCSE Too short notice to find someone else to live there, and wasn’t allowed anyways 2021-06-04 18:29:28 Me and my sister lol https://t.co/hcc3O216Y2 2021-06-04 18:10:27 Someone talked me into recording it, so I'll upload and post the video later if people want to see it 2021-06-04 18:06:31 Dr. Talia https://t.co/WLViwLy8bF 2021-06-04 17:59:33 Waiting through the gossip sesh when they kick you out lol 2021-06-04 17:56:25 RT @larsr_h: Watching @TaliaRinger defend her thesis and I really love the content & 2021-06-04 14:59:34 @codydroux Haha this is a good one 2021-06-04 14:35:26 @GrammarTable Mostly because I always pronounce this as "would've" no matter how it is written 2021-06-04 14:34:53 @GrammarTable 1 or 3, but 2 sounds so weird 2021-06-04 14:24:50 This offer also holds for anyone who left Google AI because of the way the Ethical AI team was treated within the last 6 months 2021-06-04 14:15:26 Or "hey at least after this talk, you don't have two days full of virtual 1:1s" 2021-06-04 14:13:11 Or maybe "just show up and it will be over before you know it" 2021-06-04 14:12:17 So many people messaging me today like "you'll do great!" Easy for you to say when you're not the one throwing together a defense over 2 weeks late in a pandemic. I recommend "you'll survive" or "you won't fail" 2021-06-04 14:08:00 Reminder that LGBT Israelis existing is not some government plot or pinkwashing or whatever 2021-06-04 14:03:33 Same https://t.co/uFcK6b0jQW 2021-06-04 13:55:48 RT @mathhombre: Literally dreamed this https://t.co/2vI9qGH47r 2021-06-04 13:23:19 @kamatsu8 We just need to stop splurging on avocado toast 2021-06-04 13:18:26 The US during the pandemic https://t.co/0trMXn6R3E 2021-06-04 13:12:48 It's almost like after the first recession we went through, boomers just hired each other back and left my generation to rot. And then there was a pandemic and second recession. Building wealth? Some of my friends are struggling to afford food https://t.co/sPOnCG4jeH 2021-06-04 05:40:25 @JAldrichPL @SouthernSummr Yeah, I think a healthier view of both species and human sex is one of clusters 2021-06-04 05:39:16 Constructive logicians talk about sexual dimorphism part one https://t.co/2DKggLp5lX 2021-06-04 05:35:45 @JAldrichPL @SouthernSummr But then again I'm an intuitionist! 2021-06-04 05:35:14 @JAldrichPL @SouthernSummr I don't think it's reasonable to discretely bin without acknowledging complexity is inherent and exceptions are humanly defined 2021-06-04 05:31:59 @JAldrichPL @SouthernSummr I definitely think going from clusters to discrete bins defining everything else as an "exception" is a bizarre human thing, one I don't understand the desire to do at all 2021-06-04 05:30:53 @JAldrichPL @SouthernSummr And one of my girlfriends was not considered intersex but had three X chromosomes. It's exceptionally complicated. 2021-06-04 05:28:49 @JAldrichPL @SouthernSummr I don't know, I think it's kind of all BS, like even what we call intersex versus not is a weirdly human defined thing. I have PCOS and produce lots of extra testosterone, but we've decided that's fine to be a woman while having internal testes produce it makes you intersex 2021-06-04 05:27:45 RT @hisham_hm: @JAldrichPL @TaliaRinger @SouthernSummr my favorite statistic on this topic is that there is about the same amount of inters… 2021-06-04 05:25:26 @JAldrichPL @SouthernSummr We say "humans have two nipples" but this is super false many of us have three or more. The obsession humans have with trying to simplify everything when nature rarely makes anything simple is really weird 2021-06-04 05:24:01 @JAldrichPL @SouthernSummr Humans love to simplify things. Evolution rarely makes things that simple. Humans don't even all have the same number of bones. There are bones that 10% of us have, 40% of us have, and so on 2021-06-04 05:22:52 @JAldrichPL @SouthernSummr It isn't just "a little" blurry though, it's exceptionally blurry, just doctors will go as far as to cut genitals of kids without consent just to assign a sex 2021-06-04 05:21:17 @SouthernSummr It clusters in a way that makes it easy for humans to say "OK then there are 2, and everyone else is an exception we'll pathologize" but the binning is a human thing 2021-06-04 05:20:16 @SouthernSummr Sex isn't discrete it's continuous and there are multiple dimensions along which it varies, since it refers to a whole bunch of usually aligned things like chromosomes, hormones responsiveness, hormone production, genitals, and so on 2021-06-04 05:18:00 @robs_cse But I'll align with women when discussing things like discrimination, harassment, abuse, lack of representation in the field, and so on and just say I'm a woman in those situations because it's easier than explaining in what sense I'm a woman and in what sense I'm not 2021-06-04 05:16:27 @robs_cse I just accepted it makes me a gray agender woman. My experiences and presentation are that of a woman and that's relevant but I don't have anything inside me that "feels like a woman" and that's cool. People can call me any pronouns they want 2021-06-04 05:14:18 @SouthernSummr Evolution didn't discretely bin humans into two sex categories. Humans did that, and then went on up develop entire sets of norms for each category, and hurt people who defied them 2021-06-04 05:12:36 RT @kenklippenstein: After the stimulus checks, reports of food shortages fell 42% and "frequent anxiety and depression" fell by over 20%,… 2021-06-04 05:12:06 RT @NPR: A single California wildfire last year wiped out up to 14% of the world's mature giant sequoia trees, a new draft report finds. ht… 2021-06-04 04:57:40 I super love how much my generation and younger generations have just kind of wholly rejected the notion of a forced gender binary because honestly like who thought "there are two kinds of people" was ever a good idea 2021-06-04 04:56:27 I still don't have a sense of gender identity internally or whatever but seems like a lot of people do and it makes the world really interesting, adds this whole dimension of identity and expression, seems cool 2021-06-04 04:55:01 I hope in the future the only people who care about sex assigned at birth are doctors. Seems so irrelevant to basically every other situation. I love how these days I just can't tell at all even if I try, and I don't care at all because why would it matter to me 2021-06-04 04:39:35 I'd be panicking if I had enough energy to panic 2021-06-04 04:31:33 RT @niftyc: This is a major milestone for researchers working to stop racial and other forms of discrimination online. The work that these… 2021-06-04 04:31:31 RT @niftyc: In an "important victory for civil rights," today the US Supreme Court ruled that journalists & 2021-06-04 03:32:08 RT @ChrSzegedy: @anton_hilado @TaliaRinger Not just the proof, but Adam Brown's unexpected connection between the mathematics of bouncing b… 2021-06-04 03:16:31 RT @Browtweaten: tchaikovsky during the 1812 overture: https://t.co/9StWmsMgzR 2021-06-04 03:09:45 RT @anton_hilado: Math Twitter: What's your favorite proof? 2021-06-04 03:09:40 RT @MathAndCobb: My favorite theorem is Euclid's proof of the infinitude of the primes. Because it is profound, elegant, intriguing, constr… 2021-06-04 02:30:06 DM me or email me if interested. @timnitGebru and @mmitchell_ai can probably vouch for me being a reasonable person. (I hope!) 2021-06-04 02:27:36 My credentials: I went on the job market this past year and got hired by UIUC in the CS department. I am very fresh out of that process and feel very comfortable giving good advice about it. I also just want to see that whole team end up happy. 2021-06-04 02:25:51 I'm happy to review your materials, help you find places to apply, help you find places hiring in Ethical AI, watch your job talk and give feedback, find contacts at universities that are hiring, and even try to initiate off-cycle hiring if it's urgent. 2021-06-04 02:24:22 I did just notice that this article says some of the Ethical AI team members are considering academia. Blanket offer: if any current or former Ethical AI team member wants help with faculty job applications next year, I will very gladly help. https://t.co/6qZLELV5bw 2021-06-04 01:58:25 RT @ThePlanetaryGuy: In Scenario 2, Venus enters a runaway greenhouse, but not because of its proximity to the Sun—but from a self-inflicte… 2021-06-04 01:58:19 RT @ThePlanetaryGuy: But in Scenario 2... Venus might have been just like Earth. *If* Venus survived its early childhood, recent models su… 2021-06-04 01:57:56 RT @ThePlanetaryGuy: It all comes down to one simple question: Why is Earth's sibling not its twin? (Credit to @efrankplanetary for that… 2021-06-04 01:43:03 RT @ThePlanetaryGuy: Psst. Interested in learning more about #Venus—and why @NASA's decision to go back there with the #VERITAS and #DAVIN… 2021-06-04 01:37:30 RT @SupernovaMomma: There should be no guilt, shame, and blame in those convos though. Just acknowledgement, responsibility, apology, and f… 2021-06-04 01:37:13 RT @SupernovaMomma: Two things can be true. 1)You did your best. 2)Your best caused harm. It’s a hard truth. 1) You are a product of yo… 2021-06-04 01:35:58 RT @ZJAyres: @ravenscimaven I still get anxiety I'm going to fail my defense four years later... 2021-06-04 01:35:42 @ravenscimaven Good luck!!!!! 2021-06-04 01:35:34 @ravenscimaven It's not just you lol https://t.co/qB83hhFePV 2021-06-04 01:34:44 I do have people coming from around the world since it's virtual, which is interesting. 2021-06-04 01:33:23 Is it normal to feel so nervous the day before my defense even though basically nobody ever fails these things? 2021-06-04 01:18:59 y'all gonna hear about Lambek's theorem though 2021-06-04 01:16:49 bonus slides in case anyone really wants to see this, but so that I don't have to burden anyone who doesn't lmao https://t.co/eN0PxaqSyP 2021-06-04 00:39:23 RT @samkeslr: @ryangsims link to the full list of favorites from @axios: https://t.co/G6LBMisrSw 2021-06-04 00:39:10 RT @ryangsims: elementary schoolers got to submit names for Colorado's new snow plows and damn https://t.co/SO64lSDtFR 2021-06-03 23:45:41 Weird https://t.co/Ne3W5C1Smh 2021-06-03 22:43:38 I don't even want to tag anyone at this point. This all feels kind of hopeless. 2021-06-03 22:42:18 In short, this all sucks and continues to suck, I'm exhausted, please just listen to the Ethical AI team. It's the right thing to do and the rest of the research community will forgive you if you do that well even if it's 6 months late. 2021-06-03 22:40:34 I also feel like I get more and more tired every time I try to push for any change, with such an opaque structure at Google that I don't even know who makes these decisions and why. I don't even know who feels sorry internally. 2021-06-03 22:39:11 I feel like I really came to like a number of the current and former team members personally as friends during the harassment escapade I got pulled into, and I feel really sad to see their careers derailed like this. 2021-06-03 22:37:09 Not on the count of the team, to be clear. Just on the part of Google to not do right by Timnit and Meg, while the headless Ethical AI team is left with no real choice but to ask them for free labor, which they'll do because they care about the team. Seems bad. 2021-06-03 22:32:25 This is like peak invisible labor though. I find it really upsetting. 2021-06-03 22:29:36 @karger @linguangst @CantlonLab @wendyju @UMichCSE I can say that it was life-saving. I know this because I used to replace "I want to kill myself" with "I want to go to San Diego" that year, and I ended up in San Diego twice. 2021-06-03 22:28:48 @karger @linguangst @CantlonLab @wendyju @UMichCSE But I got that through informal connections. I think making it easier to access would be better. I wouldn't have thought to ask for it without the right people telling me to consider it and setting it up. Housing help would have also been really good. 2021-06-03 22:26:59 @karger @linguangst @CantlonLab @wendyju @UMichCSE This one is inspired by something I did, not during an investigation, but while waiting for my abuser to graduate, and fleeing ongoing retaliation. UW continued to pay me, and UCSD hosted me. Unfortunately, I did have to pay for housing in both cities at the same time. 2021-06-03 22:23:44 Given that they're continuing to help manage the team for no pay out of the goodness of their hearts, yes, apologizing and offering them their jobs back is the right thing to do, even if they say no. 2021-06-03 22:22:13 The outstanding asks: "They want Google to publicly commit to academic freedom and clarify its research review process. They also want it to apologize to Gebru and Mitchell and offer the researchers their jobs back — but at this point, that’s a highly unlikely prospect." 2021-06-03 22:12:23 "This has left the ethical AI unit running itself in an ad-hoc fashion and turning to its former managers who no longer work at the company for informal guidance and research advice." So they are even still working & 2021-06-03 21:48:52 RT @__apf__: the fun thing about low probability events is that they happen 2021-06-03 21:25:18 @linguangst @karger @CantlonLab @wendyju @UMichCSE 5) help setting up short-term research visits or internships for the accuser to get away while an investigation is ongoing 2021-06-03 21:20:42 @linguangst @karger @CantlonLab @wendyju @UMichCSE Examples: 1) making it OK not to get a letter from your advisor 2) funding guarantees that make it easier to switch labs or advisors 3) easier ways to transfer between universities 4) preemptive documentation & 2021-06-03 21:18:45 @linguangst @karger @CantlonLab @wendyju @UMichCSE I think before we remove confidentiality at any point in the process (with permission from the accuser, otherwise you are putting them at great risk of personal harm), we need to fix the systems that make covert retaliation so simple 2021-06-03 21:16:05 haha I'm defending my thesis tomorrow what 2021-06-03 21:14:59 @karger @CantlonLab @linguangst @wendyju @UMichCSE I'm not huge on breaking confidentiality honestly. I think it would increase retaliation extremely significantly 2021-06-03 18:04:00 I went in 2019: https://t.co/C581ugZbKw It was at UIUC 2021-06-03 18:02:11 Hi! You should apply to go to Rising Stars if you're eligible. It's great. https://t.co/oae5aEtsBY 2021-06-03 17:25:52 She's the best 2021-06-03 17:25:34 I have a lot of trouble getting out of bed when I have nobody to meet in person somewhere. Lately my mom has been ordering breakfast for me on the hardest days so that I have to get up and meet the delivery driver 2021-06-03 16:53:32 RT @morgangames: @TaliaRinger @JAldrichPL @karger @wendyju @linguangst @UMichCSE Due process is a legal construct, not something governing… 2021-06-03 16:51:49 @disconcision @sliminality If it's someone whose feelings I don't want to hurt, I just mute them 2021-06-03 16:45:10 @JAldrichPL @karger @wendyju @linguangst @UMichCSE There's also a sexist effect where evidence not from men is often just ignored regardless of the gender of the accused? The same way often when I make logical arguments, despite literally being a logic expert, people often tell me that I need to stop arguing with my "feelings" 2021-06-03 16:39:26 RT @DrTarikaBarrett: A new study shows the only section of the U.S. where boys are much more likely to outperform girls in math are rich an… 2021-06-03 16:34:58 @disconcision @sliminality I unfollow when I feel like reading someone's Tweets is making my life worse 2021-06-03 16:32:51 @JAldrichPL @karger @wendyju @linguangst @UMichCSE I don't think this is why men get the benefit of the doubt. You can present direct evidence of abuse and the man can lie and a large number of people always just believe those lies even in the face of evidence. It's a thing you get from being a white man 2021-06-03 16:29:05 RT @FreyaHolmer: I want to look into sound absorption for my DDR setup, does anyone know where to start? https://t.co/pH3SzNjUpM 2021-06-03 15:51:34 RT @Catshealdeprsn: The most beautiful flower in the garden https://t.co/p430TISF80 2021-06-03 15:46:54 RT @ChristopherJM: New @ADL report findings indicate "anti-government and white supremacist beliefs are the most commonly held extremist be… 2021-06-03 14:53:20 @asajeffrey That seems a bit unfortunate 2021-06-03 08:26:31 Probably with a bit more softness and kindness toward people around me, but I think June is supposed to be a sort of hyperbolic example of the "burn down the oppressive system" response and we need people like that or the systems will sustain themselves and harm more people. 2021-06-03 08:24:11 I will forever identify with June, even if I probably wouldn't want to hang out with her personally. I'd probably respond the same way. 2021-06-03 08:22:51 Reading online commentary on The Handmaid's Tale, the latest episode, and honestly I'm kind of surprised that so many women think choosing to heal internally is better than choosing to fight externally and burn the whole oppressive system down. Even when the system is Gilead! 2021-06-03 08:20:20 RT @justinesherry: @PatPannuto pointed out an important lesson a few days ago for harassment cases: making sure students can get out of the… 2021-06-03 06:30:16 @Daniel_Nikpayuk @va2lam I don't really understand what you're trying to say but I think "inheriting a large amount of money tends to make your life easier" is a relatively uncontroversial statement 2021-06-03 06:25:08 @Daniel_Nikpayuk @va2lam In my culture, FWIW, complaining is a form of bonding, often done with love, or to find humor in dark situations when we don't know all of the answers 2021-06-03 06:21:12 @Daniel_Nikpayuk @va2lam This is Twitter, I'm allowed to have an outlet sometimes 2021-06-03 06:20:07 @Daniel_Nikpayuk @va2lam I... I think it would be hard to follow me for more than like a day and not see that I'm frequently engaged in enacting systemic change at every level from my university to national politics lmao 2021-06-03 06:18:01 RT @tianyin_xu: Correctness is a serious topic, but correctness research is fun! @xudongsun_uiuc will show how #Kubernetes Operators could… 2021-06-03 06:17:31 My comfort with occasionally using swear words like "I'm a grown ass professor" has gone up dramatically since securing a faculty job, which means for the past six years I've been subconsciously doing the grad school equivalent of not swearing in front of my parents 2021-06-03 06:06:15 RT @laurelwamsley: Hi. I reported this guide to gender identity terms for @npr, and I learned a whole lot in the process. I hope it's use… 2021-06-03 06:03:09 RT @IntuitionisticB: @mathslogicbot ¬(¬((a⇾a)⇾b)⇿b) is not an intuitionistic tautology 2021-06-03 05:58:34 @miniapeur Standards and competition have risen, but so has funding. I don't know which have grown faster 2021-06-03 05:56:15 @sabahzero @drtowerstein I'd pull it. I wouldn't trust someone who wrote that to faithfully represent my work anyways 2021-06-03 05:50:05 Like professors used to be intimidating people with status but now a lot of them are just friends 2021-06-03 05:48:46 I met all of these grad students when I started grad school and it's so cool to see so many of us become professors and just spread out around the world I'm so excited 2021-06-03 05:40:39 @miniapeur Guilt? Humility? Impostor syndrome? Bitterness about broken systems? I'm not sure either 2021-06-03 05:34:25 Companies: rainbow logos are what the gays want The gays: wait what about fighting for trans rights to exist and use the bathroom in public Companies: logos. Logos are the only answer The gays: 46% of bi women are raped can you help build systems to-- Companies: logos 2021-06-03 05:31:43 Thank god Uber highlights the path to pick me up in rainbow colors so that I know that, uh, wait what actually is this supposed to tell me 2021-06-03 05:30:36 @jbakcoleman I think denial at the start of the pandemic was normal. Just a part of grief 2021-06-03 05:09:24 @anqiwu_angela Thanks! 2021-06-03 04:43:40 @krismicinski @palvaro TBH I think I'd score a lot higher just by taking my ADHD medicine before the test. I took the test unmedicated because it was for my ADHD diagnosis (to rule out LDs). But I also don't want to contribute to a metric I don't like much by taking a test "just for fun" 2021-06-03 04:40:48 @palvaro I've managed to do well on a lot of these, but I know for the ones I don't do as well at (like when I talk to my friends with 140+ IQs on official tests who say IQ is bullshit) I just feel mildly bitter and jealous haha 2021-06-03 04:32:15 @va2lam In this world, being born to a rich family is probably a lot more useful than being smart 2021-06-03 04:26:24 @yoavgo yeah mostly just dark humor on how stuck we are with systems none of us actually really like that much 2021-06-03 04:25:00 @krismicinski shhhhhhhhh 2021-06-03 04:21:19 @va2lam Still I'd love an IQ of 140+, it's probably a proxy for something useful 2021-06-03 04:20:48 "USNews rankings are total BS 2021-06-03 04:20:19 "CSRankings is biased against small venues 2021-06-03 04:19:45 "test coverage is a useless metric 2021-06-03 04:19:10 "conference rankings are ridiculous 2021-06-03 04:18:44 "h-index is useless (mine is for sure not 27, just variations on a theme) 2021-06-03 04:17:15 I think "IQ tests are largely biased and even what we consider to count as intelligence is largely biased, but still those tests are a proxy for something useful" is more accurate 2021-06-03 04:16:20 It's like how everyone I know with a measured IQ of 140+ is like "IQ is irrelevant" lol 2021-06-03 04:15:24 Academics be like "awards and publications don't matter and are largely random 2021-06-03 04:12:36 @alpha_convert I'm a grown ass professor I can hoard all of the dishware if I want 2021-06-03 04:12:04 @alpha_convert you can't tell me what to do 2021-06-03 04:11:51 RT @alpha_convert: Hey you! Yeah, you! This is a gentle reminder to clean the cups off of your bedside table 2021-06-03 04:02:08 @johnregehr @krismicinski For academia it's sometimes rewarded but sometimes ignored lol 2021-06-03 04:01:49 @johnregehr @krismicinski Ah forgot about that. For industry spending a lot of time just making the system itself awesome is also highly rewarded 2021-06-03 03:58:12 @_chenson__ @16kbps Also @samps sent me his thesis with the beautiful format. I think I need to ack him still actually 2021-06-03 03:57:39 @_chenson__ Thanks!! I put a lot of work into that. Huge thanks to @16kbps for introducing me to the knowledge package for cross-referencing 2021-06-03 03:55:56 @johnregehr @krismicinski I mean, if you don't have students depending on the paper to graduate lol 2021-06-03 03:37:09 @krismicinski @johnregehr Expectations were thoroughly reasonable when I worked on test input generation 2021-06-03 03:34:50 @krismicinski @johnregehr Being an applied type theorist is rough though, reviewers are often like "why didn't you both prove your system completely correct and also run a thorough empirical evaluation of its utility" and I'm often like "um because that would take 3 years" 2021-06-03 03:32:32 @krismicinski @johnregehr Yeah I remember I got into CCS once with like B/C/D or something ridiculous hahahahaha 2021-06-03 03:29:30 @johnregehr @krismicinski No it's more like aggressive philosophical purity positions 2021-06-03 03:27:57 @krismicinski @johnregehr Especially type theory reviewers oh my gosh 2021-06-03 03:27:32 @krismicinski @johnregehr I would be genuinely surprised if type theory and proof assistant reviewers don't skew lower in general 2021-06-03 03:23:52 @johnregehr I think I got like BABB or something like that this time, or BABA, I don't remember, but either way those were my best scores, there is always a C naysayer (the only one changed their mind after rebuttals this time) 2021-06-03 03:21:27 @johnregehr Is that even possible when type theorists look at your paper lmao 2021-06-03 03:20:45 @JAldrichPL Ah I never pick hot topics haha always new ones 2021-06-03 03:20:02 @krismicinski I mean, I haven't gotten one in six years, I don't know if I ever will 2021-06-03 03:19:30 @krismicinski I would think any paper that gets into PLDI and so on is genuinely pretty solid lol 2021-06-03 03:16:36 Is this process even blinded? 2021-06-03 03:14:45 Honestly, how do people get best paper awards and stuff? This has always confused me, the whole process seems so opaque and I don't understand what I would need to do to ever be considered for one 2021-06-03 03:11:31 I'm way more of a "first" person than a "best" person. So I find it really easy to pitch ideas, I basically never struggle with novelty, and I'm good at writing proposals and implementing new ideas. But I've never gotten any "best" award, and I don't know how people do that 2021-06-03 02:56:44 not necessarily faithful though 2021-06-03 02:54:45 I don't think my kind of work is ever the kind of work people care about to get "best" awards and stuff, but I put a lot of effort into this thesis to help my future students think of cool ideas and find useful background information 2021-06-03 02:49:58 OK, in other news, I've finished a complete pass on my thesis after feedback from my advisor. I haven't gotten to any of your feedback you've all given me this past week 6.1 is new! 2021-06-03 02:49:00 RT @littmath: How many of the responses to this poll will receive at least 24% of the vote total, when the poll is complete? 2021-06-03 02:47:04 all paragraphs must be full 2021-06-03 02:46:23 For the rest of my life after grad school I'm going to have nightmares about paragraphs that end with a line that has just one word 2021-06-03 02:18:22 Sometimes I type things in my writing and I'm like "where the hell did that come from, OK, I guess I'm keeping it." "but of course not everyone is me" lmao Talia https://t.co/3DkfCk8ycq 2021-06-03 02:14:27 RT @cestlemieux: Didn't think that when I left Soda Hall in March 2020, I would never be back there as a student... https://t.co/QZJp5pykls 2021-06-03 02:14:24 @cestlemieux mooooood 2021-06-03 02:14:18 @cestlemieux Congrats!!!!!! 2021-06-03 02:14:12 RT @cestlemieux: Yep, turns out I have a PhD now! https://t.co/ZUGiob4pK9 2021-06-03 01:34:53 RT @courtneymilan: Twitter hates your brain chemistry and will delightedly trash it for engagement metrics. 2021-06-03 01:31:53 The depression is kind of eating at me today even though most days are good these days. I already had boba twice, too 2021-06-03 01:28:06 RT @FineEthan: @zackbeauchamp from @isaacdecastrog https://t.co/HqLtrqMAo8 2021-06-03 01:22:03 @alpha_convert If SIGPLAN wants to require some number of hours from me every year lobbying for political changes that lead to more sustainable energy sources though I'll gladly do it 2021-06-03 01:19:30 @alpha_convert Like FFS even their PC meetings involved flying and they want to tell me I can't fly to conferences get out of here 2021-06-03 01:18:49 @alpha_convert I am extremely bitter about the more senior generation of researchers trying to dictate the experiences of the more junior generation, and glad that we've managed to fight to have a voice this past year 2021-06-03 01:14:21 "But if every single person just---" hi did we just experience the same pandemic or 2021-06-03 01:12:24 Even though choosing not to fly for academic reasons would have an insignificant dent on climate change and larger systemic regulations and incentives much higher up are the only way to save the world but everyone wants to feel like they're trying I guess, or making us try 2021-06-03 01:10:28 I know I know the previous generation could have as much fun as they wanted at the expense of the climate but we aren't allowed to have fun anymore in the world they've created 2021-06-03 01:08:43 Senior faculty: it was nice not flying for a year Me: going to France every year was literally the best part of this job please dear god resume annual trips to France 2021-06-03 01:03:25 Two years ago on this day I was in France for the @CoqLang users & 2021-06-02 23:13:44 "albeit" is my favorite word tbh 2021-06-02 22:21:43 @DavidLawler10 I mean, Lapid knew. Magician, that guy 2021-06-02 22:21:13 RT @DavidLawler10: Why didn't anyone realize that all it would take to oust Bibi was a right-winger forming a pact with an Islamist party,… 2021-06-02 21:41:05 @sibinmohan Oh, absolutely terrible. But he has so few seats he won't be able to accomplish anything he really wants to accomplish. Also the government staying together hinges on the participation of one of the Arab parties, with about as much representation as Bennett 2021-06-02 21:39:54 @vj_chidambaram I've had good research conversations over drinks but that can happen on my own time, with my own money, wherever I want to go. Zero need for it to be a part of the conference, and going elsewhere lets me choose whom to trust and bring with me 2021-06-02 21:38:55 @vj_chidambaram We should be decoupling conferences from drinking. Super weird to tie our academic lives to the use of a particular drug 2021-06-02 21:17:59 @BethShakir @Yair_Rosenberg Lol OK. Some dark humor usually helps with dark situations though 2021-06-02 21:13:06 @BethShakir @Yair_Rosenberg There are rockets all the time, not all of them escalate 2021-06-02 21:11:36 @BethShakir @Yair_Rosenberg I don't think you understood the joke? It was about Bibi doing anything to stay in power, including starting conflicts that benefit him 2021-06-02 21:03:30 @BethShakir @Yair_Rosenberg Just enough time to start another war 2021-06-02 21:01:30 @samth I'm hopeful, but who knows what kind of nonsense he'll pull 2021-06-02 20:59:44 @samth Skepticism warranted 2021-06-02 20:58:56 https://t.co/y69orD6Mx3 2021-06-02 20:58:46 RT @Yair_Rosenberg: NOT YET. They still have to vote this new government in, and Bibi has time to apply immense pressure and try to break u… 2021-06-02 20:58:42 RT @Yair_Rosenberg: Here's the video of Israeli opposition leader Yair Lapid making the call to Israel's president officially informing him… 2021-06-02 20:58:31 RT @Yair_Rosenberg: Israeli opposition leader Lapid, tasked with solving the Rubik's cube of Israeli politics and uniting the anti-Netanyah… 2021-06-02 20:52:55 אולי יש עתיד 2021-06-02 20:50:50 https://t.co/Os5eu6Yp3Z 2021-06-02 20:50:10 https://t.co/6GcffxlCKx 2021-06-02 20:49:15 https://t.co/pdW6scDwGo 2021-06-02 20:44:54 (There are still ways this can fall through apparently, just fewer) 2021-06-02 20:43:35 RT @Proof_by: @TaliaRinger I like this thread so much. I can't properly put it into words but it's like a perfect snapshot of every second… 2021-06-02 20:43:11 Go to jail. Do not pass go. Do not collect 800 shekels 2021-06-02 20:39:19 Bye-bye Bibi 2021-06-02 20:39:08 RT @yairlapid: עלה בידי 2021-06-02 20:30:56 RT @y_sarfati: @Istanbultelaviv Hope he doesn't start a war until the vote of confidence 2021-06-02 20:28:38 RT @Istanbultelaviv: Its a done deal. Just waiting for the official notice. Netanyahu is over (let's hope it holds up). 2021-06-02 19:28:36 @Jding314 Right yeah that's what I need to do though https://t.co/SV5VvWso08 2021-06-02 19:17:54 @mtesseract I've seen something that looks like `` only bigger, and also a square 2021-06-02 19:13:35 @zschutzy "we must reify and transform the smiley face" would be funny in my defense but also confusing hahahaha 2021-06-02 19:11:58 @mtesseract Oh gosh now I have 3 options haha. Is this just not standardized? 2021-06-02 19:10:44 I've seen a square before, and I've also seen little quotation mark thingies. I don't know which is standard. 2021-06-02 19:09:40 Just trying to emphasize it on a slide because when I move from category theory to type theory, I need to actually reify the proof that the diagram commutes and reason about it explicitly. 2021-06-02 19:08:26 When people need to emphasize "the diagram commutes," how do they usually highlight that? Is there syntax I can use that says "this diagram commutes" explicitly? 2021-06-02 17:29:17 @hannelita @SigplanM if it helps even one person it is a success 2021-06-02 17:16:36 RT @hannelita: A few weeks ago, I sent out the tweet below and I received several DMs with folks sharing their stories and also looking for… 2021-06-02 17:16:32 RT @hannelita: 1. Mentorship programs are awesome (see @SigplanM wonderfully organized by @TaliaRinger, targeting PL folks). Talking to som… 2021-06-02 17:12:33 RT @ChaseTMAnderson: Normalize LGBTQ+ people living in a world where they can live authentically as themselves every day. 2021-06-02 17:00:49 RT @theADHDacademic: #ADHD is realizing you've spent too long working on one paper and need to switch to working on something else followed… 2021-06-02 16:19:00 RT @djg98115: "Hey, Dan, do you actually read your students' dissertations?" "Yes, yes I do." 2021-06-02 16:12:55 RT @ev_maus: This is so important to talk about. (The numbers for bi men are also higher than gay men or straight men for rape or sexual a… 2021-06-02 15:13:13 @mattnightingale I inferred that maybe life jackets aren't cool 2021-06-02 15:12:54 @mattnightingale I remember this moment well, because I was trying to "look cool," but my family laughed at me and I didn't understand why they were laughing because I thought I was so cool 2021-06-02 14:34:28 RT @ThomasVanRiet2: https://t.co/AzFF9P0guh 2021-06-02 07:40:42 RT @kazADHD: packing with ADHD: "if i forget it, it wasn't important enough to remember and i don't need it anyway" also packing with ADHD… 2021-06-02 07:31:21 Whenever American politics feel awful I just remember the other place I'm a citizen of somehow has even worse politics https://t.co/ZndEWYm1mQ 2021-06-02 07:01:52 Retweet with a photo that proves you were a fashion icon in your youth. https://t.co/AOhZyeNiMX https://t.co/fFnLFUmPgO 2021-06-02 06:54:11 @karger @wendyju @linguangst @UMichCSE Yeah, also no matter how many women come forward with stories, somehow it's those men who always get the benefit of the doubt 2021-06-02 06:51:53 Yikes https://t.co/nMsLziEgzO 2021-06-02 06:46:49 Then I dated an emotional abuser it feels way safer to date people who aren't men, but I won't rule men out. Just, I wish all women were lesbians these days, things would be much simpler 2021-06-02 06:44:11 I couldn't even touch a partner lightly like as a hug after that for like 2 years, and couldn't enjoy relationships for another 3 years after that 2021-06-02 06:42:50 Another was a guy in college I had a crush on, and wanted to date. He couldn't accept that just because I'm bi doesn't mean I'll want to get physical with him on the first date. So he invited himself over, did what he wanted, and then told me he didn't do relationships and left 2021-06-02 06:39:37 One of my assailants used his then-girlfriend (who is also bi and whom he also abused) to lure in bi women to assault. This was a serial thing he did at parties. People would drink and then make out with her, and he would move in and take over when they weren't paying attention 2021-06-02 06:36:15 For straight women, it's 17%, and for lesbians, 13%. But for bi women. 46%. Seriously. 2021-06-02 06:35:15 You know a lot of bi women probably. Just like, 1 in 2. Imagine that. That's astounding. The amount of violence we face for our orientation is through the roof. 2021-06-02 06:33:51 Here is a serious pride month thing: 46% of bi women have been raped, and the majority of PTSD cases are survivors of sexual violence. I have experienced this specifically for being bi: men were too offended to let me say no if I'd say yes to women. https://t.co/9v05GGzEXK 2021-06-02 06:27:42 Actually not even independent, again I think it made me much more compassionate and expanded my worldview to understand what people are going through and how difficult life can be 2021-06-02 06:27:01 But that's what PTSD is like. All of this totally independent of my compassion and worldview and so on 2021-06-02 06:26:24 Over time the more I make myself talk about the traumatic events, the less I get sent into that spiral involuntarily 2021-06-02 06:25:44 And sometimes something in the environment triggers that despair or panic and I don't know why, but suddenly I don't feel like I'm in the present anymore, and instead I'm living in the memories, experiencing them in real time, with no sense of my surroundings 2021-06-02 06:24:30 Also college, from a different event. Basically my life just has these weird gaps where I know the events that happened but can't actually relate to having been there to experience them, unless I try really hard, at which point it sends me into a spiral of intense despair 2021-06-02 06:21:16 Also means I have completely dissociated from entire years of my life and cannot relate at all to the person in the memories, so my life feels like it has a huge gap in it several years long, including most of grad school 2021-06-02 06:18:31 PTSD gives me a lot of nightmares and episodes of panic and despair, and occasionally makes me freak out and express my panic on social media or in 1:1 conversations. But also makes me more empathetic toward suffering and the responses people cannot easily control 2021-06-02 06:16:54 Most of the service work I did in the past couple of years was in direct response to my PTSD, like legit helping people because the trauma was so bad I didn't want other people to have to deal with it too https://t.co/cqQtQTRqaw 2021-06-02 05:46:20 RT @bascule: @hipsterelectron @BlueSpaceCanary @mgattozzi Yep, you're left wondering if someone writing high-level Go code forgot to check… 2021-06-02 05:45:21 @ian5v Yeah that specific timeline doesn't work 2021-06-02 05:19:20 @deckeresq Tyyyyy 2021-06-02 05:18:31 I'll get a hotel if need be, but I still haven't filed for the last reimbursement so I'm kind of low on cash haha 2021-06-02 05:17:48 Does anyone have a place in Chambana I can crash next week while looking at a house? 7th through 10th, no CS students due to power dynamics 2021-06-02 05:07:46 Things nobody ever tells you: your thesis is an opportunity to fuzz your advisor 2021-06-02 05:06:55 The best part about revising my thesis is figuring out just what kinds of fun Dan is willing to let me have. "It was wonderful, and involved many trips to France" got a smiley face, but "Leroy and friends 2012" was strictly forbidden 2021-06-02 04:28:09 https://t.co/HsmWF37THH 2021-06-02 03:24:11 @hipsterelectron I think after reading all of the depression lit, I'd probably be OK at understanding the schizophrenia lit. I'm extremely curious 2021-06-02 03:16:49 @hipsterelectron I have read the relevant depression literature in great detail when I was depressed lol 2021-06-02 03:16:12 @hipsterelectron Yeah sure. I don't have a paper or anything though, it's more like "I trust my friend to have read the literature more than I'd trust myself to interpret it" 2021-06-02 03:06:36 @hipsterelectron In any case, this is advice I got from a friend who 1) has a degree in psychology, and 2) does a heck of a lot of LSD. 2021-06-02 03:06:13 @hipsterelectron I think it would be surprising to me if it were not true. Most mental illnesses are combinations of some genetic "switch" and an environmental "trigger." Depression works this way, for example. 2021-06-02 03:01:47 @emremjones But I haven't managed to figure out how exactly these decision processes work at Google, and the opacity makes it difficult to figure out where my effort is best spent. Twitter is usually the outlet I use when I don't feel like I know what else I can do. 2021-06-02 02:59:57 @emremjones They could have done much better. They probably still can do much better, at least to help the people who are still there. 2021-06-02 02:58:51 @emremjones Lately they have not made things much worse, which is nice. I have to give them that. They at some point changed course from repeatedly causing harm to kind of just not bothering to repair harm already done. 2021-06-02 02:49:48 Take note though: this is what helpful information for student safety looks like 2021-06-02 02:44:27 (Also the relevant person was a student, so no need to worry about this still being a thing at UW) 2021-06-02 02:43:38 And I did talk to a faculty member every week for an entire year. I'm really relieved he wasn't mandated to report anything at all, even though this was emotional abuse and not sexual assault---just easier when you don't have to think about mandatory reporting. 2021-06-02 02:42:10 I had so many conversations about this with UW CSE leadership in the last few years, as I'm sure you can imagine given my personality and life experiences. The confidential advocate advice would have saved me from retaliation that almost cost me my life? 2021-06-02 02:36:17 Got an email about WL from the UW CSE director and I'm just so happy these paragraphs are in it https://t.co/qLON5zA2k4 2021-06-02 01:56:16 It's basically the least optimal outcome. Google would be better off losing a lawsuit honestly. 2021-06-02 01:54:37 It's really sad, like it honestly makes me feel very sad thinking about it these days, because nobody wins from this. Timnit and Meg are hurt, Google AI is hurt, the ethics team is hurt, nobody gains anything from this approach at all. 2021-06-02 01:52:21 Truth https://t.co/feznTBVXpc 2021-06-02 01:51:52 RT @NerdPyle: “I am invisible” https://t.co/4CqriMPPRh 2021-06-02 01:51:49 RT @NerdPyle: “Dad says this spot must be popular with wedding photographers” https://t.co/moifYrBBUy 2021-06-02 01:51:44 RT @NerdPyle: “Anybody want to see a thread of our Seattle flower hunt adventures?” https://t.co/gjsHHNbwoC 2021-06-02 01:51:42 @NerdPyle Yes 2021-06-02 01:48:45 Whoever at Google made the decision to follow this approach, though, I hope you're paying attention to the consequences, which were not at all inevitable even after the events of December. 2021-06-02 01:37:32 It does say positive things about the research community, though. There are people in the community willing to do so much to stand up for each other: refuse funding, quit the jobs they've dedicated years of their lives to. That is beautiful. I wish everyone the best. 2021-06-02 01:29:48 I think it's probably too late for anything reasonable to happen, but imagine an alternate reality in which Google had just owned up, apologized, and repaired damage. Maybe a good lesson for other companies in the future on how the research community works. 2021-06-02 01:16:02 If this makes you want to try LSD, do note that if you have any family history of schizophrenia, it is good to avoid hallucinogens since they can trigger schizophrenia for life https://t.co/BNmi1xtl1C 2021-06-02 01:08:56 RT @mmitchell_ai: Google has lost @le_roux_nicolas, a researcher w groundbreaking work on optimization (the learning part of machine learni… 2021-06-02 00:10:04 20) when engaging publicly, know your goals, your audience, and what can go right or wrong 2021-06-02 00:03:41 18) be willing to not just propose things but also do them yourself, otherwise they may never happen even if people are receptive 19) design self-sustaining systems that let you outsource some of your work so that you can move on to other change if you want to 2021-06-02 00:00:57 I'm gonna shower and practice my defense but that's Talia's workplace activism 101 2021-06-02 00:00:01 17) get involved on relevant committees, even just to know what they are up to 2021-06-01 23:58:18 Oh for engaging privately that's just one option. Public engagement also has its place obviously. Just remember that private engagement is also an option. Coordinated public/private teams can be even more amazing 2021-06-01 23:56:57 16) generally just engine building---think about how to enact change that will later help you enact more change 2021-06-01 23:56:03 15) start by building systems that benefit everyone except malicious people, and that prioritize safety over punishment 2021-06-01 23:55:28 14) if speaking up about an abusive or harmful person, expect brutal retaliation 2021-06-01 23:53:27 13) work in teams 2021-06-01 23:53:19 12) UNIONIZE 2021-06-01 23:53:11 11) gauge, over time, the opinions of people around you on different systemic issues you care about, so that you know how much they overlap with you, and how much they will want to work with you to achieve common goals 2021-06-01 23:52:17 9) know the rules really, really well, so you know exactly when you're stepping outside of them, even if in a way that would not normally be enforced 10) know how systems operate 2021-06-01 23:51:05 8) don't be afraid to make friends or even just allies who have a lot of power 2021-06-01 23:50:19 6) if acting in private, coordinate extensively with those acting in public 7) make extensive use of confidential advocates, free or discounted student legal services, and so on to protect yourself preemptively before you act, and talk about the implications of your actions 2021-06-01 23:49:23 5) if you anticipate your advocacy for change backfiring, reach out to someone above you in power ahead of time and explain & 2021-06-01 23:47:56 1) ghost write for sympathetic white man with tenure 2) engage privately rather than publicly 3) know the right people to make the right changes within a system 4) own the narrative yourself in the face of attempted blackmail 2021-06-01 23:46:02 I'd love to collect these tactics somewhere actually. Post any you know of https://t.co/qrXolUxJ1A 2021-06-01 23:42:05 I'd sincerely hope for (1) your motivation is intrinsic 2021-06-01 23:40:46 This is all kind of tangential, I think. Just there are two very different ways I know of to earn a bad reputation: 1) acting like a jerk and hurting people, and 2) trying to help people but pushing too hard too early for the amount of power you have 2021-06-01 23:38:15 Also though, there are tactics you can use early on when you want to push back against systems but aren't in a position to. In academia, one I love is finding a sympathetic White Man with Tenure to act publicly on my behalf. Ghost writing tweets and letters, and so on 2021-06-01 23:36:54 But also you don't want to get pushed out for pushing too hard so you do have to be careful early on when it comes to enacting change that will help people. I've messed this up, most famously with the harasser during my job search that landed me in the newspapers lmao 2021-06-01 23:35:36 I think one of the reasons people hurt each other so much is actually because of rather than in spite of this reputation obsession. You should care more about how you treat people than what people think of you 2021-06-01 23:34:34 If you are not compassionate enough to want to do those things just inherently to be a good person, then I guess care about your reputation. But caring about reputation sets some bad incentives, like not owning up when you hurt people so as to maintain a facade of innocence 2021-06-01 23:32:51 It's maybe weird but I don't like the reputation-focused view. Be yourself, be authentic, be kind, be considerate, act like a team player, and be aware of how much power you have to enact systemic change so you can budget it appropriately and not get hurt too much too early https://t.co/RMAWzCqbRL 2021-06-01 23:27:26 RT @hannahgreene916: Fake women's pants pockets feel like a cruel cosmic joke. Who thought it would be a good idea to taunt women with the… 2021-06-01 22:49:53 Thanks Google, always doing the good work https://t.co/99wXkbZpCX 2021-06-01 22:43:04 @soldni My Twitter followers are just amazing sometimes, like when I'm confused I'll ask "hey is this thing true" and sometimes people just respond with actual proofs 2021-06-01 22:35:43 Probably the only thesis ever that uses the words "a cute proof" and "Twitter" in the same sentence 2021-06-01 22:34:05 footnote footnote footnote footnote footnote https://t.co/z4lJMinyNw 2021-06-01 21:53:06 Just did a major pass on my thesis, in particular with respect to Section 2.3.2 and Chapter 3. If anyone is excited to read it, I'd appreciate some feedback at this point, in case I introduced confusion, typos, and so on. It is here: https://t.co/XlBfGkP8QZ 2021-06-01 18:07:24 Happy Pride from your favorite bi almost-professor! https://t.co/3Kddd7KpfT 2021-06-01 17:04:05 @ZJAyres https://t.co/PsmIA9JpdA 2021-06-01 16:53:19 RT @mmitchell_ai: Happy Pride Month!! https://t.co/m3wNYLNHci 2021-06-01 16:42:48 stated -> 2021-06-01 16:26:02 AI is gay https://t.co/ogCGp2hJ83 2021-06-01 16:24:43 Proof engineering is gay https://t.co/L9DkYfi9D2 2021-06-01 16:21:01 @krismicinski Our new SIGPLAN chair 2021-06-01 16:20:18 @krismicinski Jeff! 2021-06-01 16:16:57 Also probably a good reminder that you can't wellness and mindfulness your way out of oppressive and discriminatory systems 2021-06-01 16:15:38 When my undergrad PL professor told the story of Turing's death and almost broke down in tears in front of the class I was like "oh, wow, I belong in this field, I can safely exist here" 2021-06-01 16:07:52 Computer science is just inherently gay. Why do you all have to talk about being straight in a gay field? https://t.co/Zi1nAMfO50 2021-06-01 16:06:39 Remembering when I stated The Identity Function and got a lot of pushback that was like "it's OK to be gay, but why talk about it?" from other computer scientists. As if our field were not started by a gay war hero forced to take hormones for being gay until he killed himself 2021-06-01 16:01:06 https://t.co/fJf8Ey7uab 2021-06-01 15:36:24 https://t.co/GujnW37l88 2021-06-01 15:16:58 This film was huge in my coming to terms with my orientation though. Plus it's amazing https://t.co/RViZ5EG8dh 2021-06-01 14:59:41 https://t.co/71wLkfl0tJ 2021-06-01 14:55:26 Happy gay https://t.co/iaEz5OLxjB 2021-06-01 14:53:17 RT @carrieffisher: Gay 2021-06-01 14:45:54 @lastpositivist https://t.co/NBlmmq0d7o 2021-06-01 14:44:54 @lastpositivist It's basically a long interview on Putin's life story and politics, in which he is quite transparent about his entire platform 2021-06-01 14:43:45 @lastpositivist Sort of tangential, but I recommend First Person (Putin's autobiography), which is very good for understanding his firm anti-revolutionary view at all costs, in the context of the collapse of the USSR 2021-06-01 14:19:46 RT @arankomatsuzaki: @octopyweb We didn't try enough number of comic style drawings to succeed, but we succeeded with colored pencile style… 2021-06-01 05:44:40 One good thing, I think, is that these days it's basically impossible to form a government without the Arab parties. If the parliament serves its purpose, that will mean some progress, albeit slowly since it is unlikely to be the focus right now 2021-06-01 05:43:13 The first PM in the rotation is far right and awful but only has like 6 or 7 seats IIRC. The Arab parties have about as much representation which would actually be a first, and one of the ministers I think would come from Meretz? It's a really wild potential government 2021-06-01 05:40:43 This is going to be really historically fascinating if it happens. Also a real test of a functioning parliament doing the job it is supposed to do https://t.co/wcW8iip98q 2021-06-01 05:21:03 RT @AnimalsWorId: because every timeline need more baby goats https://t.co/g23Chqkm1k 2021-06-01 05:17:27 @kearonis @arankomatsuzaki What is the prompt you used for this? 2021-06-01 05:09:40 RT @arankomatsuzaki: @braddwyer It didn't perform as well, unfortunately, but "the Nether from Minecraft rendered in Unreal Engine" looks n… 2021-06-01 05:09:25 RT @nickcammarata: @arankomatsuzaki @tayroga In the end humanity was saved by adding "Super safe." to all their requests of the AGI 2021-06-01 05:08:41 RT @arankomatsuzaki: When you generate images with VQGAN + CLIP, the image quality dramatically improves if you add "unreal engine" to your… 2021-06-01 04:29:42 I really think in the future this is the kind of thing people will look back at us for as vile, in the way we look at medieval sewage practices. 2021-06-01 04:29:02 I'm still disgusted that we work in buildings that just recirculate viruses that would not even pass between humans outside, and nobody bothered ever telling me this until a pandemic shut down the world? And also that pandemic wouldn't have happened with outdoor-like ventilation? https://t.co/AGbOXN2txJ 2021-06-01 03:38:34 "What's your research theme" "Fearlessness" 2021-06-01 03:36:12 What a beautiful toolbox you have when you have all of those things though 2021-06-01 03:31:37 I kind of like the idea of being the sort of proof automation person who can handle everything from hardcore DTT to machine learning though. Just years and years of machine learning work to catch up on before I can help students with that side of things beyond high-level wisdom 2021-06-01 03:27:26 At least it has a consistent theme of automating type theory to help real people develop & 2021-06-01 03:26:29 Normal junior faculty pick normal projects for normal people, but ongoing and near future projects I'm on are 1) a machine learning aided proof term to proof script decompiler, and 2) an implementation of automatic transport in cubical as type-directed search using e-graphs 2021-06-01 03:04:25 Well, strongly shortened future vision. But I had too much fun with the type theory. Also I had to refactor large parts of this to get it to address the right level of audience, but it seems good now 2021-06-01 03:03:06 It's basically job talk - job frame + thesis frame - future vision + type theory 2021-06-01 02:59:50 Defense is down to the right length! That gives me 3 days to practice it lol 2021-06-01 00:59:09 omg I miss this game so much 2021-06-01 00:58:55 https://t.co/Ji3uS2TedB 2021-06-01 00:58:05 https://t.co/9pL3T7NhjQ 2021-06-01 00:57:22 https://t.co/eunZSfBaP1 2021-06-01 00:57:02 https://t.co/CVl3XQscw1 2021-06-01 00:56:44 https://t.co/3MqXxu3XUp 2021-06-01 00:56:05 lest anyone forget I'm a huge nerd https://t.co/zzEpORLylW 2021-06-01 00:53:37 TEAM DRAGONFORCE! 2021-06-01 00:49:13 Yesssssssss this one https://t.co/nbqZ22ZpPC 2021-06-01 00:48:30 I wish I could find the Snow Goose video 2021-06-01 00:45:28 I too love Phil 2021-06-01 00:41:23 This is the Zetorux & 2021-06-01 00:36:44 I'm crying https://t.co/BZf40wRJK1 2021-06-01 00:35:18 Zetorux and WinDEU 2008 lmao 2021-06-01 00:34:26 Nice DDRIllini figure in there citing @peztronic https://t.co/2PWctzaGQg 2021-06-01 00:32:04 OK but can we talk about this SIGBOVIK paper??? Amazing https://t.co/qcKJEJf2hH https://t.co/X1WaoWoljJ 2021-06-01 00:22:07 My brain every time anyone says or writes "segue" https://t.co/a9uLodYGD5 2021-06-01 00:16:54 RT @_pdarragh: fun fact: Bajie *loves* water https://t.co/T9cM6V6zkE 2021-05-31 23:51:21 RT @ionathanch: @TaliaRinger Naïve cubical type theory https://t.co/xtclXPg5or 2021-05-31 23:51:18 @ionathanch Interesting! 2021-05-31 23:42:38 @ionathanch Is this cubical? What's it from? It looks pretty 2021-05-31 23:39:44 RT @thegautamkamath: How to have a highly-ranked conference: 1. Make a list of all researchers in the field, ranked by h-index. 2. Invite p… 2021-05-31 23:37:32 RT @stevemblackburn: My inbox this morning is full of discussion of the future of VEE, an excellent venue that is being destroyed by the cu… 2021-05-31 23:35:40 RT @ionathanch: there are enough paths. i've had enough. no more paths https://t.co/2imLwF8anW 2021-05-31 23:31:28 Looks like we may organize some kind of picnic with the lab & 2021-05-31 22:56:03 (Friday, after I defend) 2021-05-31 22:55:56 @HHMusicOfficial It's Friday but thanks 2021-05-31 22:53:58 Very sad about defending & 2021-05-31 22:44:36 RT @AxiomCrisis: I think I need a post for available HoTT/UF PhD positions around the world 2021-05-31 22:43:59 @KevinKaichuang I think it was the sort of "we're all guilty of this" frame. Is it pervasive enough in ML that everyone is guilty of this? Because I don't think it is in my field 2021-05-31 21:57:08 @SidikiXavier lmao 2021-05-31 21:54:20 The students are external it's just super fun to talk to multiple students in an advisory role like that and have the students own it. I can't wait for this to be my job 2021-05-31 21:46:28 @znmeb NASA does some cool work on verification! 2021-05-31 21:41:48 Feeling super happy after advising two students on a project. This is fun, I'm gonna be a professor soon guys https://t.co/F2gyNilK94 2021-05-31 20:04:08 @krismicinski @ChrSzegedy I also think the GenProg example and the work building on it was full of intellectually honest people 2021-05-31 20:03:02 @krismicinski @ChrSzegedy Case studies were for flexibility and usefulness. Which we don't really have good benchmarks for wrt proof repair style work yet. 2021-05-31 20:02:19 @krismicinski @ChrSzegedy Though, 4 case studies was not enough to my POPL reviewers. 8 was enough to my PLDI reviewers, with open code you could run and walk through yourself. But my work is proof-producing which I guess eliminates _some_ need for benchmarking 2021-05-31 20:00:43 @krismicinski @ChrSzegedy Yeah, but in the meantime I think "here are some case studies 2021-05-31 19:57:52 @krismicinski @ChrSzegedy I mean if you also make it very serious, top conference material to design the evaluation methodologies and benchmarks, that should not be as much of an issue. Just doing the work and that is doing several papers of work 2021-05-31 19:54:29 @krismicinski @ChrSzegedy How do we stop the BS eval culture 2021-05-31 19:52:54 @krismicinski @ChrSzegedy That also seems really bad 2021-05-31 19:52:17 @ChrSzegedy ML seems way more benchmark-driven, which is cool, but I think only if the act of designing those benchmarks is considered very serious research in itself (and I've read some good papers on new ML benchmarks, so I assume sometimes it is!) 2021-05-31 19:50:56 @ChrSzegedy I would worry more about this in ML, honestly. Imagine building years of ML work of incremental improvements on benchmarks that don't actually turn out to represent useful gains. That would be devastating 2021-05-31 19:30:04 @ChrSzegedy Mechanized metatheory had a good benchmark suite that drove a lot of good work, but also had the effect of narrowing focus to some problems at the expense of others: https://t.co/nrUQfiDELX 2021-05-31 19:26:41 @ChrSzegedy I feel like this was a simple mistake in metrics that become authoritative and propagated throughout the entire community for years. I really do not doubt anyone's intentions at any point. I just think it points to the amount of care that needs to go into these decisions 2021-05-31 19:24:46 @ChrSzegedy I think the danger is more one of ignorance when the benchmark suite isn't ideal. Most famously, I think about what happened in program repair, when everyone was comparing to one tool, but that tool was actually producing incorrect patches: https://t.co/AEhQF1hL5k 2021-05-31 19:07:01 @BenNiderberg @jacobkornbluh Yeah this is basically a perfect test case for a functioning parliament. It's exactly how parliaments are supposed to work! 2021-05-31 19:04:43 @anqiwu_angela @GTCSE @gtcomputing Congratulations!!! 2021-05-31 18:28:03 RT @ericlippert: @TaliaRinger It's lovely 2021-05-31 18:27:17 Can I get away with this level of artistic freedom in my normal papers, too? https://t.co/xle83dTHZR 2021-05-31 17:55:27 @noneuclideangrl Washington State is so beautiful 2021-05-31 17:53:08 Mountains out fountains out https://t.co/QSbt7LGuSJ 2021-05-31 17:35:20 @VittoMartinelli I view it is a shared commitment with someone I like and feel attracted to, to build something together 2021-05-31 17:34:19 @j_gottschlich I appreciate your actions regardless I think it's more that I don't want relying on good will and power cognizance to be so necessary 2021-05-31 17:32:37 Q: How did you know? A: https://t.co/pmbSqZzQba 2021-05-31 17:30:11 @VittoMartinelli That's all. Just the ability to have crushes on people of any gender 2021-05-31 17:29:23 @VittoMartinelli When I finally learned that it was possible, I realized that all of those things had also been crushes. So then I had my first same-gender relationship at age 16 or so 2021-05-31 17:28:35 @VittoMartinelli Starting from age 10 or so, I used to find that there were girls I really wanted to spend a lot of time with, and whose attention made me feel really good. For boys I always called those "crushes" but I didn't think I could have crushes on girls too. Until high school 2021-05-31 17:24:36 Oh here's a fun one OK. Ask me all any of your naive questions about being bi here. Like "how did you know?" and that sort of thing. I'll try to get to them throughout pride month. https://t.co/HjUA5cc6v3 2021-05-31 17:22:52 I'm #StillBisexual 2021-05-31 17:21:01 Pride month starts tomorrow https://t.co/5OAdFtzTjc 2021-05-31 17:19:30 RT @anqiwu_angela: I’m thrilled to share that I will be starting as an Assistant Professor at the School of Computational Science and Engin… 2021-05-31 17:06:32 RT @agdakx: News: Benedikt Ahrens will be joining our PL group at TU Delft this fall! More news: he's hiring a PhD student. So if you're in… 2021-05-31 17:04:54 RT @reduct_rs: Soulmates? Man on Subway Is Wearing a Conference Shirt For a Programming Language You’ve Heard Of 2021-05-31 17:02:22 RT @reduct_rs: Anticapitalist Icon? This Woman Forgot To Include A LICENSE File 2021-05-31 17:01:45 RT @reduct_rs: How to Love Yourself, Even the Part of You That Enjoys Maintaining Legacy Code 2021-05-31 17:01:02 RT @Reductress: How to Have a Hot Girl Summer Even Though the Litter Box Is in Your Bedroom: https://t.co/0tacWHcA0l https://t.co/cRbzwv81Gw 2021-05-31 16:58:45 RT @verge: Climate change responsible for about a third of heat deaths, study says https://t.co/i6YsEwe4rA https://t.co/wUvd9WEEJl 2021-05-31 16:47:03 The majority of research directors that I know are not power cognizant enough to realize their abilities to have this strong of an impact in this kind of situation, even when their intentions are good 2021-05-31 16:44:35 RT @interpretantion: as long as you made other people's lives better you didn't "miss out" on anything 2021-05-31 16:42:03 We need to think about systemic change that decouples incentives for dealing with harassment and abuse from funding IMO, rather than relying on people with funding being good actors 2021-05-31 16:38:07 Probably unpopular take: That Michigan clearly reacted to this is good in the short term, but also indicates a huge problem. This much power should not be concentrated into the hands of research directors at a few companies to begin with. https://t.co/Cjk1qxcPmI 2021-05-31 16:32:27 RT @Dev14e: "WE SHOULD TOTALLY MAKE IT A HOT GRILL SUMMER" - me talking to my housemates about buying a grill so I can make proper-ish Tand… 2021-05-31 16:30:30 @tobycmurray Honestly can we just cut the h-index crap in general 2021-05-31 16:30:11 @tobycmurray But yeah, the h-index of PC member incentive just incentivizes the broken system where senior faculty farm out reviews to students, rather than large PCs with junior faculty and senior student engagement 2021-05-31 16:27:59 @tobycmurray I see that, I'm just saying the whole thing is kind of nonsensical, it's students who drive most of the work in the community, with established faculty managing and getting funding. Why are we measuring stuff about the faculty at all 2021-05-31 16:22:57 Like at that point it's a benefit to the conference if some established advisor slaps his name on a bunch of papers. That's nonsensical 2021-05-31 16:21:44 I mean. The thing about established researcher h-indexes for people submitting is also bizarre. Are we still pretending that in CS, it's faculty who do the work and not the students they/we advise to begin with? https://t.co/LAuBQhCHAV 2021-05-31 16:14:32 RT @rupakmajumdar: I find the decision process of CORE deeply disturbing. By considering high acceptance rates and lower h-indices of PC me… 2021-05-31 15:47:43 RT @techpractical: I know when I want to protect someone's mental health, i force them into doing what's damaging it and slap a 15 grand fi… 2021-05-31 07:40:19 The house nextdoor to mine is having the most annoying Woo Girl party. The kind where every 30 seconds some girl yells "woo!!!!!!!!!" Why though, like why can't you get drunk without screaming, I don't get it 2021-05-31 07:24:52 How on earth is literally using gas chambers not "cruel and unusual punishment"? You all ever watched movies and read stories of the Holocaust with women slowly suffocating to death over 20 minutes holding up their babies in the air? No? 2021-05-31 07:21:34 @PensiveTM @MikeKofiA @nohomelikeplace Death penalty should be illegal, death penalty by GAS CHAMBERS should sure as fuck be illegal 2021-05-31 07:17:51 Americans: "the Israelis are the Nazis now" Also Americans: literally using Zyklon B to execute prisoners 2021-05-31 07:12:06 RT @MikeKofiA: @nohomelikeplace Really disturbing stuff https://t.co/OFxJrDjcSd 2021-05-31 07:04:38 RT @kamatsu8: @TaliaRinger This is an interesting read https://t.co/wIavaHL0hi 2021-05-31 06:47:15 RT @Istanbultelaviv: So what will it do: first it is will for the 4th time bring an Arab as a minister, Issawi Freij from Meretz party. Wit… 2021-05-31 06:11:34 @yudapearl I'm pretty sure they included the Israeli kids too though 2021-05-31 06:02:15 Sometimes the answer isn't, "we should do better science," it's "we should stop pretending to be science until we have put in the many years of work required to measure any meaningful quantitative effect at all" 2021-05-31 06:00:59 This is a problem I see everywhere, including in PL. Also famously SE with test-based program repair for a while 2021-05-31 06:00:08 There's this problem where we look at papers with numbers on benchmarks as more authoritative but it's often not actually as useful as looking at the approach and the qualitative results, because often the benchmarks are just bad 2021-05-31 05:56:44 As an example, I haven't seen a benchmark suite I am happy with when it comes to proof synthesis via machine learning tools. None of the suites currently capture the kinds of realistic tasks that people need help with during proof development. (REPLICA tries but it's too small) 2021-05-31 05:54:24 Thinking of good benchmarks often takes years and years of work, and in the meantime it makes sense to do qualitative evaluations, lest you derive false confidence from bad benchmark suites IMO 2021-05-31 05:53:06 Tangentially related to this: a bad benchmark suite is way worse than no benchmark suite https://t.co/4mEUjb3cNa 2021-05-31 05:48:56 RT @Rainmaker1973: Eastern Hokkaido is amazingly gorgeous with its breathtaking flower parks and gardens in spring. Higashimokoto Shibazaku… 2021-05-31 05:47:19 @avt_im I also don't feel like, say, HCI has this problem? Maybe I'm wrong, though, as that's not my field 2021-05-31 05:45:34 @avt_im If quality starts to go out the window and that kind of low level fraud becomes normal, that's a huge problem. We know what happens when AI overpromises and underdelivers. That has happened twice in history already. 2021-05-31 05:43:44 @avt_im ML right now has way too fast of a publication rate at the cost of rigor and integrity for the median publication. Like one publication per year would be reasonable for students in most CS fields. All of the ML numbers are super inflated, I have trouble believing it's necessary 2021-05-31 04:39:46 https://t.co/KxDfnObjIp 2021-05-31 04:38:46 It kind of feels like I'll be fine no matter what bullshit life throws at me 2021-05-31 04:38:27 I think this is because I set out to achieve a dream of mine and did it? And also rebuilt a very enjoyable life after an abusive relationship and breakup, even rebuilding through a pandemic 2021-05-31 04:36:30 RT @noamginger: @maliknir מה שחשוב כרגע זה להתמודד עם האיום האירני ועם זה שהילד חזר רעב מהבריכה. ומשילוּת. משילות זה גם חשוב. אין לנו זמן ל… 2021-05-31 04:36:05 RT @noamginger: ירדתי עם הילד לבריכה 2021-05-31 04:34:39 I feel very self-assured for the first time in my life 2021-05-31 04:01:21 @nikitab Yeah that's the neighborhood. Vibe I get is nice residential area, but you need to drive for coffee/groceries/work/fun. Is that right? Is it a good place to run? 2021-05-31 03:55:27 I unfortunately was taking in so much information during second visits that I don't remember neighborhoods very well 2021-05-31 03:54:22 @sibinmohan @sarielhp @socrates1024 @peztronic @ian5v @mraginsky etc 2021-05-31 03:49:23 Hi Urbana-Champaign friends. I really like living in houses of 4-5 people and have found a good fit, but it's at the edge of Champaign, like the very west side of Champaign. If I don't mind driving, is this a nice place to live? 2021-05-31 02:59:51 @krismicinski @sibinmohan @Dev14e @mynkgoel Idk, I feel like if even one student has a family member die and doesn't feel comfortable asking for an extension, that is much worse than a few people failing because of poor planning 2021-05-31 02:58:57 @krismicinski @sibinmohan @Dev14e @mynkgoel Ah, doesn't sound like a disaster then. What did you have to compromise? 2021-05-31 02:56:57 @krismicinski @sibinmohan @Dev14e @mynkgoel I don't know if it's a bad decision. It depends who's failing, in my view. 2021-05-31 02:55:47 @krismicinski @sibinmohan @Dev14e @mynkgoel Weird. What year students are these? 2021-05-31 02:54:01 @krismicinski @sibinmohan @Dev14e @mynkgoel Not even approximate deadlines as guidance? 2021-05-31 02:53:27 @krismicinski @sibinmohan @Dev14e @mynkgoel The key is to make it easy for those students even if they don't feel comfortable asking. If that means that someone who doesn't want to push themselves won't push themselves, I view that as an acceptable cost 2021-05-31 02:52:15 @krismicinski @sibinmohan @Dev14e @mynkgoel If you're not seeing those cases, probably students don't feel comfortable asking for them. I definitely asked for extensions around exceptional life events 2021-05-31 02:49:30 @krismicinski @sibinmohan @Dev14e @mynkgoel Especially if babysitting comes at the cost of students who really want to do well but have exceptional circumstances 2021-05-31 02:48:23 @krismicinski @sibinmohan @Dev14e @mynkgoel I guess I just feel like my job is to help students who need flexibility and want to learn, not babysit them 2021-05-31 02:47:23 @joepolitz @OptimistsInc @sibinmohan @Dev14e @mynkgoel Yeah that sounds worth trying 2021-05-31 02:45:30 @krismicinski @sibinmohan @Dev14e @mynkgoel If you see like 3 assignments that are late that's probably a good time to email the student like "hey are you doing OK? What's going on? Are you struggling with anything? If so that's normal and I'm happy to help" 2021-05-31 02:43:49 @krismicinski @sibinmohan @Dev14e @mynkgoel I don't get it. If someone just doesn't do the work should they not fail? That's like the prefect case for failing, and the students' responsibility 2021-05-31 02:42:37 @avt_im You can view CS as a whole as a larger field, and I don't think other research areas have this problem so pervasively. 2021-05-31 02:41:53 @avt_im I mean, why though? Why does an expert reviewer let that through? 2021-05-31 02:37:05 @OptimistsInc @sibinmohan @Dev14e @mynkgoel Which is roughly how I deal with collaborations by default. And committees. So seems fine 2021-05-31 02:36:21 @OptimistsInc @sibinmohan @Dev14e @mynkgoel Yeah I think I'd probably be like "here is an approximate deadline. Be late if you have a reasonable reason too, and I won't ask if it happens once or twice by a few days. If it happens every single time, or is more than a week late, I'll ask what's going on" 2021-05-31 02:33:25 @OptimistsInc @sibinmohan @Dev14e @mynkgoel I wonder if this effect goes away at all if you just make it a very impersonal form, and communicate that it's expected you'll sometimes be late, just if it becomes a pattern I might ask what's up 2021-05-31 02:31:40 @SMT_Solvers @ezyang I mean with JSCoq this is somewhat feasible lol 2021-05-31 02:28:51 @ezyang And no I'm not going to do this, it sounds like infrastructure hell. But one can dream 2021-05-31 02:28:37 @ezyang Hmmm, what if: - virtual env hosted online - interactive env to make quick fixes - deployment system to virtual env for testing 2021-05-31 02:19:55 @johnregehr fair 2021-05-31 01:56:22 RT @noamginger: הורדתי את הכלב 2021-05-31 01:56:20 RT @noamginger: רגע אני שותה קפה 2021-05-31 01:56:18 RT @noamginger: סליחה התקלחתי 2021-05-31 01:51:25 @sibinmohan @Dev14e @mynkgoel FWIW I think conferences should do this too. Our obsession with making up weird absolute cutoff dates and forcing people to make them is just kind of unhealthy 2021-05-31 01:50:36 @sibinmohan @Dev14e @mynkgoel I'd just be down for giving extensions to basically anyone who asks, and also making the process of asking not super personal, like just some form you fill out. It'd be clear if there's a pattern of abuse over time, at which point I'd ask the student what's up 2021-05-31 01:48:07 @sibinmohan "I have to work the few days leading up to the deadline" should be a totally valid reason to turn in a project a little late IMO 2021-05-31 01:47:39 @sibinmohan Unless we're going to make higher education free, I think acknowledging that reality is good. Though really, I think all of the strangely arbitrary and strict deadlines we set are net harmful, and we should just be much looser about deadlines in general 2021-05-31 01:46:55 @sibinmohan Monday deadlines can be good for students who have to work to support their education during the week IMO. For some people, the weekend is the only time homework can actually happen 2021-05-31 01:14:24 @dlowd @emeryberger Also, if you break down NeurIPS, this problem won't exist at this scale by definition. Just saying 2021-05-31 01:09:22 @dlowd @emeryberger I did fairly well building a reputation with small community conferences (CPP/ITP), even though my PLDI publication at the end did help boost it. I think you can build ranking systems that acknowledge that reality, too. 2021-05-31 01:07:36 @mdekstrand Yeah Artifact Eval is amazing. Rarely catches bad things, but I think what it does is make it extremely normative to qualify your claims by code, document the code, make it very transparent, and distribute it with the paper so that you can check all claims easily. 2021-05-31 01:06:14 @hillelogram Lmao 2021-05-31 01:05:12 @avt_im Shouldn't an expert reviewer at a small enough conference know to ask questions like, "did you try a different seed? What happened? What happens if you try using this data instead?" and so on, in a rebuttal period? 2021-05-31 01:03:20 @dlowd @emeryberger In particular, what made CSRankings authoritative is that it sucks less than other ranking systems. So you, too can make a ranking system that sucks less. 2021-05-31 01:02:06 @dlowd Yeah, this is a common criticism against CSRankings that I know @emeryberger is aware of. His response is usually that, well, this is a thing he threw together. If you want to use a different metric that incentivizes small conferences, fork it and do your own rankings. 2021-05-31 00:42:14 Also is it true that the ML community doesn't have something like artifact evaluation? If so, that's a huge mistake 2021-05-31 00:35:34 Aside from that, I've heard that at this scale, conferences don't really benefit students much anyways. Like, they're so big that many students I know tend to just talk to their peers from their lab. 2021-05-31 00:31:14 @lexi_lambda @samth @johnregehr Yeah it's meant to criticize the field as a whole and point to small-scale fraud. What I don't like is how much it normalizes small-scale fraud, which shouldn't be happening to begin with. 2021-05-31 00:29:16 If the ML community has this much of a problem with reviewing, IMO it needs to break down into smaller subcommunities for the sake of scientific integrity. I'm usually a fan of merging over splitting, but it seems like peer review at that scale is just not possible. 2021-05-31 00:18:32 @samth @johnregehr I'd avoid being too publicly negative about a student's statement on social media without offering constructive ways to improve it, just because of the power dynamic of publicly criticizing student work as faculty 2021-05-30 23:50:41 @johnregehr And then this shouldn't punish the reviewer. It should just point to things the next chair can build systems around. Even then, don't want to make reviewers too suspicious because of malicious papers, so this should be very occasional IMO 2021-05-30 23:49:32 @johnregehr I empathize a bit, but think this is where things like responsible disclosure and security ethics come in. We should have people fuzzing the review process once in a while, probably, but disclosing that this is happening to the chair, and retracting if accepted. 2021-05-30 23:47:08 @palvaro @samth Yeah, I think at that point, breaking down the field a bit becomes prudent for the sake of scientific integrity. 2021-05-30 23:14:46 @palvaro Yeah, I mean I believe the student has been a part of a culture like this, and is frustrated. I'm surprised that papers like this can regularly get past peer review. That shouldn't happen. Part of being an expert reviewer is knowing questions to ask to rule out this kind of thing 2021-05-30 23:09:21 But I'd be shocked if a proof assistant paper got past me with that kind of nonsense. Maybe I'll eat my words some day! 2021-05-30 23:08:38 With that in mind, I've heard that the ML community has a lot of problems finding enough expert reviewers given the number of submissions, so maybe that is the issue that makes small-scale fraud common? 2021-05-30 23:05:14 If these things are widespread, that's already a huge problem. Maybe that's the point of the post, I don't know. But I'd actually be surprised if this were totally normal? It is usually quite transparent when authors do this, and easy to call out as an expert reviewer. 2021-05-30 23:03:36 I don't want to be too critical of this since a student wrote it, and I get the point. But also, I don't think normalizing things like cherry-picking is a good idea. I always try to be very explicit about limitations. Also, my future work sections are real things I intend to do. https://t.co/dEEy6EWpRI 2021-05-30 22:25:17 (Like all interesting problems, there is an undecidable part to it. Here, one undecidable part is reifying all of the implicit reduction rules like iota. But I think maybe if you instrument the proof assistant to track those, that part might become decidable!) 2021-05-30 22:23:01 RT @jonmsterling: My tikz ch'i is getting scarily high... LOL https://t.co/zvwf2IQkAH 2021-05-30 22:21:16 @thatinterlace Absolutely! But not every possible change is handled usefully by a tool yet. (I'll add a section to my thesis next week about the kinds of repair I don't handle yet, but would like to.) 2021-05-30 22:16:13 Any equivalence that you prove must contain enough information to describe how to jump across the equivalence, including how to transform the equality reduction rules like iota. That's absolutely wild to me. 2021-05-30 22:14:06 There's a somewhat beautiful thing that falls out of this too, if you think about it. Which is that equivalences are actually very deeply tied to equality, _even when_ your language doesn't make that connection explicit. 2021-05-30 22:03:28 The result is a powerful proof repair tool that doesn't rely on univalence, but that supports any change that can be described by a type equivalence. Paper: https://t.co/ot12Kr0wwP Thesis draft: https://t.co/XlBfGkP8QZ PLDI 5 minute talk: https://t.co/p3P8i1vSNU 2021-05-30 22:01:43 And we can do this for any equivalence 2021-05-30 22:01:05 Iota shows up kind of hidden, in the proof that the diagram commutes. But if we reify it explicitly, even when it's definitional, we can actually just port it directly from unary to binary natural numbers, like we would with anything else, just jumping across the equivalence. 2021-05-30 22:00:15 In other words, definitional iota becomes propositional iota. 2021-05-30 21:59:32 So imagine the shape of your inductive type changes! If you're jumping across the equivalence, skipping those functions, you'll run into a problem: things that hold by iota that the type checker could just check for you on the left side, now on the right side you must prove! 2021-05-30 21:58:25 One of the reduction rules that the type checker runs for definitional equality checking is this rule called "iota." How this works is intimately tied to the shape of the inductive type. 2021-05-30 21:57:02 In other words, the type checker figures out definitional equalities for you automatically, but you have to show propositional equalities yourself. 2021-05-30 21:56:43 Two things are definitionally equal if they reduce to the same normal form 2021-05-30 21:55:51 This is because the language I'm in is what is called "intensional": it distinguishes between "definitional" and "propositional" equalities. 2021-05-30 21:54:55 OK, but doing this turns out to be not entirely straightforward. The biggest challenge occurs when you see an inductive datatype totally change structure, like with unary and binary natural numbers above. 2021-05-30 21:54:14 You can then kind of jump across the equivalence directly without ever calling the functions that convert back and forth. This works for repair, and it doesn't rely on univalence. Though you do lose out on nice information you'd get if you were in a language with univalence! 2021-05-30 21:52:58 The trick is bascially to split up equivalences between A and B into things that talk about A, and things that talk about B. You can always do this thanks to this theorem called Lambek's theorem, and what you get is equally expressive. 2021-05-30 21:51:40 This gets around that issue, but it's really specific, like it just handles unary and binary. It turns out, though, that you can do this for any two equivalent datatypes: https://t.co/ot12Kr0wwP 2021-05-30 21:50:41 OK so back in 2000, like before univalence even existed, two smart people actually defined a tool that transforms between unary and binary: https://t.co/mu3oWDhhpf 2021-05-30 21:49:40 The way this is usually done in languages that have univalence, using a function called "transport," doesn't let you do that. 2021-05-30 21:48:46 It gets worse, though---I'm not just interested in deriving equivalent functions or proofs, but rather I'm interested in repairing them. That is, I want to be able to totally replace unary numbers with binary numbers 2021-05-30 21:48:12 But the language I'm in doesn't let me do this in general by just calling a function---for that, I'd need this property of my language called "univalence." 2021-05-30 21:47:04 This means that for any function or proof you write about unary natural numbers, you can an equivalent function or proof about binary natural numbers, and similarly in the opposite direction. That's cute, in theory. 2021-05-30 21:46:26 Unary and binary natural numbers are equally expressive. So you can write two functions (here "update" and "revert") that get you back and forth between them, and are mutual inverses. That is called a "type equivalence" between them. 2021-05-30 21:45:45 Explanation: Datatypes can be defined inductively in this fancy language that I like. On the left is the inductive definition of unary natural numbers 2021-05-30 21:39:21 iota is bae https://t.co/vqMrITgrlE 2021-05-30 18:34:02 RT @Yair_Rosenberg: If you are wondering what on earth is going on right now in Israeli politics, what the plan is to replace Netanyahu, an… 2021-05-30 18:29:27 RT @certifiablyrand: @aanand_03 @TaliaRinger @JeffDean It's still online but I don't have a hidden link to it from my homepage anymore: ht… 2021-05-30 18:00:35 RT @hourlyFox: https://t.co/JxLgdgHmm1 2021-05-30 17:59:49 RT @arash_tehran: Also this, of course, won't be an overall right-wing gov. Of the base of the gov in Knesset, roughly one third of it will… 2021-05-30 17:59:08 RT @AnshelPfeffer: “Netanyahu is trying to take the entire state of Israel to his personal Masada” was Bennett’s key line. I wonder who cam… 2021-05-30 17:48:19 @rob_rix I'm changes in shape!! 2021-05-30 17:33:18 Tag yourself I'm shape and change https://t.co/wlfjihRK5L 2021-05-30 17:25:48 I watched the whole thing. Highly recommended. https://t.co/byqWxrnKKf 2021-05-30 17:16:20 RT @AdalwinAmillion: Therapist: Linear Mandarin is not real, it cannot hurt you. Linear Mandarin: https://t.co/2Ro09KhpJv 2021-05-30 16:03:52 RT @HilaCodes: Andrew Wakefield is bad, actually*. This video has been in the works for so long that I'm actually shocked it's finally out… 2021-05-30 07:47:03 RT @mariana057: Today my wife and I met our new neighbor who invented the desk globe. Small world! 2021-05-30 07:18:25 @ev_maus Glad! 2021-05-30 07:07:39 @aanand_03 @JeffDean @certifiablyrand The article is definitely a joke, so I wouldn't take anything it says seriously. I don't know the context, I just found it when looking at Robert Rand's profile. Note how he gave himself an Erdos number of 1 2021-05-30 06:53:19 @aanand_03 @JeffDean Dude @certifiablyrand this is gold https://t.co/HXVteVcXXi 2021-05-30 06:49:48 @aanand_03 @JeffDean They seen to struggle a lot with theses, for whatever reason. https://t.co/2P2wXsR888 2021-05-30 06:46:11 @andrejbauer I like Twitter for the social interaction, research conversations, community news, activist organization, and so on. Just also need a balance where every so often I'm just getting puppy photos and wholesome memes to take a break 2021-05-30 06:40:28 @ev_maus I wouldn't say wholesome though, I get politically pretty intense! 2021-05-30 06:30:36 RT @AlbertQJiang: @TaliaRinger https://t.co/aFEmoIMzPs 2021-05-30 06:30:22 @AlbertQJiang My favorite author, T. under Construction 2021-05-30 06:29:00 RT @mustach_io: @TaliaRinger Did you know that although my invention is not as well cited as dr dan’s angels, I invented the Bacon Cooker w… 2021-05-30 04:29:35 Please don't fix the Hell's Angels bug that's our favorite funny mistake and Dan wants to keep it 2021-05-30 04:28:17 @JeffDean Thanks!! 2021-05-30 02:07:58 https://t.co/EVYbSfRhWd 2021-05-30 01:45:37 @jennplusplus Insurance nonsense 2021-05-30 01:38:25 Just went through the giant pile of mail in my room I'd been procrastinating on because it's overwhelming. Like six copies each of the same bill from my doctor's office. If they'd sent it just once, I would have paid it weeks ago 2021-05-30 01:37:46 I wish billing offices understood that sending me multiple copies of a bill or calling me and leaving messages just means I'll get to the bill more slowly, since now I have more things to sort through, and this is all ADHD stuff 2021-05-30 01:26:39 @johnregehr Is there real pizza out there and not the abomination that is deep dish? @hillelogram? 2021-05-30 01:24:35 Finally unpacked from my second visit three weeks ago 2021-05-30 01:15:49 @glassEyeballs https://t.co/pBFmN01laW 2021-05-30 01:05:11 What should I order for dinner 2021-05-30 00:59:59 @dlowd Neurosymbolic Confusion 2021-05-30 00:58:33 RT @dlowd: @TaliaRinger “The PIs are uniquely qualified to complete this project, primarily because they don’t understand it at all.” 2021-05-30 00:58:28 RT @dlowd: @TaliaRinger “In this project, we propose to combine ML and PL in a novel way — specifically, we exploit the end-to-end differen… 2021-05-30 00:58:26 RT @dlowd: @TaliaRinger “The resulting system, Deep Confusion, exhibits a human-like ability to be totally perplexed.” 2021-05-30 00:28:05 So being a type theorist really isn't about being smarter IMO. It's about being more okay with feeling confused. 2021-05-30 00:25:51 It's just more than okay to feel clueless and confused. It's extremely normal. If you don't, you're probably missing something that should have confused you. 2021-05-30 00:25:05 I talk to more people, and they send me more papers and talks, explaining the same things through different lenses. Over time I start to understand what the hell was going on say, 3 years ago. But I never know what the hell is going on in the moment. 2021-05-30 00:24:07 The second time around, I find I understand a lot more than I had thought I'd understood. In fact, subconsciously, I'd used concepts from this paper in my own work without even realizing it! The intuition had burrowed into my mind somehow. 2021-05-30 00:23:03 I let those papers sit a bit, mull over them, let my mind wander. Sleep. Send a few emails to type theorists asking them questions. Go back to my work. Maybe two months later I'm thinking about type theory and I'm like "that reminds me of that POPL paper" so I pick it up again. 2021-05-30 00:22:07 So when I read a modern type theory paper, I almost never expect to understand every part of it. I do my best to get the intuition, the high-level takeaways, the implications. I read the core technical sections to the best of my ability, but expect I won't fully get them. 2021-05-30 00:16:53 I've found that getting into type theory work means letting go of my desire to understand everything I read or do. Accepting that feeling totally clueless is normal. Over time you become less clueless, but if you ever fully understood it, it wouldn't be theoretical research. 2021-05-30 00:15:09 But theory in motion is always hard. You're confused because type theorists are also confused. The papers are hard to read because people are still figuring out what the hell they're doing and how to best explain it. 2021-05-30 00:14:09 In 20 years or so, once definitions have converged reasonably, and edge cases have been worked out, you might look back and be like, "why was I so confused? This isn't so hard." 2021-05-30 00:13:10 I think one of the most common mistakes for people learning state of the art type theory is thinking it's their fault when they don't understand everything. I promise, even the experts don't really know what they're doing. That's what makes it state of the art theoretical work. 2021-05-30 00:10:38 RT @mer__edith: This. Including the colleagues who suddenly stop replying. The people whose work engages "justice" who politely explain tha… 2021-05-30 00:06:24 RT @ilya_poet: Arseny Tarkovsky (yes, father of the filmmaker) whose bday is this week, was a younger friend of Tsvetaeva and she dedicated… 2021-05-30 00:04:50 Basically anything that increases faith in humanity 2021-05-30 00:04:17 I like activist content, but I'm getting a little depressed recently, and I think I'd benefit from a balance with a bit more wholesome content. What wholesome accounts do you recommend following on Twitter? 2021-05-30 00:03:13 RT @EM_RESUS: Normalize kindness for no reason 2021-05-30 00:02:03 RT @hourlyFox: https://t.co/MwKfUYZaKk 2021-05-29 23:58:42 @johnregehr Must have been an arxiv preprint 2021-05-29 23:56:44 Dan's #3 publication is the best though https://t.co/dKE2YY5TTr 2021-05-29 22:36:49 RT @schemeprincess: Reflecting on my experience, can I say that I have never in my life encountered as much sexism and bullying as when I w… 2021-05-29 22:36:19 @schemeprincess Do you have confidential advocate systems yet? 2021-05-29 22:35:49 RT @schemeprincess: Coincidentally, last week was also our annual review period at @MSFTResearch, so I committed to enacting concrete chang… 2021-05-29 22:15:19 RT @16kbps: it would be nice to be able to express constraints like "list of length > 2021-05-29 22:11:54 RT @16kbps: because type annotations are not evaluated, there's nothing stopping anyone from writing a dependent type system for python 2021-05-29 22:09:28 @BjornerNikolaj I can say with confidence that my advisor has not written any of my code, or any section of any of my papers lol 2021-05-29 21:45:11 Google Scholar thinks my advisor wrote my thesis lol https://t.co/ceOCi88CQS 2021-05-29 21:25:18 RT @arash_tehran: The best news for Israeli left, progressives, pro-democracy and pro-peace/anti-occupation camp will be that the no. 1 on… 2021-05-29 21:24:53 RT @arash_tehran: The 'gov of change,' if it ever comes to be, will likely last less than a year.Its main job will be to usher Israel into… 2021-05-29 20:32:38 Oh man I'm going to need this in the future https://t.co/CwsHJltWXY 2021-05-29 20:07:09 Same though https://t.co/4NuzjwRJEF 2021-05-29 19:41:26 RT @patbaumgartner: This is so true and worries me very much! 2021-05-29 19:40:11 RT @DrCABerry: #FullProfessor #RedTamAchievementUnlocked #NoireSTEMinist https://t.co/hnVl8twYGz 2021-05-29 19:27:53 They weren’t even excited, they seemed moderately annoyed, like "we don't really get much out of this, people just don't like chess players wasting food after matches" 2021-05-29 19:26:28 Slept 12 hours and dreamed a lot. My favorite dream was of a restaurant where famous chess players would come and bring you their leftovers. They'd explain both the food and the chess game. I'm so confused 2021-05-29 07:22:37 RT @azhir_io: New kitten alert. https://t.co/6VMqPC2wRe 2021-05-29 07:11:55 RT @Haaretz: הם שומעים, זוכרים ואפילו מרכלים: מבט לעולמם הנסתר של הצמחים https://t.co/5tk4XWEzRT 2021-05-29 06:52:32 RT @maxvonhippel: On May 29 @carloangiuli will be presenting "Homotopy Type Theory 101" to the #BostonComputationClub (https://t.co/TwCscKU… 2021-05-29 06:52:09 RT @carloangiuli: @TaliaRinger @ev_maus @jonmsterling I’m actually giving a talk tomorrow that sounds like it would be helpful to you! I pl… 2021-05-29 06:50:12 Please save this Tweet for post nuclear apocalypse dark humor if the internet still exists 2021-05-29 06:47:58 RT @barisanhantu: For the same reason, people might not want to seek a professional diagnosis. In many countries, the downsides of diagnosi… 2021-05-29 06:05:12 Can we just hit the reset button sometime and like start from scratch with open borders, that would be nice 2021-05-29 06:03:45 This Twitter thing makes it too easy for people to catch only parts of conversations, which too easily leads to painful situations. Sigh 2021-05-29 05:56:20 @hamza0867_ben @iyad_elbaghdadi @yawaramin OK, sorry if I said anything that hurt you 2021-05-29 05:40:19 @hamza0867_ben @iyad_elbaghdadi @yawaramin Oh OK, I'm glad to hear that. Yeah sorry, it's just really painful for me to talk about that 2021-05-29 05:39:44 @hamza0867_ben @iyad_elbaghdadi @yawaramin It's just helpful to not do that, that's all 2021-05-29 05:38:12 @hamza0867_ben @iyad_elbaghdadi @yawaramin Anyways, you'll find I won't excuse much. I find the whole thing thoroughly depressing. I also want to fucking die every time someone erases the history of my family's trauma though, or paints them in the same way as the British, who acted deliberately from a place of power 2021-05-29 05:36:42 @hamza0867_ben @iyad_elbaghdadi @yawaramin The British were like "oh cool though let's pretend to help the Jews by helping this fringe national movement so we can break up the Ottoman Empire, draw some ethnic lines, create eternal conflict, and run away" 2021-05-29 05:35:47 @hamza0867_ben @iyad_elbaghdadi @yawaramin Imagine a world where we never placed any limit on refugee resettlement? Israel probably wouldn't have happened, honestly. 2021-05-29 05:35:10 @hamza0867_ben @iyad_elbaghdadi @yawaramin The Nazis, which is why Ashkenazi Jews definitely don't want to go back to Europe. I don't know what there is to love about this. It's depressing. The Zionist groups were the only people willing to take in homeless Jewish Holocaust refugees. 2021-05-29 05:30:47 @hamza0867_ben @iyad_elbaghdadi @yawaramin If there were a place to go back to, sure, maybe. 2021-05-29 05:29:47 @hamza0867_ben @iyad_elbaghdadi @yawaramin I don't know, but my grandparents were homeless Holocaust refugees, they didn't have a place to "go back" to. 2021-05-29 05:28:19 @iyad_elbaghdadi @hamza0867_ben @yawaramin And Netanyahu should fucking rot in prison it's just the literally advocating for Jews to get out of the region thing that isn't OK coming from anyone. One democratic state, one day 2021-05-29 05:27:22 @iyad_elbaghdadi @hamza0867_ben @yawaramin "go back to where you came from" I'm not really OK with that coming from anyone and I don't think I should be. 2021-05-29 05:25:10 @iyad_elbaghdadi @hamza0867_ben @yawaramin Not yours. Yours is fine. This one: https://t.co/w8D43KXTWr 2021-05-29 05:19:10 @nbhansen @CoqLang I recommend chapter 2 of my thesis (draft): https://t.co/XlBfGkP8QZ 2021-05-29 05:17:05 @nbhansen @CoqLang But, there are actually two languages inside of the Coq system. Here I show one of the two languages, the one that feels more like a conventional programming language, just with some cool extra features. 2021-05-29 05:16:03 @nbhansen Here: @CoqLang 2021-05-29 05:13:26 Broke: syntax highlighting Woke: semantics highlighting 2021-05-29 05:05:46 @landay 1PM :) 2021-05-29 05:04:23 @nbhansen What you end up getting from this is the ability to say very concisely "my function takes two lists of length n, and returns a list of length n" and have the compiler enforce this for you, and return a proof back to the user in the end! 2021-05-29 05:03:20 @nbhansen Oh, you can do super wild things in this language if you want. Like, what if you stick the length of a list in the type itself? Like this? https://t.co/9acJiz1ZoD 2021-05-29 05:02:09 @nbhansen (Also, thanks. Sorry, just couldn't resist a chance to explain inductive types.) 2021-05-29 05:01:27 @nbhansen Here's the type of natural numbers and the corresponding induction principle, for reference. "S" here means "successor." So you can basically read it as "1 + n." https://t.co/qGYOghlPEi 2021-05-29 04:53:31 Impromptu classes Talia teaches on Twitter https://t.co/QLCGHY24xh 2021-05-29 04:53:10 @nbhansen Everything I highlight here just aligns the base and inductive cases with the same colors and superscripts, so that you can quickly see which parts of each term correspond to each other. 2021-05-29 04:52:26 @nbhansen (I super prefer the other way around, where the induction principles are "primitive"---encoded directly in the language---but YMMV.) 2021-05-29 04:52:02 @nbhansen In fact, any language usually encodes just one of them, and implements the other as syntactic sugar. Here, the induction principle list_rect used on the bottom right is actually just a constant that unfolds to a function that looks like the thing on the bottom left. 2021-05-29 04:50:24 @nbhansen With a clever enough translation, these two things actually turn out to be equally expressive---though there are some caveats, like that arbitrary recursion is disallowed. You must show that your recursive function terminates. (There are tricks to reason about nontermination.) 2021-05-29 04:49:19 @nbhansen OK, so in this language, there are two ways to write functions about your list datatype now, like the length function. You can write your function: 1. by breaking into two cases, one for nil and one for cons (left), or 2. using the induction principle (right). 2021-05-29 04:47:31 @nbhansen (There's a beautiful and deep reason for that!) 2021-05-29 04:47:13 @nbhansen It will return a proof that P holds for any possible list. This is a lot like the induction principle for natural numbers that you're probably used to, just moving from numbers to lists. 2021-05-29 04:46:26 @nbhansen This function says that, for any property P about lists, if you pass it two arguments: 1. base case: a proof that P nil holds , and 2. inductive case: a proof that for any element and list, if P holds for the list, then P still holds when you cons that element in front, then: 2021-05-29 04:44:17 @nbhansen From this datatype, in the fancy language I work in, this function called an induction principle is generated automatically. 2021-05-29 04:43:22 @nbhansen The thing in the top left is a datatype that defines lists recursively: a list is either empty (nil) or the result of taking some element and sticking it in front of another list (cons). 2021-05-29 04:40:59 My brain does really intense structural parsing of proof terms, so this is kind of how I see things. Over time in my papers, I've been trying to use visual cues that mimic my internal experience in some way (though my internal experience doesn't use color 2021-05-29 04:38:06 Here's a pretty thing I did last minute in my thesis https://t.co/YJOTv3oXDI 2021-05-29 04:28:07 Academic culture is using em dashes for emphasis---like this---habitually in every day conversations. 2021-05-29 04:23:48 Still gloating about this, though https://t.co/jTW8Vowr1R 2021-05-29 03:18:06 2021-05-29 03:15:43 - my orientation being labeled and assumed based on who I am happen to be dating at the time---and same with whether or not I get to belong in the LGBT community 2021-05-29 03:15:07 - coming out to new friends or early dates and having them immediately ask me details about my sex life 2021-05-29 03:14:16 - people assuming I must be poly, even though I can't imagine ever feeling attracted to more than one person at the same time 2021-05-29 03:13:46 - coming out to partners and having them respond "that's so hot" yikes 2021-05-29 03:12:51 - coming out to people I love (gay-friendly people at that) and having them respond by telling me that I'm lying or going through a phase or confused 2021-05-29 03:11:20 - girls on my swim team refusing to change near me in high school in the locker room, despite all of the above caveats about how I'd never be attracted to them to begin with - men not taking my attraction to women seriously 2021-05-29 03:10:17 - girls on my swim team refusing to share a bed with me in high school, making me sleep on a hard cot, assuming I'd hit on them, even though they weren't attractive and I very rarely feel any kind of attraction to begin with (just when I do, it's intense and gender independent) 2021-05-29 03:08:02 - being fetishized by men - unicorn hunting - my orientation being viewed as inherently personal or sexual, despite the fact that I am demi, leaning toward ace 2021-05-29 03:05:47 Biphobia sucks. Some experiences: - being assaulted several times in very illegal ways specifically because men felt entitled to my body as a bi woman - being told over and over again I'm not real or I'm just making things up - being unwanted by women who assume I'll cheat 2021-05-29 03:01:41 I don't know who this person is, but teenage me would be in absolute tears to see the number of out bi people these days, given how many times I was told I don't actually exist https://t.co/XZccbGif4D 2021-05-29 02:57:16 @SMT_Solvers No way, peers often leak info that can lead to severe retaliation 2021-05-29 02:56:34 This seems to be the only other genre of lie I'm capable of telling, outside of ones that serve to protect people's lives. I can tell very, very innocent lies to protect the feelings of someone I very deeply love 2021-05-29 02:55:54 Now she thinks I love guava juice, and every time I visit her, she buys me guava juice for after my runs. I don't tell her 2021-05-29 02:55:07 When she asked, I told her that I had gone to the store to get guava juice because I like having juice after I run. I hid the shampoo from her. I didn't want her to feel really, really guilty for not buying shampoo, as I knew she would 2021-05-29 02:54:30 Very, very occasionally, like once a month or so, I tell a lie to help someone I love not feel as bad. For example, when visiting my grandma in Israel, she forgot to buy me shampoo. I didn't want to reveal to her that I bought shampoo, so I also bought guava juice 2021-05-29 02:45:44 Is this something neurotypicals do more easily? Like I can't stop myself from crying when I'm sad, so it makes sense that if a lie makes me feel guilty (does not serve some higher good like protecting someone else), that feeling will just overflow and the pain will spill out 2021-05-29 02:43:41 People who can just look other people in the eye and lie to them even when someone else's life or safety doesn't depend on it scare the shit out if me. Even small, harmless lies 2021-05-29 02:42:29 Like really, it feels too terrible to lie, I don't think I even remember the last time I did it 2021-05-29 02:41:41 I don't understand how people can tell even small lies unless someone's life or safety depends on it https://t.co/GqGfabZRgY 2021-05-29 02:31:20 Feel free to brainstorm ideas ITT. Safety should come before punishment, but punishment can be a form of safety for future students, if enacted in a way that does not cause retaliation. 2021-05-29 02:30:32 Build confidential advocate systems. Ensure confidential advocates are not mandatory reporters. Reform the formal complaint process that follows from that (I have fewer ideas for how to do this). Form peer groups of survivors to find solidarity and support in the community. 2021-05-29 02:28:27 Same thing with the SIGARCH/Florida case that culminated in suicide. Abusers are absolutely everywhere. At your university, in your department, there are probably many. Maybe they're nice to you. Maybe you'd never guess. Until it's safe to speak up, you'll never know. 2021-05-29 02:25:14 Or company. Or SIG. It's everywhere in CS, I promise. 2021-05-29 02:24:35 I keep seeing folks not at Michigan say "that's great, but now Michigan doesn't have to change the system." Michigan isn't unique, here. This is deep and systemic. If you are in a position of any power, use this opportunity to change the system at your own university. 2021-05-29 02:22:02 RT @niloufar_s: A (small) relief, but I also keep thinking about what we could have achieved if one serial abuser hadn't taken up SO MUCH e… 2021-05-29 02:21:17 RT @dmetaxak: THAT'S RIGHT. Now do the rest of them, and the system. https://t.co/AP9jBGaLdZ 2021-05-29 02:18:54 RT @timnitGebru: I can, with 100%, certainty, speaking from experience, say that saying something, whether its on your own or others' behal… 2021-05-29 02:12:28 @hamza0867_ben @yawaramin @iyad_elbaghdadi In fact, the realization of Israel _at all_ is Jews taking "go back where you came from" to its extreme, so it's probably generally not a good thing to advocate for 2021-05-29 02:11:07 @hamza0867_ben @yawaramin @iyad_elbaghdadi In any case, your pinned post is telling Israelis to "go back where they came from" which is exactly what Israelis are afraid of, being kicked out and having literally nowhere to go, because there isn't any country to go back to at all 2021-05-29 02:09:32 @hamza0867_ben @yawaramin @iyad_elbaghdadi Way, way better than the way most of Europe treated us, though! They just murdered us mostly 2021-05-29 02:08:59 @hamza0867_ben @yawaramin @iyad_elbaghdadi Yeah, that was a long time ago though, and Jews were in the last 50 years kicked out of all neighboring countries, which is why Israel has such a huge Mizrahi population. Also, despite thriving back then, we still had to pay a Jewish tax, be second class citizens etc 2021-05-29 01:14:12 @ev_maus @carloangiuli @jonmsterling Oh, cool! 2021-05-29 01:13:28 @ev_maus @carloangiuli @jonmsterling I think I'm the only weirdo using it for external automation right now, but I love it. I discuss the distinction in my thesis a bit 2021-05-29 01:10:58 @ev_maus @carloangiuli @jonmsterling (Plus it works for repair, which is crucial for me) 2021-05-29 01:10:38 @ev_maus @carloangiuli @jonmsterling You lose out on general guarantees that way, and you lose information in the final result, and you have to deal with some messy things to get it right. But otoh you can get much faster functions and smaller proofs this way for certain kinds of changes 2021-05-29 01:09:40 @ev_maus @carloangiuli @jonmsterling Externally, if you're in a type theory consistent with HoTT, you should always be able to take applications of that induction principle and view them as a rewrite system, and implement that as a transformation. This is kind of what chapter 4 of my thesis does 2021-05-29 01:08:16 @ev_maus @carloangiuli @jonmsterling Basically, when it's internal, by picking a super super smart definition of equality, you get an induction principle that lets you do really amazing rewrites generically and beautifully, even with dependent types. It's cute 2021-05-29 01:06:02 @ev_maus @carloangiuli @jonmsterling Internal := in the type theory itself External := in the "metatheory" basically, like you write programs that transform programs and the correctness statement of that transformation would be in HoTT, even though the programs it generates are not 2021-05-29 01:05:06 @ev_maus @carloangiuli @jonmsterling Its a transformation if you externalize it (I do in my thesis!), but the cool thing about HoTT is that you can literally implement a function that transports across equivalences, and that's super cool. From my perspective, it's useful whether internal or external 2021-05-29 01:02:54 RT @admitscut: @ev_maus @TaliaRinger @EgbertRijke ‘s course materials are good too https://t.co/8esCC0n8ff 2021-05-29 01:02:21 @ev_maus @carloangiuli @jonmsterling But I think I'm one of few people interested in that problem right now, though Jon is interested in helping so I have a lot more confidence I can do it now 2021-05-29 01:01:39 @ev_maus @carloangiuli @jonmsterling Really I think cubical type theory just nailed the definition of equality in an extremely beautiful way. But it's not just beautiful 2021-05-29 01:00:24 @ev_maus @carloangiuli @jonmsterling IMO absolutely, but it needs a lot more automation, which is something I'm hoping to help with soon 2021-05-29 00:58:28 @ev_maus @carloangiuli @jonmsterling Understanding the theory beneath those is a good start, though these days it's important to note that cubical type theory exists, so you can have your programs not get "stuck" on univalence 2021-05-29 00:56:41 @ev_maus @carloangiuli @jonmsterling I think in terms of applied homotopy type theory, the important things to understand are: - type equivalences (including the adjoint) - transport - why equality with dependent types is usually so hard in intensional type theory - why univalence makes it much nicer 2021-05-29 00:53:28 @ev_maus @carloangiuli @jonmsterling Haha not just that, a new foundation for mathematics. Quite an undertaking! 2021-05-29 00:50:57 @ev_maus @carloangiuli @jonmsterling And I still get confused a lot. I think with really deep and new type theory like this, though, it's important to accept sometimes feeling confused, because even the type theorists are still confused (still trying to solve some fundamental problems, agree on terminology, etc) 2021-05-29 00:49:33 @ev_maus @carloangiuli @jonmsterling (https://t.co/XlBfGkP8QZ) I do recommend trying to read the HoTT book to some degree. But at least for me, I found the concepts kind of---they were like very dense liquids that my brain slowly absorbed over 6 years 2021-05-29 00:47:22 @ev_maus @carloangiuli @jonmsterling I personally had a lot of trouble with the HoTT book. I acquired most of my HoTT knowledge from talking to type theorists. These days, there might be much better resources. My thesis draft has the high level "HoTT for engineers" explanation, but not much mathematical context 2021-05-29 00:45:34 RT @ev_maus: So, I've finally decided that since I liked homotopy theory as it related to algebraic topology (i.e. it's still some of the m… 2021-05-29 00:45:15 @ev_maus So, the people I trust most on this who are on Twitter are @carloangiuli and @jonmsterling 2021-05-29 00:40:05 @amyjko Yeah, I'm just going to wait a few years, see if I'm engaged or married, and if not just have kids myself, then figure out the rest later. At least ADHD takes some stress away by virtue of not being able to process plans on that large of a scale lol 2021-05-29 00:34:29 RT @shabberwocky: @blkgirllostkeys "if you cared about my time you wouldn't be late" I CARE SO MUCH PLEASE STOP I CAN'T HELP IT 2021-05-29 00:33:04 @amyjko Yeah, that makes it hard to find relationship stability though, especially since I want kids before then 2021-05-29 00:22:19 @amyjko I think the job insecurity leading up to tenure sounds scary, but I'm less worried after that. And honestly these days not super worried about tenure, though I imagine if I have major setbacks in the first few years, it will get very stressful 2021-05-29 00:17:15 @amyjko ouch 2021-05-29 00:03:44 Very non-exhaustive acknowledgments slides for defense next week https://t.co/yxXehnGSge 2021-05-28 23:00:58 @arvindsatya1 My entire international cohort of incoming faculty has a Slack channel, and I think we're all planning to stay in touch! 2021-05-28 22:54:36 RT @nikitab: @pvineetha @krismicinski @TaliaRinger I didn’t think to negotiate for one when I was hired, but perhaps I could get a retentio… 2021-05-28 22:33:51 @virtualcourtney I think I just need a therapist who aligns with me culturally, and views emotions as useful tools for changing the world that ought to correlate well with the external situation, rather than feeling good being a goal in itself, which I think is a bad idea for the world 2021-05-28 22:31:53 @nikitab I suppose in Urbana-Champaign, most therapists probably have worked with academics just by virtue of living in the area 2021-05-28 22:29:53 @virtualcourtney Often when I feel really bad, I don't want to focus on my internal response, I want to plan ways to improve systems so that no other person will have to go through what I went through in the future. I think the feelings are useful because they help me do that 2021-05-28 22:29:21 @virtualcourtney I haven't really found therapy to be helpful for anything work related even as a graduate student. Very helpful for other things, though. Talking to someone who treats academics may help. I think part of the problem is that a lot of American therapy is too internally focused 2021-05-28 22:28:15 I was falling asleep and went to get coffee with a friend, and ordered a honey oat milk latte, and they wrote "HOL" on it and I was falling asleep so I was like "how did they know I like higher-order logics" 2021-05-28 22:26:51 RT @jonmsterling: Having a lot of fun playing with Lean 4, which looks extremely promising! I found a bug and filed a ticket, though. https… 2021-05-28 22:26:36 RT @dlowd: It’s been a challenging week for many, so please enjoy this trolley non-problem. https://t.co/HYVYMNfE37 2021-05-28 22:23:43 @joepolitz hahahaha yeah I do have a mentor through this, but the relationship is a bit different 2021-05-28 22:22:26 @rg9119 Oh, interesting! Does it say anywhere on the website what specifically is captured by "diversity"? Just want to make sure I belong if I sign up and am not intruding 2021-05-28 22:18:42 Kris truck meme is best meme https://t.co/hrBjO1SDBu 2021-05-28 22:18:01 @srivoire @dlowd Yeah I've never really had a good experience with this. Useful for other life things like abuse, but for work they always seem profoundly confused about what the job entails 2021-05-28 21:36:44 @pbzcnepu I think the leadership has changed a lot 2021-05-28 21:31:10 @znmeb I'll have a few, but I don't know in what capacity they'll be comfortable with that kind of thing. It's just nice to be like, "oh man, the political situation really sucked last week, so I've been distracted and I've had trouble focusing" even just to get an affirmative nod 2021-05-28 21:29:32 Mentors? Peers? It always made me feel better to explain why I was tired or stressed to my advisor, I think that's the part of the relationship I can't replace with just myself, whereas I feel totally fine guiding my own technical plans for the future 2021-05-28 21:28:08 So when I start my faculty job, who do I talk to when I've had a stressful and tiring week and just want to explain what happened and how it impacted my work, to someone who actually understands my job (so not a therapist)? 2021-05-28 21:26:44 RT @schemeprincess: The letter can be found here: https://t.co/AO4FFBdj5P 2021-05-28 21:25:36 RT @schemeprincess: Everyone that has spoken out about the students at Michigan, the students have given permission for the rest of the CS… 2021-05-28 20:43:37 THE INVISIBLE LABOR you have NO IDEA https://t.co/97ioIj3u33 2021-05-28 20:43:04 RT @dmetaxak: 3. Be vocal 2021-05-28 20:42:58 RT @dmetaxak: 2. Fix the processes in your workplace so they actually do something. The last thing I ever wanted to do was describe where a… 2021-05-28 20:42:46 RT @dmetaxak: Now what? 1. If you've never experienced such abuse, reflect on that privilege—and the fact that, despite what this has cost… 2021-05-28 20:42:42 RT @dmetaxak: So where does this leave us? As anticipated, I've gained nothing, and the process has cost me time, energy, dignity. WL r… 2021-05-28 20:42:37 RT @dmetaxak: 3. The consequences range from meaningless to insufficient. @ACM will still allow him in our community after a few years away… 2021-05-28 20:42:22 RT @dmetaxak: 2. The costs were high: For the last year and a half I have spent 5-20 hours/week dealing with this—while finishing a PhD, on… 2021-05-28 20:42:13 RT @dmetaxak: And again, the system failed us: 1. The process is rigged for the powerful 2021-05-28 20:41:56 @dmetaxak (I'm always thinking about systems, sorry if this is too intense if you're still feeling the trauma.) Aside from confidential advocate networks like CARES, I think another good thing would be some kind of private peer support network where it's easy to find people to talk to 2021-05-28 20:40:20 RT @dmetaxak: 3. Where could I find support or solidarity? Avoiding him as best I could thereafter, it took me years to learn how many othe… 2021-05-28 20:40:08 RT @dmetaxak: The system failed me then: 1. How could this have been avoided? He was well-respected and well-liked 2021-05-28 20:40:05 RT @dmetaxak: Since 2016, I have been so careful, so patient, so diligent about the process, but @UMichCSE's statement yesterday—so inadequ… 2021-05-28 20:39:42 RT @samps: This is an awful reminder of how academia can fail people who are harassed & 2021-05-28 20:33:28 @fullyabstract @IntuitionisticB Oh no, not since April! Hope he's doing OK 2021-05-28 20:32:53 @fullyabstract @IntuitionisticB is on that lol 2021-05-28 20:18:15 I think part of it is also writing my thesis, and looking back on where I actually started when I started all of this, and where my work is now. And how in so many papers the future work sections were just completely spot-on, like 100% I knew what I was doing subconsciously 2021-05-28 20:17:24 I've gained a lot of confidence in my research plans and ideas ever since getting a faculty job, and I wonder if it's just the external signal that other people also believe in me. It's a strange experience, but it feels good 2021-05-28 20:14:21 @fullyabstract https://t.co/Dqk20WNcqH 2021-05-28 20:02:08 RT @thatinterlace: huge if true https://t.co/uIKcwIYsRY 2021-05-28 20:00:29 Meanwhile in the pure symbolic world https://t.co/CHO3mYUbui 2021-05-28 19:57:41 @znmeb @gilbert_dev I hope neurosymbolic gets popular, because it means programming languages researchers will have a lot to contribute, and it will be fun 2021-05-28 19:56:14 @znmeb @gilbert_dev It would be good to see these with GPT-3 though. IIRC it still struggles with a lot of these examples, but still curious 2021-05-28 19:45:10 @znmeb @gilbert_dev This is why I believe in a neurosymbolic future~ 2021-05-28 19:44:32 @znmeb @gilbert_dev I think it's simultaneously extremely impressive that current NNs can start to imitate language so well given enough data, and extremely unimpressive how they fail to truly generalize anything basically ever 2021-05-28 19:17:23 @tomyudi Prescribe medication to our computers? 2021-05-28 19:14:43 It also seems to resemble my thought process and use of internal language as I drift off to sleep 2021-05-28 19:11:59 I kind of wonder of the resemblance between speech patterns in certain mental illnesses and Markov chain bots 1) says something about thought statefulness in those mental illnesses, and 2) is what causes certain platforms to often mark people with those illnesses as spam 2021-05-28 18:48:17 @spimees Not sure about mandatory reporting yet, but I offered to participate in the process of changing it instead of just asking 2021-05-28 18:47:24 @spimees Oh she is super supportive and was grateful haha. She is different about this kind of thing, I think we are similar, very practically oriented 2021-05-28 17:45:24 @certifiablyrand I'm just going to hire a photographer for fun photos in mine 2021-05-28 17:42:14 @codydroux damn how did you read my mind 2021-05-28 17:39:45 yikes https://t.co/uRTKt19TNn 2021-05-28 17:34:57 SMH when will I ever wear this hat https://t.co/4xT7njog5Y 2021-05-28 17:33:57 It came with a goofy hat https://t.co/tLqbMfHnDS 2021-05-28 17:31:29 Ordered this cute new outfit on the internet, but apparently not allowed to wear it until next week https://t.co/OZDvvWKJdR 2021-05-28 17:28:34 Had the final PUMPKIN PATCH project meeting today. How wonderful working with all of you! Of course, the PUMPKIN PATCH project will continue---new adventures abound https://t.co/A8yMAoiEoq 2021-05-28 17:26:15 RT @kremedelakitty_: men, anytime a woman comes forward: PROOF??? also men, when there is proof: https://t.co/gc2MIm96ee 2021-05-28 17:24:56 RT @genevievemp: I reported the event in detail to AAAI in 2017 via a video call. Unexpectedly, Walter also emailed me an apology. I sent a… 2021-05-28 17:24:37 RT @genevievemp: I was at that bar because of HCOMP 2016. I was networking, trying to learn about jobs and collaborations. Because of Walte… 2021-05-28 17:24:18 RT @genevievemp: I can’t be anonymous anymore. I’m one of 25 people with this experience. I followed official channels to establish his pat… 2021-05-28 17:24:05 RT @genevievemp: Walter Lasecki assaulted me. He pinned me against a bar, he put his hand up my skirt and his fingers in my underwear. He g… 2021-05-28 17:14:03 @ShriramKMurthi @rjmh @hillelogram @DRMacIver Haha, the really unfortunate part is that I don't have the code released because of Amazon legal problems. I have been hoping that the ideas are discussed and built on nonetheless, and it would bring me joy if anything in there sparks useful ideas for an open source and free tool 2021-05-28 17:08:18 RT @josh_sokol: this figure I just encountered in a scientific paper has split my entire day into the Before and the After https://t.co/j8b… 2021-05-28 16:59:31 @ShriramKMurthi @rjmh @hillelogram @DRMacIver It's here: https://t.co/MhbJv08ZsU 2021-05-28 16:58:44 @ShriramKMurthi @rjmh @hillelogram @DRMacIver Based on what you mentioned about a generator built on Alloy, though, probably reading the Iorek paper is still useful for related work 2021-05-28 16:58:15 @ShriramKMurthi @rjmh @hillelogram @DRMacIver For Iorek (declarative solver-aided PBT, no probability distributions) I used it to augment existing tests. However, these days I think the coverage metrics are unhelpful 2021-05-28 16:33:59 @yawaramin @iyad_elbaghdadi (Oh a lot of purple dot folks are 2SS supporters, I just meant it's useful as a way of diambiguating for those who are 1SS supporters) 2021-05-28 16:28:10 @yawaramin @iyad_elbaghdadi Sorry that's long! It's something I think about a lot 2021-05-28 16:27:45 @yawaramin @iyad_elbaghdadi But this is mostly for Americans and Europeans with no roots in the middle east. I think Palestinians I've spoken to have been much more reasonable about this than Americans, so maybe the problem is Americans hijacking the language maliciously 2021-05-28 16:26:29 @yawaramin @iyad_elbaghdadi Otherwise I think the language needs to be more explicit and precise, not as a form of tone policing, but just as a way of signaling Israeli Jews are safe participating in the process. Also because Israelis are very direct communicators who take language at face value 2021-05-28 16:25:05 @yawaramin @iyad_elbaghdadi I can never tell who I'm talking to, which makes it hard to engage and know I'm not hurting my family. The purple dot has been helpful for that since I know people who use it are signaling the former and not the latter, so it has made me really happy 2021-05-28 16:23:59 @yawaramin @iyad_elbaghdadi What makes this hard is that the language is ambiguous enough is that some people who say "decolonize Palestine" or "free Palestine" or "from the river to the sea" agree with me in that they eventually want a single democratic state, and some want to kick all of the Jews out 2021-05-28 16:22:04 @yawaramin @iyad_elbaghdadi The only thing that will end this is Israeli Jews feeling convinced that, whatever the outcome is, Israeli Jews will be safe. I think this is possible, but will take a massive education campaign both for Israelis and Palestinians, and also a change in dialogue by the world 2021-05-28 16:18:25 @yawaramin @iyad_elbaghdadi Of course, this incentives Netanyahu and friends on the right wing to start conflicts so that Palestinians retaliate, and they can use that to feed back into the fear rhetoric 2021-05-28 16:17:52 @yawaramin @iyad_elbaghdadi (Though it's worth noting that "Zionism" has different meanings to different people, not all of which entail an ethnic nation-state.) 2021-05-28 16:15:53 @yawaramin @iyad_elbaghdadi But if you surrounded Israel by a bunch of WASPY Americans you'd get the same fear. It's about not having a single country in the world where Israeli Jews know they can be safe. Fear is what drives Zionism 2021-05-28 16:14:18 @yawaramin @iyad_elbaghdadi IMO, addressing this reality without using it to justify blockading and bombarding Gaza is prudent, especially if the goal is a single unified state. I'm aware of course that racist tropes about Arabs often paint all Arabs as dangerous, so it crosses into racism very often 2021-05-28 16:12:29 @yawaramin @iyad_elbaghdadi So many American Jews came from the more privileged Jewish families that made it out by their own will without first becoming refugees. So I think they do not have the same family trauma 2021-05-28 16:11:22 @yawaramin @iyad_elbaghdadi I think the difference in American and Israeli views of this is one of proximity to genocide historically. More Israelis than Americans had family in Nazi death camps (that's how my grandparents ended up there) or got pushed out of neighboring middle eastern countries 2021-05-28 16:09:32 @yawaramin @iyad_elbaghdadi I think this is a mistake: you can protect ethnic groups in a democracy without a majority, and even grant them some autonomy. Jews can get that in a binational single state, I think. But my views on that are unpopular in Israel 2021-05-28 16:08:03 @yawaramin @iyad_elbaghdadi Also, the fear point is relevant not just for history, but for the present day. Israeli Jews by and large do not trust _any_ country without a Jewish majority to not eventually genocide them. To be honest, even growing up in the US, I'm surprised American Jews feel safe here 2021-05-28 16:05:34 @yawaramin @iyad_elbaghdadi I think that's fine and extremely true, I just think you can get that more accurately and without backfiring by avoiding language traditionally used for powerful groups expanding empires---unless it's about the UK's role. 2021-05-28 14:19:47 @yawaramin @iyad_elbaghdadi I'll follow though, it's always good to see information that challenges you in some way 2021-05-28 14:13:59 @yawaramin @iyad_elbaghdadi I largely agree on the eventual goal though, just not on how to get there, I think. Language really matters, practically, even if it sucks that that's true 2021-05-28 14:08:04 @yawaramin @iyad_elbaghdadi Also it would be nice if the UK would pay reparations and help Palestinian refugees, I don't understand why they get to run from all of the disasters they start 2021-05-28 14:06:34 @yawaramin @iyad_elbaghdadi For Jews, it's reasonable fear from history of refugees having nowhere else to go, and the more afraid people feel, the more Netanyahu & 2021-05-28 14:01:41 @yawaramin @iyad_elbaghdadi Ah, I don't think the colonial view that doesn't recognize Israel as a state maintained out of fear by an ethnic group with 2000 years of oppression & 2021-05-28 13:51:04 @yawaramin @iyad_elbaghdadi Thanks for the recommendation! Will do 2021-05-28 13:49:53 @Jon44444444 @yonahwolf @Mottel The ideology is antisemitism, not political party 2021-05-28 13:47:45 @Mottel Yeah, people justify their antisemitism with whatever political ideologies they align with. The antisemitism is primary, though. 2021-05-28 13:41:55 Every time I see the purple dot for Jewish-Arab peace and equality in Israel/Palestine I automatically follow people because everyone I follow gives me a bit more faith in the future of the world 2021-05-28 13:38:37 RT @Yair_Rosenberg: Most Americans, Jewish and otherwise, have no idea what European Jews put up with for being Jewish on a regular basis.… 2021-05-28 13:36:49 Up after 3.5 hours of sleep for a mentoring committee meeting. Do I run and then nap, or nap and then run? Unfortunately as an extrovert, nothing wakes me up like talking to a bunch of people, so I don't feel like it's early at all 2021-05-28 12:20:12 RT @CaseyHo: White guy learns for the first time that bringing up his Asian wife in a conversation about racism is actually an incredibly t… 2021-05-28 07:48:44 RT @newscientist: Monkeys can change their accent to communicate with another species https://t.co/RvWzOlrDG8 https://t.co/yVKCTj2qUr 2021-05-28 06:00:50 @jeanqasaur @mustach_io @owenarden @0xabU @danrkports @johnregehr @barik Seriously I almost committed suicide in 2019 from the consequences of retaliation. Be extremely careful. Safety must always come before punishment 2021-05-28 05:59:55 @jeanqasaur @mustach_io @owenarden @0xabU @danrkports @johnregehr @barik Retaliation is a huge concern and can be seriously deadly if you try to enact consequences or even confront the abuser without the permission of the abused 2021-05-28 05:56:42 @0xabU @danrkports @johnregehr @barik It's not hypocritical to speak up. But better to use your position to push for systemic changes that help people address abuse safely, like confidential advocate systems. 2021-05-28 05:55:18 @0xabU @danrkports @johnregehr @barik There are probably abusers in almost every CS lab. The problem is systemic. 2021-05-28 05:33:30 @VicGardenhire77 How are you feeling? 2021-05-28 05:18:29 Can we please all just pick one thing that's broken about the world tomorrow, and try to make it a little less broken in a way that does not impose unnecessary harm? Please? Just one day? 2021-05-28 05:17:24 I just want people to treat each other with kindness and compassion, to protect people who need to be protected, to help people who need to be helped. There is just too much suffering, and it is really heartbreaking. 2021-05-28 05:15:41 Not really feeling okay today, though. I just feel so sad about the state of the world, and how humans treat each other, or just sit by while other humans suffer, and try not to get involved. And how hard it is to actually make anything better. 2021-05-28 04:48:04 SIGARCH: https://t.co/ZRqlGtjcZu 2021-05-28 04:47:41 SIGPLAN: https://t.co/aSRL3vFFWu 2021-05-28 04:47:23 Oh, yeah. Now seems like a good time to remind people about CARES, and how it's possible to implement it not just at the SIG level, but also within departments (but please do be careful about the interplay with mandatory reporting). https://t.co/FFmFku6l0j 2021-05-28 04:46:03 I haven't graduated yet and I'm already trying to enact systemic change. I mean they hired me, presumably they knew what they were getting 2021-05-28 04:45:25 In case you're wondering what kind of professor I'm going to be: After my Illinois interview, but before my job offer, I sent the department head a long email full of ideas for her CARES plans. Now I'm trying to see if the notion of mandatory reporter can be flexible for CARES 2021-05-28 03:46:06 https://t.co/dYDpIUYsZy 2021-05-28 03:24:21 @zacharylipton I think you'll find that most of us are not surprised, and many of us have experienced these things before despite not talking about it much 2021-05-28 03:23:06 https://t.co/IoYhsJMk2S 2021-05-28 03:02:39 @roydanroy Also I don't know how this community works, but is this not something you could just ask the PC chair? This seems like a chair question 2021-05-28 03:01:42 @roydanroy If the answer is really "I can't think of anything," I think that's fine. If it's not, then I think you should write about it and explain what could be done to mitigate those harms. 2021-05-28 03:01:12 @roydanroy I don't know the people who do the ethics reviews, but I would stay true to the spirit: Think really hard about how what you do could be misused unethically, like what uses of your algorithm you'd love to rule out if you could just somehow do that by waving a magic wand. 2021-05-28 01:56:18 By "simple" by the way, I don't mean "easy to understand, even with no context," sorry if that makes people feel bad! I mean "described in very few rules, the majority of which are trivially derivable for a type theorist, and only 2-4 of which are interesting." 2021-05-28 01:51:40 I proved it after I implemented it, but then Carlo explained to me *why* it's true. I like my feelings. They're good and useful. I don't know where mathematical "feelings" come from, but I'm grateful I have them even when I lack the language to explain them 2021-05-28 01:50:10 The categorical intuition came from Carlo Angiuli. I didn't have it, I just had a "feeling" that you could always jump across the equivalence instead of converting back and forth, and that just two reified equality rules could fix the definitional equality problems 2021-05-28 01:48:41 It looks cute but it's actually like really deep and difficult tricks. But the result is that my transformation is sooooooooo simple https://t.co/yY3jLzOgBY https://t.co/G7nskQUmJa 2021-05-28 01:36:59 https://t.co/OKRIlvFtZk 2021-05-28 01:11:01 https://t.co/LQkIwrS1tk 2021-05-27 23:11:48 Very sad today, I hope the world is better tomorrow. 2021-05-27 22:31:54 @VicGardenhire77 I can stop commenting now if it is making you feel worse. If you need someone to talk to, though, I am happy to look for more resources for you. 2021-05-27 22:31:14 @VicGardenhire77 To make sure I could find local resources for you, I looked in your profile at a video you posted recently, which listed that your hometown is in Olympia. I don't know if you still live there, though, and I know nothing else about you. 2021-05-27 22:23:01 @VicGardenhire77 The Olympia crisis line is a different number from the one I first sent, and they are very responsive. I think it's likely they answer if you call: 1-800-270-0041. I think they are also kind and compassionate. 2021-05-27 22:12:28 @VicGardenhire77 Sorry for insulting you. What I meant was that I don't understand your situation and probably will not. But there are probably people out there who can, and I hope you will find them. I hope the resources help you find them. Please stay safe. 2021-05-27 22:05:11 @VicGardenhire77 I think these are things you really should talk to someone about. If you are in Olympia, and can't reach anyone for an appointment or any friend for help, the local crisis line to call is 1-800-270-0041. 2021-05-27 21:49:23 @VicGardenhire77 Yeah, I promise I'm not doing this from a sense of internal obligation. It just hurts a lot when you're suffering and nobody sees your suffering. My commenting here is just an expression of humanity, nothing more. 2021-05-27 21:45:08 @VicGardenhire77 It also feels like the last thing you want when you are suicidal. But suicidal ideation is often a desire for escape. There are other ways to escape, and those are things you can talk to mental health professionals about. 2021-05-27 21:44:20 @VicGardenhire77 I don't know much about your situation, but whatever problems you are experiencing, you can only solve them if you can first feel the will to live. It is important to seek help for suicidal intent, and a sign of strength---though it quite difficult and scary in the moment. 2021-05-27 21:39:51 @VicGardenhire77 A quick search suggests you are in Olympia? Perhaps these people will be more compassionate than some random hotline: https://t.co/r4R1liqDhq 2021-05-27 21:39:15 @VicGardenhire77 It is true I know nothing about your situation, and I'm sorry if I suggested that I do. I just have been suicidal before and it feels really bad. I've also made it out of that OK, and I know even though it takes years, it is worth coming out on the other side. 2021-05-27 21:34:32 @VicGardenhire77 It can take years to feel better, but when you finally do, it is really worth it. 2021-05-27 21:33:55 @VicGardenhire77 I would look at local numbers to call in your area, then. Or, if you have friends nearby, ask if one will accompany you to the ER or to mental health triage. It helps to have someone with you so that someone can advocate for you and make sure you are treated well. 2021-05-27 21:25:34 @VicGardenhire77 Suicide prevention hotline: 800-273-8255 https://t.co/rmSsa2YZIg Please take care of yourself and get the appropriate help you need. I have been there. Life can get better. 2021-05-27 21:15:33 @yoavgo 2021-05-27 21:15:16 She fixed the mistake with the Second Intifada numbers. That makes me genuinely happy https://t.co/HY9nzYj6Vg 2021-05-27 21:12:30 @h_jackson_ @MIT @nytimes I'm deleting the other comments just to prevent accidental Twitter dogpiling by people who don't see the update 2021-05-27 21:08:06 @ilyasergey @yalenus @NUSComputing Congrats!!!!! 2021-05-27 21:08:01 RT @ilyasergey: Got some big news this morning: I’ve been promoted to an Associate Professor with tenure @yalenus and @NUSComputing. 2021-05-27 21:06:39 @h_jackson_ @MIT @nytimes Just saw you fixed this---thank you so much, @h_jackson_ ! 2021-05-27 19:04:13 RT @KevinKaichuang: @LughSpear @TaliaRinger https://t.co/e8yamhu7wx 2021-05-27 18:04:44 @EliSennesh Also some Ashkenazi Jews succeeded in being viewed as Plain Old Europeans for a while before we got genocided for not being Europeans. It's complicated, which is why I always appreciate free-form ethnicity fields so I can just write "Ashkenazi Jewish" 2021-05-27 18:00:39 @EliSennesh Though there's also an internalized "Judaism is just a religion and we are just plain old white people" thing from the previous generation of Ashkenazi Jewish Americans probably in response to antisemitism, and the success in ever being viewed that way is relevant to race 2021-05-27 17:59:27 @EliSennesh I think it's just an antisemitic thing where people never want us to belong to anything, and assert our history is always a lie, no matter where we try to go and how we try to fit in. From my view Judaism will inherently always be middle eastern, even if diaspora groups are huge 2021-05-27 17:56:58 @EliSennesh Yeah that's another thing that's strange and confusing to me, especially considering the number of indigenous middle eastern groups that never left the region and are not muslim, and the history of Islam in the region 2021-05-27 17:50:13 Usually don't need to think about this when middle eastern people count as white, but when that isn't true I have to ask myself what people actually mean by "middle eastern" and I don't really know, but I think it's "Arab" 2021-05-27 17:49:05 In my head I jokingly describe my heritage as "middle eastern European" - genetically half middle eastern half southern European - culturally half middle eastern half eastern European - residentially all three of those, at different points in time 2021-05-27 17:43:10 Usually in the end I check both "white" and "middle eastern" because I think it's the most accurate, but if people mean Arab I'd appreciate them just saying Arab 2021-05-27 17:42:26 I mean, not just IMO, also like based on genetic studies. But the problem is that's not what people mean. I think they mean Arab. I'm not even sure they would consider Mizrahi Jews to count, even though they never left the middle east 2021-05-27 17:41:10 I still struggle super hard when "middle eastern" is a category in addition to white. Like, if my parents are from the middle east, and my heritage is about half middle eastern historically, am I middle eastern? IMO ethnic Jews are all at least partly middle eastern inherently 2021-05-27 17:35:16 RT @roscosmos: Государственная комиссия на космодроме Восточный приняла решение о пуске ракеты-носителя «Союз-2.1б» с 36 космическими аппар… 2021-05-27 17:30:55 @SMukherjee89 I grew up with a Holocaust-impacted definition of whiteness, and was profoundly surprised in early grad school when I learned that many Indian people consider themselves brown or Asian, not white! And I know how incredibly diverse India is. I just was taught Indians were white 2021-05-27 17:26:41 @SMukherjee89 The obvious point being that it's the groups we define based on these differences that are entirely social, and those groups vary substantially around the world and over time 2021-05-27 17:25:06 @SMukherjee89 I used to be confused by the "race isn't real" discussion because like, obviously, there are some differences between the groups we define. For example, white people by and large have lighter skin than non-white people. I was just missing the obvious point though 2021-05-27 17:19:05 The answer is "many Jews are white for many definitions of white, but all Jews are not white for some definitions of white" which makes sense only because race is a social construct 2021-05-27 17:16:55 It's easy, just ask the question "are Jews white?" and you'll immediately be confronted with people yelling at you inconsistently for different things, in a way that explains quite naturally all of the social constructs involved 2021-05-27 17:14:43 Do these people not realize that race as discrete binning of people by largely superficial traits is a fairly incoherent and constantly shifting social construct meant to sustain power dynamics that benefit some groups of people at the expense of others? https://t.co/BphBUpjDtl 2021-05-27 17:09:13 @sarahhollowell @mekki I've had this discussion a lot about hybrid conferences. It mostly seems that few people want to put in extra effort to make smooth hybrid experiences. Most assume it will fail and give up before trying 2021-05-27 17:07:28 RT @sarahhollowell: I just don't understand why opening back up as people get vaccinated also has to mean closing down all the accessibile… 2021-05-27 16:53:25 @elfprince13 @shubtri707 The robots are in the basement though 2021-05-27 07:54:20 RT @rob_rix: @TaliaRinger e.g. every single one of the hundred presentations of sequent calculi for any logic you like, each unique in its… 2021-05-27 07:40:21 @LughSpear Literally the entire thread applies to this, like every comment, so much that it is clear you are being deliberately obtuse 2021-05-27 07:37:51 @SMT_Solvers Drugs wouldn't surprise me. Same problem as male suicide rates. No ability to express emotions or seek support leads to unhealthy coping mechanisms 2021-05-27 07:32:51 @SMT_Solvers That's a different problem, mostly related to how men are not allowed to show any emotions except anger, and are taught to externalize all blame and harm others rather than accept the consequences of their own actions 2021-05-27 07:30:52 @LughSpear Lol read the rest of the thread 2021-05-27 07:28:19 RT @jonrosenberg: @nypost Look at this guy, sleeping in beds, at a temperature. Who does he think he is 2021-05-27 07:11:22 https://t.co/03NqGAAp6s 2021-05-27 07:10:27 Not to mention the exceptional tolerance of harmful behaviors from male "geniuses" that, if encountered in equally smart women, would lead to them being fired or otherwise punished 2021-05-27 07:06:25 And abuse and harassment driving women out of many fields dominated by male "genius" 2021-05-27 07:05:18 This of course not even accounting for historical lack of access to education, and both historical and ongoing lack of credit for accomplishments 2021-05-27 07:03:14 Women are still largely expected to be able to clean, cook regularly, organize things, and have good social manners and so are constantly judged as failures for symptoms people put up with in men all the time 2021-05-27 07:01:32 This means that women often go undiagnosed and untreated. Furthermore, social norms for women punish neurodiversity disproportionately, making it difficult for it to manifest as "genius" 2021-05-27 07:00:37 This is why the "men at the extrema of neurodiversity are more common than women and that's why male genius is more common" dialogue annoys me like crazy. Men are more common at the extrema because the diagnostic criteria are informed by studies of men https://t.co/WxJ3gCx6iI 2021-05-27 06:57:25 @amberbrinp Word 2021-05-27 06:57:17 RT @amberbrinp: @TaliaRinger Yeah, it was wild to me that other students could walk into class, pay attention to the lecture well enough to… 2021-05-27 06:55:53 @amberbrinp I was also "gifted" which kind of made things worse because I could literally never study, do all of my homework the day of in the class before the class it's due, and cruise through high school with Bs looking like I just don't try much 2021-05-27 06:54:25 @amberbrinp Likewise hyperactivity would manifest mostly in terms of not standing still and getting in trouble but for a lot of women because of social norms it's talking a lot or being hypersocial 2021-05-27 06:52:58 @amberbrinp I think there is also a lot of medical sexism involved. The diagnostic criteria were historically informed mostly by studies of men, so that for example outbursts of anger were recognized but outbursts of sadness were not 2021-05-27 06:51:10 @amberbrinp But nobody told me I might have ADHD I only noticed once I started to care that other people could pay attention and learn from lectures, but I had to teach myself everything from scratch after lecture 2021-05-27 06:50:04 @amberbrinp I sucked at everything until college because nothing I studied was interesting until college-level math and computer science and then WOW 2021-05-27 06:34:43 With that in mind I would not under any circumstances ever wish to "cure" my ADHD, god forbid. Managed well, it's a beautiful gift. I just need medication, coping mechanisms, and accommodations to navigate NT-built systems sometimes. That's fine 2021-05-27 06:23:43 @wilton_quinn Yeah the stereotype for emotional disregulation is outbursts of anger but I think that's because the diagnostic criteria are informed by studies on mostly men 2021-05-27 06:22:46 @wilton_quinn Ah gotcha 2021-05-27 06:21:04 @wilton_quinn Oooh which one is the diff 2021-05-27 06:19:03 RT @just_mary_tweet: @danidonovan How confusing it is to not understand NTs giving instructions How *much* people's negative behavior like… 2021-05-27 06:16:54 RT @ShadowclawFC: @danidonovan Also the sheer force of every emotion I feel, no matter how fleeting it's always ramped up to 11... I tend t… 2021-05-27 06:16:50 RT @ShadowclawFC: @danidonovan The neverending thoughts, multiple threads running at full speed all the time, anytime there's even remotely… 2021-05-27 06:12:00 Can you believe I wasn't diagnosed until age 19? Lmao 2021-05-27 06:11:39 15. The absolute lack of any mental process saying "maybe I shouldn't do that" basically ever 2021-05-27 06:11:05 11. How impossible it is to make a phone call or book an appointment 12. How insanely stressful it is to plan simple daily things like when I will clean 13. How impossible it is to be on time 14. How impossible it is to go to bed 2021-05-27 06:09:40 8. How I can procrastinate on eating or using the bathroom for the entire day by being stuck on a thought or task 9. The inability to stop myself from crying in public when I'm sad 10. How even mild rejection from people I care about feels like the death of a loved one 2021-05-27 06:07:54 5. How hyperactive I can be in my head while physically being exhausted 6. How 16 hours can pass in 5 minutes if the task is interesting enough 7. How I can zone out and suddenly have written an amazing program or page of a paper but have no memory of the process of doing it 2021-05-27 06:06:33 2. How impossible it is to follow talks and lectures 3. How impossible it is to follow conversations with many people 4. How quickly all of my thoughts go go go 2021-05-27 06:05:24 I think neurotypicals would be surprised by: 1. How a single sentence someone else says can mean me totally losing any semblance of connection to the present moment and instead withdrawing into my head fleshing out some new idea I'm positive I have to implement https://t.co/4LggiheTjf 2021-05-27 06:03:00 RT @danidonovan: I think neurotypicals would be surprised by: 1. The sheer chaotic velocity of my brain moving between semi-related though… 2021-05-27 06:00:31 RT @ZJAyres: I know deep down there are people that judge me for speaking about my mental health. That I need to be quiet. Maybe they think… 2021-05-27 05:48:43 RT @wilcoxjay: PL twitter may enjoy this: hacked together a little System F playground for my students!! https://t.co/PBqCVCUAHN 2021-05-27 05:46:39 Good luck, Twitter friends who are submitting papers to NeurIPS! https://t.co/8VGBXo6C2Q 2021-05-27 05:41:47 @shubtri707 I'll go say hi in case they're miserable 2021-05-27 05:40:34 NeurIPS? 2021-05-27 05:39:23 Is there a deadline or something? The machine learning people are still in their lab 2021-05-27 05:34:23 RT @jonmsterling: Thick math!! https://t.co/PmBsZjsumt 2021-05-27 05:34:12 @jonmsterling This is pretty 2021-05-27 05:11:03 @ian5v I can't figure out how, but I'd love to, because I don't trust this website 2021-05-27 05:07:53 @ian5v Yeah send me your email address and I'll give you a link. Getting to those invites slowly though, have a backlog 2021-05-27 05:05:57 @ian5v Here's a lightning talk, so just 5 minutes, to give you a quick sample: https://t.co/p3P8i1vSNU 2021-05-27 05:02:24 @ian5v I do many slides that all flow into each other like a single animation. It's a really strange way to give research talks, very different from most people, and it takes a lot of effort to keep it easy to follow that way. But generally people enjoy them, and I enjoy giving them 2021-05-27 04:56:50 @wilton_quinn Thanks it uses a Google Slides format but I went with the light blue a long time ago to highlight significant or important things, and I haven't really gone back because I like it. It's fun to have a style 2021-05-27 04:54:48 @ian5v Like the stick man has shown up in most of my talks now for a year and a half. I am kind of attached to him. 2021-05-27 04:53:54 @ian5v They're based off of talks I've given for a few years now, but the slides here are all from the last couple of days. Just I have a fixed style and content I reuse which makes it easier to make new talks from those building blocks 2021-05-27 04:52:42 @VittoMartinelli Also good for ADHD people like me who space out and need to figure out what is going on in the middle of the talk 2021-05-27 04:52:07 @VittoMartinelli Ah this is an animation trick, I write heavily animated but static slides, so it helps a lot to have fixed parts of the slide that don't move so that you actually perceive the things that do move as animation 2021-05-27 04:50:49 https://t.co/NUrq2HnEmg 2021-05-27 04:49:24 @kamatsu8 Exactly. So "the collective art of truth" is a fine description of math, insofar as you accept truth as being an art 2021-05-27 04:40:41 @kamatsu8 Truth in itself is an art 2021-05-27 04:39:37 https://t.co/5Ft7Lr4fDP 2021-05-27 03:28:24 mathematics https://t.co/6VsUUzDgtQ 2021-05-27 03:27:06 @ilyaraz2 When I was choosing where to be a professor, agonizing over a choice between two places I knew I'd be happy, my mom told me, "whatever choice you make will be the right one." That made me feel much better. 2021-05-27 03:18:39 This is also why metareviews and survey papers are so important. They tell the stories of how fields move. Stories invisible when we find ourselves within them. https://t.co/rt81qZfHKd 2021-05-27 03:16:02 @RaghavMalik15 The collective art of truth 2021-05-27 03:11:46 @RaghavMalik15 Yeah. It's like being confused about calculus in the 1600s. Everyone was confused about calculus, if they even knew calculus was a thing. 2021-05-27 03:04:55 @RaghavMalik15 Yes, that was the first time that I had that experience, too. Higher category theory is absolutely full of this. I used to think it was commentary on my cognitive abilities, but now I just know it's indicative of the early days of a field in motion. 2021-05-27 02:29:09 "As such, this article may not represent the views of all researchers in the fields equally. This kind of variability is unavoidable when a field is in rapid flux." 2021-05-27 02:28:59 "There is a large overlap between the work referred to as [HoTT], and as [univalent foundations]. Although neither is precisely delineated, and the terms are sometimes used interchangeably, the choice of usage also sometimes corresponds to differences in viewpoint and emphasis." 2021-05-27 02:27:51 I've spoken about this before, which is why I had slides ready to go. But this particular Twitter thread is inspired by this beautiful excerpt from the Wikipedia page for Homotopy Type Theory: 2021-05-27 02:12:05 But just remember this: When it feels like nobody agrees on definitions, conventions, and notation, this is not a bad thing. This is movement. 2021-05-27 02:08:00 This is why proof maintenance is such a big deal, and a lot of the motivation for my thesis work---automatically repairing broken proofs of program correctness. Proof repair. 2021-05-27 02:07:23 This process never converges. Everything is always moving, and moving quickly. By the time it converges, it is about to break again. 2021-05-27 02:06:33 So the world of proof engineering---that is, engineering computer-checked proofs about programs---is different in that every day, your manager asks you to invent a new calculus. The last one wasn't quite good enough. You need a new calculus by tomorrow. https://t.co/UrszlGhUjb 2021-05-27 02:04:49 When we cross the Curry-Howard bridge from proofs to programs, we see a difference. Most people say that difference is movement. But this is not true 2021-05-27 02:02:33 Math moves like a tree, so we don't always perceive its movement. 2021-05-27 02:01:00 Math is always like this, in flux. We just don't notice it when we're in the moment. 2021-05-27 01:59:54 But the notation kind of sucked, so this dude Gottfried Leibniz came around and was like, "bro, I've got better notation" and that's what we're used to these days. He actually also claimed to have invented calculus, and Newton kind of hated him for it TBH. https://t.co/qDaI4RPkBO 2021-05-27 01:58:15 Isaac Barrow had this smart student---Isaac Newton. And Isaac Newton fleshed out the rest of the theory involved. https://t.co/EndFL96LZT 2021-05-27 01:56:52 Isaac Barrow was like, "bro, your calculus isn't sufficiently general," and decided to prove a more general version. https://t.co/327pyjfjHv 2021-05-27 01:56:23 People say that Newton invented calculus in quarantine or whatever. But this is false. Actually this dude James Gregory in the 1600s was like "hey here's calculus" and then this other dude Isaac Barrow looked at it. https://t.co/uDbSmWWllg 2021-05-27 01:53:27 What may be surprising is that calculus was once like this too. 2021-05-27 01:53:09 What is an "ornament" in type theory? I can give you all of the definitions of ornaments that have ever been written about. They don't even always make the same foundational assumptions. 2021-05-27 01:52:00 What is an "inductive type"? Good question. I can give you a lot of examples, and a lot of different ways of modeling them. But when it all comes down to it, I have absolutely no clue. 2021-05-27 01:50:43 In the moment, math---and really all theoretical research---is not like this at all. At any moment there are like ten different definitions, ten ways of reasoning about the concepts each entail 2021-05-27 01:49:49 Looking back on math, it always feels like all concepts are firm with fairly uniform definitions, terminology, and conventions. But this is just because math is really, really old, and so these ideas have converged over time. 2021-05-27 01:40:46 Oh, and the thing on the left of the photo is a diagram of a particular "type equivalence," which is a concept from homotopy type theory (HoTT) that I use in my research. 2021-05-27 01:39:40 (One thing I've started doing since getting a faculty job is explaining jokes so as not to alienate people who don't get them as much. It probably seems silly, but I do often feel left out when I don't get "inside jokes.") 2021-05-27 01:38:26 For people who are not addicted to the internet and memes: https://t.co/6o5mNKmBq6 https://t.co/KJLo7VOeeE 2021-05-27 01:32:33 For people who are not programming languages nerds: https://t.co/Ao7VA9jbQY 2021-05-27 01:28:56 HoTT Girl Summer https://t.co/2MlJfOklIP 2021-05-27 01:12:10 RT @hourlyFox: https://t.co/NruHRxKaLL 2021-05-27 01:03:59 Are you really a doctor if you don't go through the hooding ritual? 2021-05-27 01:03:03 Should I try to talk my advisor into hooding me in a park? 2021-05-27 00:59:48 RT @NPR: About 400,000 more women than men have left the workforce since the start of the pandemic. Now, a new business coalition wants co… 2021-05-27 00:55:59 your last saved meme is your moral philosophy https://t.co/HniJfZTdLb https://t.co/ad9BRDbbmU 2021-05-26 23:19:53 https://t.co/nxLxBv5daX 2021-05-26 22:30:23 https://t.co/KZVJriffLx 2021-05-26 22:05:40 It's amazing that people fight it for way longer than I had to and survive. It was like 1.5 years for me of severe depression & 2021-05-26 22:03:35 There is nothing I have experienced in my life that was worse than severe depression & 2021-05-26 21:59:56 True especially since a lot of the casualties of mental illness are going to come later, once the pandemic ends and many people find themselves chronically depressed despite the removed external stimulus https://t.co/aWTF7EGrhT 2021-05-26 21:44:08 The most interesting thing I learned about myself this pandemic is that I'm willing to pay an outrageous amount of money to have boba delivered to me while I work 2021-05-26 21:43:17 I feel so called out https://t.co/EQl6sVmb1L 2021-05-26 21:24:22 RT @danrkports: Noticed that @schemeprincess has an appointment just labeled "burrito" on her calendar. Always make time for the important… 2021-05-26 20:45:03 https://t.co/v2ASuFqz8R 2021-05-26 20:41:02 @TzuHanH yes yes yes good idea 2021-05-26 18:43:13 @vdean314 It always makes me uncomfortable as a reviewer how easy it is to tell whether a paper is by an American author or not just based on things like this 2021-05-26 18:38:52 @krismicinski @Firesnakious why are people like this 2021-05-26 18:33:05 I don't agree with anyone in this article, really, but the emotional impact as a leftist Israeli-American (dual citizen) with immediate family in Israel are very similar, especially when I know that family has been trying to unseat Netanyahu for years now. 2021-05-26 18:25:17 The quote in the article “my heart has been ripped apart on multiple incompatible jagged edges” kind of summarizes how I feel about all of the dialogue about Israel/Palestine in the US this past week https://t.co/AMbh0L240F 2021-05-26 18:14:22 I'm scared https://t.co/5uH3hc1a41 2021-05-26 17:26:56 If you think the Brood X cicadas are bad, wait until you see the Millennial and Brood Z cicadas 2021-05-26 17:19:54 @SouthernSummr Only for government workers, soldiers, hospitalized people, etc. Not students 2021-05-26 17:17:22 @MonniauxD I'd probably fly to Israel just to vote against Netanyahu though lol 2021-05-26 17:16:51 @MonniauxD From Wikipedia: "absentee ballots are restricted to soldiers, prisoners, sailors, overseas diplomats, disabled persons, and hospitalized people" 2021-05-26 17:14:11 How many Israeli citizens study in the US? If there is another election, how many votes would we consist of if we all go to Israel to vote? 2021-05-26 15:57:18 @ArjunGuha @samth I think at a lot of institutions, it is one of few things tenured faculty can still get fired for, though I don't know the details 2021-05-26 15:41:58 @ArjunGuha @samth True, but at risk of getting fired, even if tenured. I wouldn't expect that from people, though I think it is noble to do so 2021-05-26 15:20:44 @syardi @samth @ArjunGuha Ideally what you want is someone who can personally accompany them to help, or contact a mental health triage system for them. The ER is very scary as a mental health patient. And imagine the incentives for Black students when police are involved in that process, also. 2021-05-26 15:17:56 @syardi @samth @ArjunGuha Yeah honestly even mandatory reporting for suicidal plans is one of those things with good intentions and bad implementation. Who do you tell? Are there police involved? Will the student be treated like a person at the ER as a mental health patient? 2021-05-26 14:57:34 @syardi @samth @ArjunGuha That's wonderful, I think even better would be promising confidentiality until next steps are determined, I often didn't reach out to people if I was afraid confidentiality would be broken 2021-05-26 14:52:13 @samth @ArjunGuha @syardi Yeah I'm going to talk to my dept head about this, she is very very enthusiastic about this kind of thing 2021-05-26 14:47:31 @samth @ArjunGuha @syardi Yeah, but with a system like CARES you can translate the legalspeak into something reasonable for humans and put it on the website, or even run an automated check before you sign up to speak to someone whether that person is a reporter for you specifically 2021-05-26 14:45:11 UMass seems to have gotten this through their union contract (faculty are unionized) https://t.co/fuNqb4DSdB 2021-05-26 14:42:32 @samth @ArjunGuha Yes I'm pushing for systemic change before I start my job this will be a long 6 years but hey, they hired me knowing who I am 2021-05-26 14:41:18 @samth @ArjunGuha Yeah, sounds like UMass got this through the union contract. My concern is mostly for CARES, which UIUC CS just adopted. It would be great to set it up so that you don't talk to mandatory reporters 2021-05-26 14:38:34 UMass and Michigan both have reasonable mandatory reporting policies https://t.co/0OkzBmnsD1 2021-05-26 14:37:19 @samth @ArjunGuha Are they mandatory reporters for other people's RAs? 2021-05-26 14:34:04 If you want to push for systemic change in your university, you can push for not all faculty being mandatory reporters. This opens up the possibility of systems like CARES being implemented more safely within departments https://t.co/TlFRK9XZA3 2021-05-26 14:32:13 @samth @ArjunGuha Wow that's awful that universities do that 2021-05-26 14:30:09 @johnregehr @ArjunGuha Yeah if there's leeway for universities defining who is a mandatory reporter, that opens up the possibility of systems like CARES being implemented truly safely within departments 2021-05-26 14:21:33 @ArjunGuha I'm curious how it worked out that way. Do universities have control over who is a mandatory reporter? 2021-05-26 07:00:42 RT @Stefania_druga: THIS 2021-05-26 06:46:10 @znmeb @BotSentinel @conspirator0 BotSentinal looks useful, too 2021-05-26 06:32:22 Do I have friends who do botnet research? Looking for detection tools 2021-05-26 05:43:32 @tonyxty 0 2021-05-26 05:24:18 I mean that's a version of the Turing Test I guess https://t.co/lgf8vi3xzy 2021-05-26 05:08:14 This is not a comment to Sarah whom I trust 100% unconditionally. This is just for faculty in general, especially those of you who want to help but have never experienced harassment or abuse 2021-05-26 05:06:03 If you are a mandatory reporter, and plan to follow the really terrible law you're supposed to follow, you should always have names of people on hand who students can talk to who are not mandatory reporters and who can swear confidentiality until (if ever) reporting is safe 2021-05-26 05:04:19 Making retaliation a fireable offense doesn't stop abusers and harassers from retaliating, it just means they retaliate in ways that don't involve getting caught 2021-05-26 05:02:59 Like it's often way, way worse than any harassment or abuse, and this is something a lot of people who haven't dealt with it often don't think about. It's why I'm vehemently against mandatory reporting, but I can't change Title IX, I guess 2021-05-26 05:02:03 A quick tangent about this: When Title IX mandatory reporting is relevant (a law with good intentions and extremely terrible outcomes), super please disclose to people upfront before they tell you things, if you're going to report. Retaliation can kill https://t.co/Xh1TLFOn5g 2021-05-26 04:38:08 a e s t h e t i c 2021-05-26 04:37:11 https://t.co/a5iYY4vQ55 2021-05-26 04:25:21 Talia slides no context, defense edition https://t.co/D1xJAdsRB8 2021-05-26 04:14:05 RT @sarahchasins: To students in my various academic communities: If you experience harassment, I will have your back.  I will do everythin… 2021-05-26 02:14:33 @krismicinski @samth @kamatsu8 Yes, compilers have plenty of Abstract Interpretation in them, Kris. Come on, you taught me this 2021-05-26 02:12:33 Also cute https://t.co/aRcd7aa6vx 2021-05-26 00:41:05 It's cute https://t.co/OF2KZC1WyC 2021-05-25 23:21:28 @znmeb @racketlang I just don't know if there is a music library for Racket lol 2021-05-25 23:21:08 @znmeb @racketlang Very cool, then, I think you'd enjoy the wide range of things you can do with Rosette, it is quite powerful 2021-05-25 23:19:58 @znmeb @racketlang @ben_kushigian is my goto for computer music stuff 2021-05-25 23:18:08 @znmeb @racketlang It could be fun to do something with music that way, setting constraints that ought to sound good but vary in details by execution but I don't know how one would implement that 2021-05-25 23:17:27 @znmeb @racketlang What I like about Rosette in particular is that it lets you state high level things, but call out to a constraint solver (like SAT but fancier) to actually choose values for you. So you can say, "this should be a number greater than 5" or whatever and it will figure it out later 2021-05-25 23:10:47 @znmeb @racketlang https://t.co/0hL3HIBNFa 2021-05-25 23:09:40 @znmeb @racketlang You should try Rosette, it's really fun 2021-05-25 23:06:05 @znmeb There are a lot of good Lisp dialects! Especially @racketlang and the solver-aided language Rosette build on top of it: https://t.co/wuP2VtZjJD 2021-05-25 23:03:20 Once parties resume, if you live in NYC and are lonely, I weirdly highly recommend the Lisp meetup, which for some reason exists https://t.co/72ptBVC17W 2021-05-25 23:01:47 @zeeshanlakhani @RemivanTrijp Like another lifetime honestly 2021-05-25 22:52:56 My body, that is 2021-05-25 22:52:48 It's making weird noises 2021-05-25 22:51:03 My body decided last night it only wanted to sleep 2.5 hours. I feel vaguely weird now 2021-05-25 22:48:01 RT @EndedXl: The only impression I have for Computational Type Theory (by Harper) is the *Ninja Move*. Are there other interesting example… 2021-05-25 22:24:41 @imspacekitteh if you email me I'll give more context 2021-05-25 22:18:46 @emilymbender @timnitGebru @mmitchell_ai @RealAbril "Here's a fun subject! Literally the livelihoods of two women haha" yikes that interviewer 2021-05-25 17:48:15 @samth @jer_gib @vj_chidambaram @programmingconf That's good! I just think calling this a "journal" sends confusing messages since there are a lot of bad things about journals that we don't want. But I guess it helps Europeans and some smaller American schools so maybe we need to keep calling them that 2021-05-25 17:45:51 @jer_gib @samth @vj_chidambaram We _can_ do this in PL through a journal, but it's a weird and unusual process, and I'd rather just have this be the conference process, period 2021-05-25 17:45:22 @jer_gib @samth @vj_chidambaram This specifically https://t.co/JrFoOPjfLE 2021-05-25 17:44:00 @jer_gib @samth @vj_chidambaram But conferences shouldn't have attendance requirements for presentation, it should always be OK to just send in a YouTube video of a talk 2021-05-25 17:43:22 @jer_gib @samth @vj_chidambaram No, what I want is a conference system that takes the best parts of journal reviewing and leaves the rest behind. The journal is an outdated format IMO 2021-05-25 17:42:40 @samth @vj_chidambaram The key differences between revisions and resubmissions being more guarantees addressing reviewer feedback will actually help you get your paper in, a way to signal that your paper is *almost* past reviewer bar for a top conference and just needs a bit more work, etc 2021-05-25 17:40:37 @samth @vj_chidambaram Yeah, I do think the general point stands though. Right now I think conference acceptances are artificially low and so largely random and adversarial. But I think a revisions process would do better for addressing this while maintaining a high bar than just accepting more papers 2021-05-25 17:30:37 @vj_chidambaram Not fully objective, but more importantly, blinded 2021-05-25 17:28:34 @vj_chidambaram No, but having top conference papers sends an objective signal that you are probably pretty good, in spite of other biases against you. I appreciated the ability to signal this 2021-05-25 17:27:15 @mustach_io @vj_chidambaram Best papers are way more biased than reviews, especially since they are pretty much always done by small committees post unblinding IIRC 2021-05-25 17:25:59 @vj_chidambaram I think I'd rather just get rid of the once per year part. Some communities have done this already. Still have conferences, but have one submission process for any of the annual conferences in a field, with a revisions process so reviewing isn't adversarial 2021-05-25 17:21:32 RT @rene_mobile: https://t.co/p1O55VzYEi - Active collusion in academic peer review is disturbing. We knew it was always possible, but quit… 2021-05-25 17:21:19 @vj_chidambaram I do agree though that acceptance rates are too low right now, and that it makes the whole process more random than it ought to be 2021-05-25 17:20:46 @vj_chidambaram Hmm, how do you judge content fairly, though? All sorts of biases factor in, especially once author names and institutions are revealed 2021-05-25 17:14:57 @ZJAyres Thank you for being vulnerable and using it to push for systemic change 2021-05-25 17:14:17 RT @ZJAyres: The metrics we judge one another and compare people by can do damage. If we don't talk about these things, we fail a whole hos… 2021-05-25 17:14:13 RT @ZJAyres: Until we acknowledge that the academic system we built can do serious harm, we cannot work towards changing it. No-one should… 2021-05-25 17:14:09 RT @ZJAyres: The 'publish or perish' article circulating today has got to me more than I would like. Here's why. People do "perish" in high… 2021-05-25 17:09:41 RT @BjornerNikolaj: @TaliaRinger I still can't get past the point that it remains possible to be "behind times" in use of ancient Greek let… 2021-05-25 17:09:19 RT @tastarte: @tangled_zans @bentnib @TaliaRinger ... but! Most early computer researchers did a lot of different stuff. There wasn’t the d… 2021-05-25 16:48:33 RT @tastarte: @tangled_zans @bentnib @TaliaRinger @mysdick As for @bentnib’s point: the ALGOL effort was key because it reified PLs: machin… 2021-05-25 16:43:57 RT @tastarte: @tangled_zans @bentnib @TaliaRinger @mysdick ... namely ‘Mathematical Science of Computation’. I argue (in a forthcoming pape… 2021-05-25 16:43:55 RT @tastarte: @tangled_zans @bentnib @TaliaRinger @mysdick While originally a tool for AI research, others found the languages useful in ot… 2021-05-25 16:38:27 Chair Jeffrey S. Foster Tufts Vice-Chair Anders Møller Aarhus Members-at-Large Isil Dillig University of Texas at Austin Antony Hosking Australian National University Cindy Rubio-Gonzalez UC Davis Sukyoung Ryu KAIST Alexandra Silva University College London 2021-05-25 16:35:39 Wow what an amazing committee!!!! https://t.co/umtclGES7h 2021-05-25 16:31:21 For a second I thought the President of Russia was the account posting about Sears and I was quite amused https://t.co/UlKCnCaoPr 2021-05-25 15:28:44 @StanfordAILab The most unfortunate casualty of the AI Winter is this beautiful keyboard 2021-05-25 15:20:16 @samth @brendanzab @jonmsterling @kamatsu8 That's very fair 2021-05-25 15:17:31 @samth @brendanzab @jonmsterling @kamatsu8 But ease viewing say my software engineering job at Amazon that way. Just for research it feels like the more I interact with the community, the more good ideas come back, and I _want_ people to eventually _change_ my goals 2021-05-25 15:16:05 @samth @brendanzab @jonmsterling @kamatsu8 Maybe that's an industry thing, I would have trouble viewing research that way 2021-05-25 15:12:50 @jonmsterling @samth @brendanzab @kamatsu8 I have no idea how to implement cubical but I want to so badly because I love building automation, hence why I'm happy you're willing to work with me on automating cubical 2021-05-25 15:11:47 @samth @jonmsterling @brendanzab @kamatsu8 But with a big enough community, referring those people to the right community members for that job seems like a reasonable and easy call 2021-05-25 15:10:53 @jonmsterling @samth @brendanzab @kamatsu8 I've basically never had anyone show up ready to write code for my kind of work. I've also never had a bad experience helping people learn how to do that, and later watching them do things I would have never been able to do without them 2021-05-25 15:09:02 @samth @jonmsterling @brendanzab @kamatsu8 I wouldn't say it's rare, it has been my experience 100% of the time someone has been interested in my work 2021-05-25 15:08:26 @samth @jonmsterling @brendanzab @kamatsu8 Maybe people just haven't worked on the skill enough of helping newcomers? 2021-05-25 15:07:50 @samth @jonmsterling @brendanzab @kamatsu8 In my experience, I would say that sentence has been false every single time I have experienced it, and the returns of helping people who were confused or couldn't do anything at first absolutely massive 2021-05-25 15:05:08 @brendanzab @jonmsterling @kamatsu8 @samth It's ironic given its history, but it's really wonderful. These projects should always aim to be maximally accessible to people who want to help and do cool things and learn 2021-05-25 15:02:45 @brendanzab @jonmsterling @kamatsu8 @samth IDK what I love about the Coq community is that I've always felt so respected within it, at least in France. They are actively trying to make contributing as approachable as possible, recognizing for example gender imbalances in open source contributions 2021-05-25 15:00:43 @brendanzab @jonmsterling @kamatsu8 @samth With a coherent enough community you can just get them in touch with the right people 2021-05-25 14:59:48 @brendanzab @jonmsterling @kamatsu8 @samth All of us have huge amounts of things on our plates, that isn't a good reason to alienate people who are interested in our work 2021-05-25 14:58:43 @jonmsterling @brendanzab @kamatsu8 @samth Yeah I just don't. I'm happy to talk to anyone who wants to build on my work, that is inherently exciting. If I don't have time, I'll say so, but that's on me, not them 2021-05-25 14:57:12 @jonmsterling @brendanzab @kamatsu8 @samth Yeah but this kind of thing scares away well meaning people, but doesn't scare away any of the personas because they are not usually self aware 2021-05-25 14:55:29 @jonmsterling @brendanzab @kamatsu8 @samth Like now I'm wondering how he perceives me and my ideas? What am I doing wrong? I just wanted one conversation so I could kick off a project. I deliver all of the things I have ideas about eventually, though often after years of thinking and talking to people 2021-05-25 14:54:10 @jonmsterling @brendanzab @kamatsu8 @samth I hate it, this is the sort of thing that scares people away from contributing to projects and approaching researchers 2021-05-25 14:52:14 RT @NWSSeattle: A young Bald Eagle dropped by this morning to say hello. #wawx https://t.co/s40l75D1Rx 2021-05-25 14:50:49 @samth @bentnib @jonmsterling @kamatsu8 Like they paid me to come to France twice for my work, it's just Inria is a bit chaotic 2021-05-25 14:49:19 @samth @bentnib @jonmsterling @kamatsu8 Haha fair, but I think it's not on purpose, I think it's just how the team operates 2021-05-25 14:42:25 @samth @bentnib @jonmsterling @kamatsu8 This is why my work is on 8.9.1, for example. 2021-05-25 14:41:43 @samth @bentnib @jonmsterling @kamatsu8 The hard part is not that the teams don't care about it, but that Coq moves quickly and breaks backwards compatibility all the time. Keeping your automation up to date is really hard 2021-05-25 14:40:55 @samth @bentnib @jonmsterling @kamatsu8 There are a lot of people working on automation in Coq, just most of the fancy stuff is in external plugins & 2021-05-25 14:36:26 @jonmsterling @brendanzab @kamatsu8 @samth I don't know, I don't get it either. I'm confused by community separation in general, I like to switch between proof assistants as needed, I even advised a student on an Isabelle/HOL project 2021-05-25 14:35:05 @jonmsterling @brendanzab @kamatsu8 @samth Oh I just mean for cubical 2021-05-25 14:33:05 @jonmsterling @brendanzab @kamatsu8 @samth Have a bunch of those in PUMPKIN Pi and at this point a pretty general framework for building them, though obviously with limitations since the general problem is intractable. Good for specific classes of equivalences though, and very fun to write 2021-05-25 14:31:41 @jonmsterling @brendanzab @kamatsu8 @samth Fair, I just think for your every day humans, automation and usability are what will matter. Another thing I'm happy to help with is building search procedures that discover and prove type equivalences automatically 2021-05-25 14:29:42 RT @bentnib: @jonmsterling @samth @kamatsu8 @TaliaRinger Sean Wilson's Edinburgh PhD thesis was on applying proof planning (specifically ri… 2021-05-25 14:27:54 @jonmsterling @brendanzab @kamatsu8 @samth Maybe Daniel Selsam would though, that's worth trying 2021-05-25 14:27:22 @jonmsterling @brendanzab @kamatsu8 @samth Yeah, but they now seem totally uninterested in converging. I'm happy to translate some of the ideas over to cubical, I'm generally good at that kind of thing, but it helps to have experts on both sides of the bridge interested 2021-05-25 14:24:39 @jonmsterling @brendanzab @kamatsu8 @samth Interesting, I suppose that could lead to extra challenges. I think @BasspittersBs was open to talking more about this at some point, as was his student 2021-05-25 14:18:01 @jonmsterling @brendanzab @kamatsu8 @samth I've tried to talk to Leo de Moura about this, but he never seems open to taking to me I think he has some grudge about univalence 2021-05-25 14:16:40 @jonmsterling @brendanzab @kamatsu8 @samth I think they put more thought into efficiency, but also some of the extra design involved was due to the fact that they don't get hcongr_ideal, which you do get in cubical 2021-05-25 14:14:33 @jonmsterling @brendanzab @kamatsu8 @samth They just don't use the word in the paper: https://t.co/UsIxDxQijp. 2021-05-25 14:13:15 @jonmsterling @brendanzab @kamatsu8 @samth I think they do for congruence closure 2021-05-25 14:11:21 @EliSennesh אז אפשר 2021-05-25 14:10:10 @jonmsterling @brendanzab @kamatsu8 @samth It is, but we have the work in Lean to build off of, except our job is easier since our notion of equality & 2021-05-25 14:08:01 @jonmsterling @samth @kamatsu8 But not in its AI context because I had no idea any of this proof automation stuff was AI, I just like proof automation 2021-05-25 14:07:20 @jonmsterling @samth @kamatsu8 I'm tired, I may have mentioned it in QED at Large though 2021-05-25 14:06:26 @jonmsterling @samth @kamatsu8 Proof planning! Good keywords to look up 2021-05-25 14:03:29 @EliSennesh Also this is in Hebrew to stop people from being terrible 2021-05-25 13:59:52 @EliSennesh I'm a citizen, are you? I need an antibody test though once I arrive 2021-05-25 13:56:44 Why are humans like this https://t.co/UByvMrGtnC 2021-05-25 13:50:45 RT @hodgesmr: The problem with the “this meeting could have been an email” meme is every single one of you brags about your inability to ke… 2021-05-25 13:47:44 RT @Foone: So when you push down on the switch, the contacts touch. https://t.co/7NZaaZye0N 2021-05-25 13:47:41 RT @Foone: looking down the switches, they've got a circular section with a spring, and a cross-bar which holds the little contacts apart.… 2021-05-25 13:47:11 RT @Foone: I'm working on cleaning up my H19 terminal. So the keyboard is secured to the bottom of the case with 6 screws, and now it hinge… 2021-05-25 13:44:11 @emilymbender @alkoller My ML/PL nerd sense is tingling 2021-05-25 13:35:24 Everything is better with octopuses https://t.co/QEYHKmxDW7 2021-05-25 13:32:23 @codydroux Probably 2021-05-25 13:22:00 RT @ccshan: Everyone deserves a second chance. It doesn't have to come from me again. Not everyone gets what they deserve. 2021-05-25 13:19:11 @thatinterlace @XenaProject I talked about it and wrote about it until it was normal and not extremely traumatic. And I built a new life. It's "moving on" but it's absolutely not "forgiving" 2021-05-25 13:18:25 @thatinterlace @XenaProject I chose not to dwell, but to make peace with what happened, accept how terrible it was, and move on with my life, knowing that this is someone I should continue to avoid, and writing things down when most lucid so I can ensure I don't get manipulated later, or gaslight myself. 2021-05-25 13:17:01 @thatinterlace @XenaProject Forgiveness to me is trusting that it the same situation were to arise again, the person would not act the same way. Forgetting is just not remembering it. In this situation I chose neither 2021-05-25 13:07:32 @thatinterlace @XenaProject But I've found a lot of people's notions of "forgiveness" is much weaker than mine. For example, if you are not on speaking terms, or would not trust the other person in similar situations again, in my view you have not forgiven. 2021-05-25 13:06:17 @thatinterlace @XenaProject I found the biggest sense of relief when I decided not to forgive. Only then could I move on. 2021-05-25 13:05:03 @XenaProject Abusers often wait until you're useful to them again and then try to lure you in and discard you as soon as they perceive you as not useful, so it's important to remember everything. 2021-05-25 13:03:39 @XenaProject Abusive people don't deserve the benefit of the doubt without expressing very sincere remorse and commitment to change. The onus is on them to apologize and change, not you to forgive and trust. 2021-05-25 12:59:08 עד ה-13 ביולי 2021-05-25 12:53:51 RT @getsthegoods: 4 years ago the founder of proud boys made a video “10 things I hate about Jews” https://t.co/kY4WLyrrxZ 2021-05-25 12:53:36 RT @getsthegoods: The local paper even touched on the subject. https://t.co/ElwEInFII1 2021-05-25 12:50:11 אהיה בארץ בסוף יוני 2021-05-25 12:42:23 RT @jcreed: In this episode of "huh I wonder if this categorical construction is good for anything": Say objects are finite sets and morph… 2021-05-25 12:41:49 RT @ta_broderick: In Fall 2020, I gave the lectures for MIT's 6.036 "Introduction to Machine Learning" course. You can find all of my lectu… 2021-05-25 12:28:33 @lindsey @silentbicycle I do not! Why? 2021-05-25 12:03:37 RT @polikarn: Our work on "Cyclic Program Synthesis" has been selected as a distinguished paper at @PLDI'21! Stop by session 4A on Thu June… 2021-05-25 11:54:09 Keep detailed records of the abuse when you remember it so that years later when you're like "maybe that person wasn't so bad and isn't worth being afraid of and staying away from" you can look back and read and stop yourself from that mistake 2021-05-25 11:39:19 RT @dadakomachi: .... https://t.co/FXvVRrCCsE 2021-05-25 11:38:15 RT @JoshuaLambier: Troubling signs that "horrific" government cuts to UK universities in the post-pandemic period will aim at disciplines t… 2021-05-25 11:36:53 RT @StefanSzeider: Today I received my author's complementary copy of the Handbook of Satisfiability, 2nd Edition (now 2 volumes). Many tha… 2021-05-25 08:17:47 We are literally killing ourselves by sitting in those disgusting offices all day with windows closed and poor circulation and filtering all because of some old controversy over how viruses spread 2021-05-25 08:16:21 https://t.co/9bU7xmypfc 2021-05-25 08:13:10 Better Circulation 2022 let's do it 2021-05-25 08:02:29 RT @RedPandaEveryHr: https://t.co/JsZmv7kDty 2021-05-25 08:00:45 I think masks will help and people should wear them if they have to be out in public while symptomatic with a cold or the flu. But I think the majority of the effect was from not being inside of our disgusting buildings that recirculate each other's viruses 2021-05-25 07:59:35 To be fair this also involved literally not working or seeing friends in person and, for many people, not having any food security or hope for the future so https://t.co/hYbE6CiPRI 2021-05-25 07:32:47 Honestly kind of annoyed previous gen Jews in the US tried the "maybe if we just assimilate enough they'll like us" thing because we already tried that in pre-Holocaust Germany and it doesn't work. It's not working here either, it just gives people an excuse to hate us more 2021-05-25 07:26:08 @BAPearlmutter @samth Thanks I hate it 2021-05-25 07:24:20 This too, good opportunity to lose followers who don't believe in basic things like the right for Jews to not be scared of dying for existing https://t.co/MLoiCheCaD 2021-05-25 07:22:37 I lost followers for this in the morning and I'm glad because it's a way to lose actually antisemitic followers. So retweeting in hopes to lose more of those https://t.co/zz8F9oZmO7 2021-05-25 07:21:23 RT @moonriseTK: @TaliaRinger kinda like how in special reativity ppl disagree on what speed the other is going and how fast time is going b… 2021-05-25 06:59:32 @yisongyue Very cool! Good luck with the submission, looking forward to it 2021-05-25 06:57:36 @yisongyue (Feel free to sleep whenever, I'm just musing right now, in a mood to explore connections between fields and ideas.) 2021-05-25 06:57:02 @yisongyue OK, so very ignorant question, but can you fix the modularity itself as not a static thing but something that can vary too, and somehow learn that? 2021-05-25 06:52:43 In every path is new knowledge waiting to be found and understood 2021-05-25 06:51:53 You meet all of these people who each know about all of these little dots, and you from that can make a huge social knowledge graph and find paths between the dots 2021-05-25 06:49:40 @yisongyue Or is that what you are saying you can view metalearning as doing? I don't have quite the same vernacular since most of my exposure to ML has been in the past 6 months, so sorry if ignorant questions 2021-05-25 06:47:59 @yisongyue Can you learn the right modular structure in itself hahaha 2021-05-25 06:46:45 @yisongyue Ah, so there's a continuum there of how you can even define the problem 2021-05-25 06:45:08 @yisongyue Ah, that makes sense! I'm in a domain where I very rarely have a large amount of training data, but I do have a lot more structure, so I can often compensate with structure. I'll have to read! 2021-05-25 06:43:07 And it's beautiful because you can see all the other little dots now and make sense of them and move them around and the world is just so much more real and textured 2021-05-25 06:42:19 There's a sense in which finishing a PhD is like waking up from a long dream. You spend six years zoomed in on this tiny thing, and then you zoom right back out and you learn about the whole world, but you're equipped with this incredible beautiful knowledge about this tiny part 2021-05-25 06:39:07 @yisongyue You being the field in general, I guess 2021-05-25 06:38:54 @yisongyue Is this a kind of thing you can handle? Breaking down larger changes into possible sequences of smaller changes? 2021-05-25 06:38:24 @yisongyue On isolation: Sometimes I have too many things changing at once, and the problem gets so much more difficult. This is an outstanding problem in my own work, common when analyzing say Git repos---insufficiently granular history to break down large changes 2021-05-25 06:34:46 @yisongyue Cool, I'll read! I'm curious if any of the techniques I have can carry over to other domains, especially once you start getting partly symbolic. And of course similarly in the opposite direction 2021-05-25 06:30:17 @RemivanTrijp Ah I always think about it for its functional programming value 2021-05-25 06:24:07 Some people call this "god," some people call this "beauty," some people call this "math," I'm not sure it really matters 2021-05-25 06:20:18 Just, I think those connections, all of them, every single one of them, cannot be wholly explained from within the systems we are capable of defining and reasoning about within our universe. Just outside of our universe, in the metaverse 2021-05-25 06:19:07 And of course, in its fullest sense, it is fundamentally unknowable But I still think it exists! 2021-05-25 06:18:00 The more I learn about how everyone solves the same problems through different lenses and all of them are connected and you can translate between them, the more I believe in truth. But truth lies not in any one system, but rather in the connections and translations between them 2021-05-25 06:15:32 @RemivanTrijp Like what? 2021-05-25 06:13:44 RT @xnningxie: Excited to share that Perceus won a @PLDI 2021 Distinguished Paper Award! Having both a Distinguished Paper Award at POPL (… 2021-05-25 06:07:09 @RemivanTrijp Shoutouts to @zeeshanlakhani 2021-05-25 06:03:07 @almostmeenal I do clean because my housemates rely on me doing some of that, so maybe I can schedule that at night lol 2021-05-25 06:02:18 @almostmeenal Ah, I just don't do any of that stuff, unfortunately, like ever. I haven't unpacked from a trip a few weeks ago yet 2021-05-25 05:58:17 @AdrianoDAlessa3 But how can you motivate yourself to do any of those things they all suck 2021-05-25 05:57:47 @almostmeenal Good point my ADHD housemate is often up until like 3 AM 2021-05-25 05:57:05 @MarisaVeryMoe There are so many cool logics out there. FOL is one of my least favorites by far 2021-05-25 05:55:19 @MarisaVeryMoe I hope soooo modal logic is fun it's good for distributed systems 2021-05-25 05:54:22 How do ADHD people who sleep make the exciting hyperactivity go away at night once medicine wears off 2021-05-25 05:52:49 @HenkPoley Programming languages lol 2021-05-25 05:51:49 @RemivanTrijp I went to a Lisp meet-up in NYC to make friends in 2017 because I was lonely and wanted to meet PL nerds. It was ALL PL nerds. Super wonderful we ate pizza and got drunk on a roof near Central Park but still 2021-05-25 05:47:33 RT @drfeifei: @michellearning Congratulations Michelle! So proud of you! You did a fantastic job! https://t.co/Jp3nPf4T8g 2021-05-25 05:47:10 @michellearning Congrats!!!!! 2021-05-25 05:28:54 @SMT_Solvers Any equivalence you can define I've gotchu 2021-05-25 05:24:36 @HC_solomon ADHD is great when managed well 2021-05-25 05:20:06 Everything is so interesting and exciting! 2021-05-25 05:15:36 @ian5v Happy to talk about research at some point if they are interested 2021-05-25 05:14:29 @ian5v So cool I love this!! 2021-05-25 05:10:12 @TedPavlic Good points! Really good things to think about---there's a very literal sense in which higher dimensional paths are proofs by way of homotopy type theory. So it's likely work in each field can carry over to the other nicely 2021-05-25 05:08:00 @ian5v woahhh that's so cool how 2021-05-25 05:07:16 @rgivan1 Sounds good! I'll be slow with email until after my defense (June 4th) but am excited to talk to people about cool future ideas after that as I start building a group of students 2021-05-25 05:02:18 @rgivan1 Very interesting! I'd love to try it out, and read more about the foundations. Do you have reading about this? How expressive is the logic? What's the trusted part of the system? 2021-05-25 04:57:34 @rgivan1 Oh cool! If you ever want to talk about that I'd be happy after I defend 2021-05-25 04:56:26 @rgivan1 But that's enough for a lot of tasks, if you have a lot of data and don't mind confusing robustness bugs and can pour absurd amounts of money into energy efficient hardware. It's just silly and you're not always in that situation to begin wtih 2021-05-25 04:54:46 @rgivan1 And more importantly lots of good work going on right now. But yeah they'll merge back I'd bet on it. There is no reasonable sense of "generalize" in which ML tools actually generalize right now, they just fake it with a lot of data, very inefficiently at that 2021-05-25 04:53:32 @rgivan1 I agree and I think neurosymbolic is the future of everything, but it's going to take a few years for people to understand that the limitations of current technology is fundamental, though there are some good neurosymbolic papers out there already 2021-05-25 04:52:10 @kamatsu8 But I'm ADHD with a very big H 2021-05-25 04:51:46 @kamatsu8 Yeah sometimes it's that, sometimes it's not even conscious it's just distraction, like OH WOW I HAVE IDEAS and I have to share them with everyone and suddenly there's a project today I am hyper 2021-05-25 04:49:31 @TedPavlic I actually think there's even a sense in which path finding could be viewed this way, because there is a natural analog between paths and proofs, so a change in a datatype is just a change in the environment in which you are searching for a (higher-dimensional, in my world) path 2021-05-25 04:44:37 But I want more please talk to me if you care about automation & 2021-05-25 04:43:39 @TedPavlic ooooooh lifelong learning 2021-05-25 04:43:20 Yeah actually I have a bunch of ideas 2021-05-25 04:42:40 Do other AI/ML problems ever have things that work until the environment very slightly changes and then they don't oh shit OK this might actually be a cool way to think about robustness 2021-05-25 04:41:29 Are there obvious domains where if you look not at a moment in time but at change over time you gain a lot of information? How can tools learn from this information? What can they learn? What information does the change carry? 2021-05-25 04:40:58 Is this lens being considered already in other problems (besides program repair, obviously)? If so, what are they? How is the information about change used? 2021-05-25 04:40:19 Another thing I have been thinking about: my work (proof repair) reimagines proof automation to view proofs as fluid entities that change over time, rather than looking at them only at one moment in time. Move outside of proofs: What AI/ML problems could benefit from this lens? 2021-05-25 04:37:12 @rgivan1 Thesis: https://t.co/XlBfGkP8QZ 5 minute talk of most recent work, but for PL audience: https://t.co/p3P8i1vSNU 2021-05-25 04:36:16 @rgivan1 I don't know how much of my work you know, but I build proof automation in Coq that fixes proofs in response to change. I'm not used to calling it "AI" but there's probably a lens under which it is. It's weird because the logic is very different, I don't do any backtracking 2021-05-25 04:33:27 @rgivan1 But for "common sense" I don't think going symbolic makes sense, because I think "common sense" is precisely dealing with the counterexamples, whereas "mathematical intuition" and "structure sense" are symbolic 2021-05-25 04:32:38 @rgivan1 Yeah I think symbolic AI is complementary to rather than good for implementing "common sense." It's good for quickly generalizing the things that can be truly generalized, and getting strong guarantees, and understanding things, and needing few to no examples 2021-05-25 04:22:11 I didn't build the keyboard yet OFC but I'm obsessed I have to do it 2021-05-25 04:21:45 In case you're wondering what ADHD is like, 3 hours ago I set out email SIGPLAN-M mentors. Instead I: 1. decided to build a replica of the SAIL keyboard with custom bindings, 2. found a collaborator for an awesome project, and 3. emailed a colleague at UIUC about students. 2021-05-25 04:13:32 @SMT_Solvers Figure 25 2021-05-25 04:12:49 @SMT_Solvers Check out the vector type here for example: https://t.co/XlBfGkP8QZ 2021-05-25 04:12:05 @SMT_Solvers list < list < 2021-05-25 04:11:14 @SMT_Solvers What do you mean by parameterized types? Like polymorphic type parameters? polymorphism = types that depend on types dependent types = types that depend on terms 2021-05-25 03:44:11 @brendanzab @kamatsu8 @samth I think keeping them efficient will be one of (the only? once a type theorist is involved) the challenges of adapting them to cubical 2021-05-25 03:37:06 I think I'll be a good professor 2021-05-25 03:31:32 @kamatsu8 @jonmsterling @samth Yeah the first 2 are things I'm really good at. The last one I still struggle with or willfully ignore sometimes lol 2021-05-25 03:28:28 If any potential students are interested in the way SMT solvers and rewrite systems work, but also interested in dependent type theory, I have a really cool project that I think could revolutionize DTT, with really wonderful potential type theory collaborators already 2021-05-25 03:27:03 @kamatsu8 @jonmsterling @samth Actually I think I did the getting in touch part too because I'm too extroverted for my own good lol 2021-05-25 03:26:28 @kamatsu8 @jonmsterling @samth Yeah fair, I mean same for me, but honestly it was really hard sometimes 2021-05-25 03:24:01 @jonmsterling @kamatsu8 @samth I'm not on speaking terms with one of the egg authors either which makes things even harder, because that is the best e-graphs library, but if I can avoid talking to him I should be OK 2021-05-25 03:23:18 @jonmsterling @kamatsu8 @samth It has so far just been hard for me to find the right group of people with the right expertise. I want an e-graphs expert, and also a cubical type theory expert. I'm neither? I'm just really good at proof automation and at bridging concepts from different areas 2021-05-25 03:22:11 @kamatsu8 @jonmsterling @samth What this means is that congruence closure carries over to solve equalities but perhaps more interestingly, you can use the same automation to automate transport as type-directed search the same way people usually implement rewrite systems in simpler logics 2021-05-25 03:20:43 @kamatsu8 @jonmsterling @samth https://t.co/acIQpx23dn 2021-05-25 03:19:50 @kamatsu8 @jonmsterling @samth The real benefit you get is generalized congruence 2021-05-25 03:17:59 @jonmsterling @kamatsu8 @samth I'll email you or something 2021-05-25 03:17:49 @jonmsterling @kamatsu8 @samth Yeah OK so let's talk after my defense. This can honestly fit into my DARPA grant given a student, since I think I can reuse it in my repair automation, just externally 2021-05-25 03:12:36 @jonmsterling @kamatsu8 @samth But you need to tweak the notion of equality and carry proofs. I've wanted people to work with me on this for a while but it's a very significant undertaking, so if we can put something together I'd be very happy 2021-05-25 03:11:57 @jonmsterling @kamatsu8 @samth Path equality actually gives you the natural generalization of congruence to higher order dependently typed logics for free though so there is no theoretical barrier at all to reusing the infrastructure inside of SMT solvers 2021-05-25 03:10:31 @jonmsterling @kamatsu8 @samth I'm really good at proof automation, but I'm really bad at getting type theory details right, so I really need some type theorists on board 2021-05-25 03:09:52 @jonmsterling @kamatsu8 @samth Oh honestly my big barrier is that I don't feel comfortable implementing cubical. If you want to collaborate on e-graphs for type-directed automatic transport in cubical I would really love to, it would help me very significantly in my own work 2021-05-25 03:08:02 @kamatsu8 @samth Category theoretically I think when you move to prove a negation, you move into the opposite category, so initial things become terminal, and you lose information that way But the *fact* that you can do that is itself information so maybe the information itself is a red herring 2021-05-25 03:06:43 @kamatsu8 @samth Proof automation reusing information is my jam so I've obsessed over this for many nights 2021-05-25 03:06:10 @kamatsu8 @samth Interestingly, this happens in constructive proofs that go through by way of provable DNE over decidable domains. It remains to be seen whether the fact that DNE is provable over the domain is enough information to reuse in itself 2021-05-25 03:04:32 @kamatsu8 @samth But, if you look at proofs that go through by double negation elimination, even with explicit rather than ephemeral proof terms, information carried in the negation of the negation is not necessarily information that will help you in proving anything else 2021-05-25 03:03:14 @kamatsu8 @samth There are 2 orthogonal axes IMO. If you reify proof terms in Isabelle/HOL like with HOL-Proofs, you can get a lot more power for reuse & 2021-05-25 02:58:42 @kamatsu8 @samth But right now it isn't type directed search yet, mostly because I had to graduate. It should be. I need a student who cares about this lol 2021-05-25 02:58:05 @kamatsu8 @samth Turns out when you have proof terms, by externalizing transport you actually get to use it for repair too so that's cool 2021-05-25 02:57:20 @kamatsu8 @samth My PLDI paper actually implements a rewrite system based on exactly this, just not with e-graphs yet, and metatheoretical so you get the benefits in Coq 2021-05-25 02:56:27 @kamatsu8 @samth So effectively every single proof carries lots of example information you can extract and generalize. The same happens with changes over time. You basically never have to search for a proof from scratch, you can build off of existing blocks 2021-05-25 02:55:45 @kamatsu8 @samth Constructive logics also carry information you can reuse inside of other proofs 2021-05-25 02:55:24 @kamatsu8 @samth Cubical + e-graphs boom you're done 2021-05-25 02:54:45 @kamatsu8 @samth You get the same exact thing out of path equality in cubical, but nobody who likes cubical cares about automation 2021-05-25 02:53:36 @kamatsu8 @samth You can still reconstruct proof objects in Isabelle/HOL in an interesting logic though so idk why it matters 2021-05-25 02:52:20 @kamatsu8 @samth Yeah true. TBH I think automation is often much easier in constructive logics, and the fact that classical proof assistants are more automated is an artifact of history 2021-05-25 02:49:18 @samth AI is, that is 2021-05-25 02:49:04 @samth What even is intelligence though? I certainly wouldn't say any ML system today has intelligence. That's why it's a ridiculous name for a field 2021-05-25 02:48:17 RT @samth: @TaliaRinger Another trend was that no one knew how hard various problems were, so everything was initially AI. Then people disc… 2021-05-25 02:47:22 @kamatsu8 @samth Why it's just a different logical system who cares 2021-05-25 02:46:15 @samth @kamatsu8 How so? 2021-05-25 02:44:53 RT @samth: @TaliaRinger I think it's more that people came to see PL as something independent of AI. It's striking to read about the histor… 2021-05-25 02:44:50 @samth So weird 2021-05-25 02:44:17 @kamatsu8 @ChrSzegedy At UW it was just implementing Pacman plus some "wow FOL is just graphs" stuff and the whole time I was like "ugh why FOL though there are so many better logics" 2021-05-25 02:42:50 @kamatsu8 Fascinating 2021-05-25 02:42:07 @kamatsu8 @ChrSzegedy I have a weird amount of faith that neurosymbolic learning is the future of both PL and ML and will actually reunify the fields a lot, though I also don't want that because ML is overwhelmingly large and full of even stronger personalities than PL 2021-05-25 02:40:59 @kamatsu8 I mean, @ChrSzegedy seems to think it's still an intelligent thing for a computer to do. I actually think AI is breaking away from just ML and becoming kind of a merger again, as people open themselves up to AI by any means. I think it's a terrible name for a field, though 2021-05-25 02:36:58 @kamatsu8 Still not into FOL though so limiting 2021-05-25 02:36:10 @kamatsu8 Yeah there's no difference we are just lying to ourselves 2021-05-25 02:35:13 @kamatsu8 That's almost reassuring 2021-05-25 02:35:02 @kamatsu8 OK so funding cuts just lead to renaming fields that continue to be useful lmao 2021-05-25 02:34:13 @kamatsu8 Maybe closer to 1000 IDK the exact number 2021-05-25 02:33:37 @kamatsu8 I read thousands of papers about this, literally, and never once thought of it as AI 2021-05-25 02:33:11 @kamatsu8 Wow 2021-05-25 02:32:02 Help I'm having an identity crisis https://t.co/IRIxNtKEEw 2021-05-25 02:31:26 @kamatsu8 I think the ATP folks who do search with backtracking are probably closest to classic symbolic AI. But with absolutely incredible developments since 2021-05-25 02:30:28 @kamatsu8 I don't distinguish between proof assistants and PL at this point, so that's hilarious. So like the stuff I wrote about in QED at Large, a lot of it was historically considered AI? 2021-05-25 02:29:46 Were there PL labs before the 1970s? 2021-05-25 02:27:51 Looks like PLDI was born from a compilers conference. Still curious if a lot of AI folks migrated into PL, and of the history of POPL 2021-05-25 02:26:07 If so the irony of so many PL people winning Turing Awards doing work from a "dead" field is hilarious 2021-05-25 02:23:27 Also $\simeq$ 2021-05-25 02:21:05 @palvaro https://t.co/cTslxvxvrm 2021-05-25 02:20:14 The first POPL was what, 1973? Literally the first AI winter? Guys? Who has been around that long and can talk about this 2021-05-25 02:17:55 @hillelogram @johnregehr My keyboard will have HoTTkeys instead 2021-05-25 02:16:53 @supersat @smkmeetup After I have a paycheck lol 2021-05-25 02:16:38 @supersat @smkmeetup I ... if it's vintage... maybe 2021-05-25 02:16:01 @_pdarragh @taehatypes Ahhhh yes 2021-05-25 02:15:10 I mean many of us choose way more interesting logics these days than FOL which is kind of boring IMO but still 2021-05-25 02:13:17 Honestly after reading the history of the Stanford AI Lab, I'm now even more confused about what the difference between PL and symbolic AI is. Like Lisp was an AI thing? Really? Did we just rename huge swaths of Symbolic AI to PL to avoid the stigma of the AI winter? 2021-05-25 01:52:48 @kamatsu8 I'll make one that looks like this but is beautifully modern 2021-05-25 01:50:41 @supersat @smkmeetup I'll be a professor, I can have expensive hobbies 2021-05-25 01:50:20 Needs the latest typed lambda calculus and reduction rule Greek bindings and some HoTT keys 2021-05-25 01:49:24 @benno_stein I think I want to make my own though so I can update it to the latest typed lambda calculus bindings and reduction rule bindings, and add some HoTT bindings 2021-05-25 01:48:00 @benno_stein Should be shinier though 2021-05-25 01:47:49 @benno_stein Ahhhhhhhhhh want want want 2021-05-25 01:39:25 Hiiii who knows how to make keyboards and how do I start, I want to make a modern replica of this but update it to my key bindings but keep the absolutely beautiful color and shine and so on https://t.co/vspcZSMBlU 2021-05-25 01:37:27 Maybe I'll build a modern replica??? Who knows how to build keyboards??? 2021-05-25 01:34:03 Outing myself as completely obsessed with both Greek key bindings and vintage computers. Also this is one of my favorite colors 2021-05-25 01:32:55 How does anything get this beautiful https://t.co/bnTtNFiHa1 2021-05-25 01:31:32 It's so beautiful I want to cry. And it has such nice key bindings 2021-05-25 01:30:57 How much do I have to pay to acquire a @StanfordAILab keyboard? Do you all still have these anywhere? I'll donate to your lab for one 2021-05-25 01:29:04 @kal_muzaffer Good idea to bookmark this tbh 2021-05-25 01:27:15 @BjornerNikolaj The color is so beautiful too. So aesthetic omg 2021-05-25 01:26:30 I want this with every inch of my soul https://t.co/25SOSxLLtp 2021-05-25 01:26:13 @BjornerNikolaj It's a bit behind the times on all of the bound Greek letters, but I'd take it in a heartbeat 2021-05-25 01:25:23 @BjornerNikolaj As someone who is SUPER into vintage computers and SUPER into all of these keys I want this with every inch of my soul 2021-05-25 01:21:05 I still think I'm going with this solution from @supersat though because why wouldn't I https://t.co/sPwS3EkvsY 2021-05-25 00:17:28 @dj_hci @MSFTResearch Congrats!! 2021-05-25 00:02:08 @supersat Life goals https://t.co/OYoYcaG5tb 2021-05-25 00:01:27 RT @supersat: @TaliaRinger https://t.co/alEHUCgyBr 2021-05-25 00:01:26 @supersat Lmao 2021-05-24 23:45:30 @kczat @ChrSzegedy lol nothing gets a graduating PhD student excited like someone caring about their thesis work 2021-05-24 23:41:24 Oh man nothing gets nerds excited like keyboard bindings, love you all 2021-05-24 23:41:00 @kczat @ChrSzegedy I'll have graduated :) but happy to talk whenever after my defense, plus I have a grant with some UW folks so should probably be back on campus for collaboration in the future 2021-05-24 23:40:00 Still kind of proud of my 1 minute explanation of type equivalences & 2021-05-24 22:55:43 @benno_stein (honestly though as a professor I'll have my own office, so maybe that's a good time to get a mechanical keyboard) 2021-05-24 22:55:03 @benno_stein but then I'll have a mechanical keyboard and I'll need to type very loudly and also tell everyone I have a mechanical keyboard 2021-05-24 22:53:45 @Fish_CTO gotcha 2021-05-24 22:52:57 @arthurxavierx This is mostly for making slides, writing emails and Tweets, etc 2021-05-24 22:52:13 @Fish_CTO what is diamond 2021-05-24 22:51:40 thanks Google https://t.co/LZbwonbYX9 2021-05-24 22:50:10 @norootcause I get the words "turnstile" and "turntable" confused and it always sounds like some kind of DJ thing 2021-05-24 22:49:17 @calborean I don't need it much for my work 2021-05-24 22:48:07 @opdroid1234 What is that 2021-05-24 22:46:53 Also and 2021-05-24 22:46:01 RT @dj_hci: Honored to be one of the Microsoft Dissertation Grant recipients this year! Thanks, @MSFTResearch. https://t.co/2QbhmNnCSs 2021-05-24 22:45:56 I really need ∀, ∃, Π, Σ, Γ, ⊢, →, α, β, δ, η, ι, λ, ξ, and ω as permanent key bindings on my keyboard honestly 2021-05-24 21:46:50 RT @shirtpiggypig: interest check, located in seattle area: -1 technomotion pad that's only ever been used inside private homes (with bar… 2021-05-24 21:44:48 If I don't know you personally, then please send me an explanation of why you want to come, just to prevent Zoom bombing. I'm super open to inviting the community, I just also want to make sure the space is safe 2021-05-24 21:42:32 June 4th, 9:00 AM Pacific FYI 2021-05-24 21:42:21 @codydroux June 4th 2021-05-24 21:34:39 @MonniauxD Yeah, Zoom 2021-05-24 21:33:54 Who wants an invitation to my thesis defense? If so, DM me your email address 2021-05-24 19:11:20 BTW these movements are still pretty unpopular, I just think if you care the best thing you can do is support their growth 2021-05-24 19:04:36 https://t.co/qzNhOOD7Rd 2021-05-24 19:04:16 This is what the purple circle by my name is for, and honestly I find it works well as a signal that engaging in dialogue is safe, so people don't assume weird things about each other's politics when we bring up real problems 2021-05-24 19:01:42 But supporting these movements does not mean using them as token Good Jews (it is mixed Jewish Arab anyways), but rather just actually helping them do the things they want to do, amplifying messages, joining from abroad but mostly listening and helping, and so on 2021-05-24 18:59:22 This reaction is strongly justified by basically all of Jewish history, too. It's quite rational. So work with the Israelis who want change from within 2021-05-24 18:58:40 IMO the path to peace & 2021-05-24 18:52:06 @fractalherbs BTW I recommend @omdimbeyachad for a nice Israel internal movement for Jewish-Arab peace and coexistence 2021-05-24 18:23:41 @fractalherbs Yeah Hamas definitely means "kill all the Jews" when they advocate for one state, though most Palestinians I've spoken to mean "one democratic state" 2021-05-24 18:19:23 RT @fractalherbs: @TaliaRinger I think a lot of wasp American leftists don't understand Jewish history, and also, are displacing their own… 2021-05-24 18:19:11 @fractalherbs And I think OSS is so unpopular that I'm willing to compromise on TSS since it is at least better for Palestinians than the status quo. OSS would take like 30 years of intense education efforts on both sides and nobody has incentive to do that 2021-05-24 18:17:55 @fractalherbs And also because it ignores reality. I'd consider myself post-Zionist 2021-05-24 18:17:34 @fractalherbs Yeppppp same, though I don't even say "anti-Zionist" because that term is simultaneously used for people who want one democratic state one day, and people who actually just want the Jews to get kicked out of the entire region or killed 2021-05-24 18:04:36 Antisemitism as a whole is also worth reading about. I don't know a good source, for now here is Wikipedia which has its own sources: https://t.co/4gXKm3tldr 2021-05-24 18:00:38 RT @elfprince13: @TaliaRinger Racial politics in the US are f***ed in so many ways, but the treatment of Jews teleporting from you "we don'… 2021-05-24 17:59:40 RT @annolik: @TaliaRinger have you heard about this: https://t.co/xApUv7Pew4 ? (here is also some source in english: https://t.co/6urft3NoL… 2021-05-24 17:59:37 @annolik Yes!! For sure I know about that, it is awful 2021-05-24 17:58:16 Please learn about it before Jews start dying in the US again 2021-05-24 17:57:04 The idea that antisemites cling to is that we are superior in some respects (intelligent, powerful), but that this actually makes us inferior because we are evil, controlling, selfish, and manipulative. Thus our intelligence and power are just forces of evil 2021-05-24 17:55:00 It actually weaponized any success against us. It is much safer to be a nobody Jew than to be a rich, famous, high profile Jew in a position of power 2021-05-24 17:53:49 This also makes it square less with internal racism dialogue you are used to, because that dialogue paints other "races" as inferior in all respects, and leaves privilege to the majority oppressor white "race." But antisemitism never worked like this historically 2021-05-24 17:52:07 But since it is imported, Jews with recent European families get to be white in most contexts here, so we simultaneously benefit from the internal racial hierarchy, even though the guardians of whiteness also want us dead for existing. It's a weird place to be 2021-05-24 17:50:39 For example, the college interview was invented to keep Jews out of college in the US, since too many of us were getting in by test scores. So elite universities wanted to screen us out by interviewing for Protestant Character or some nonsense 2021-05-24 17:49:11 In the US, it's weird imported racism, which is why it doesn't fit into the conventional leftist American understanding of racism. This means it is less embedded in structures here, too, though not absent https://t.co/dXfewtqBZ3 2021-05-24 17:47:26 It just shouldn't be hard to be against all kinds of racism, and antisemitism is for sure a kind of racism, in that it casts The Jews as an evil, manipulative, selfish, cunning, and parasitic race intent on starting war and harming host countries for our own benefit 2021-05-24 17:44:54 The antisemitism on the American left right now is alienating in a way that antisemitism on the American right isn't, since I don't identify with the American right. Last time this happened, it pushed me out of leftist politics for years, due to a feeling of being disrespected 2021-05-24 17:15:49 @typer_uma I think I'm too gay for this, because my first thought was "because women are pretty and so are machines" Or maybe just gay enough to internalize and repeat the objectification? 2021-05-24 17:11:33 Was anyone else afraid of these https://t.co/BYbw6LdqX0 2021-05-24 17:08:55 RT @fun_ext: As a woman, do you think the name of the proof assistant Isabelle (because of which the tool is often personified as female) is 2021-05-24 17:07:14 I'm scared https://t.co/2rVsbRdvIi 2021-05-24 17:05:10 @ravenscimaven Everything about this scares me. Reminds me of Furby 2021-05-24 17:03:31 RT @disconcision: annual reminder that you're all pseudo-coders 2021-05-24 16:31:40 If your reaction to "please stop attacking Jews for being Jews" is "what about Israel" like half of the comments in this thread, you're a jerk to put it lightly https://t.co/TCxfSvBBLx 2021-05-24 16:28:34 @See_U_Soon_Boi @v1kramkr @LikeTheProphet @patskareshet @NFWMBBY_ @The_first_part @POTUS That's literally what everyone in this thread is doing, though. POTUS is like "hey stop attacking Jews" and everyone is like "but Israel" 2021-05-24 16:26:28 RT @POTUS: The recent attacks on the Jewish community are despicable, and they must stop. I condemn this hateful behavior at home and abroa… 2021-05-24 16:12:57 RT @CertoraInc: @wilcoxjay's lecture on mathematical induction https://t.co/C0YFWrNwJj shows the magic of formal verification. Induction u… 2021-05-24 16:03:09 RT @vardi: Save the trees! https://t.co/nbp5z0KEPG 2021-05-24 16:00:18 RT @tony_ferraro7: i wasn’t productive at all during the pandemic but it was actually my first pandemic so i think that’s fine 2021-05-24 08:14:04 In general, that is. Until all refugees worldwide are able to find homes and have free movement. This will continue. Forever. If you care, learn history. DM me please. 2021-05-24 08:13:12 That is a microcosm of the Israeli Jewish experience, one that white Americans who grew up far from war and genocide cannot begin to understand. It's also why war and genocide lead to more war and genocide and perpetuate endless cycles of human suffering. 2021-05-24 08:12:25 In particular, it isn't at all ironic 2021-05-24 08:10:47 DM me and I'll send you a link to my grandfather's autobiography, if you want to understand how someone can go from a Nazi death camp survivor to a proud Israeli soldier. I think that history is extremely important to understand. 2021-05-24 08:06:22 Also, sorry I disable comments on these, I don't want to be dogpiled, I'm sharing my frustration with the new generation of American and European Israel/Palestine experts who read a few Twitter feeds. 2021-05-24 08:01:21 can't* 2021-05-24 07:57:28 You were lucky to not grow up in a family of genocide refugees now sit and listen before you tell other genocide refugees what to do, thanks. 2021-05-24 07:56:48 Middle Eastern friends can keep doing whatever you want that's cool I just don't like white Americans and Europeans jumping into things without taking the time to understand them and applying frameworks that don't transfer and cause more pain. 2021-05-24 07:55:37 Peace will require different solutions, you can just copy paste your cute South Africa knowledge onto Israel/Palestine and expect it to work. Even though the injustice is at least as bad, the history is radically different, and that matters a lot to how different measures work. 2021-05-24 07:53:12 Like, you grow up to grandparents who fled death camps to the only country that would take them in after years of homelessness, in a war zone they didn't really know much about, and then you can take the moral high ground. Until then learn our history and think practically. 2021-05-24 07:49:35 So I'm sick of white liberal Americans and Europeans projecting western racial politics onto other countries FFS. Things can be wrong and still differently wrong. If you don't learn Jewish history, don't expect any ideas you think of to work. 2021-05-24 07:47:46 That is what makes this "complicated." It's not ethnosupremacy, it's fear rooted in millenia of oppression with a genocide many of our grandparents narrowly survived that exterminated most of the Jewish population in Europe, and so on. 2021-05-24 07:45:23 Any solution to Israel/Palestine that ignores the history of Jews and especially Israeli Jews (where most Holocaust refugees and mizrahi refugees went) as an oppressed ethnic group literally terrified of being exterminated (after a genocide that almost succeeded) will fail. 2021-05-24 07:27:54 Can anyone draw Lambek's theorem maybe I'll get a tattoo once I understand it better 2021-05-24 07:19:33 @samth @arachnocapital2 I'm surprised, what do universities get out of this? 2021-05-24 07:18:27 @johnregehr @IsilDillig Yayyy looking forward to it 2021-05-24 06:46:21 @hillelogram @arsduo @Meowtima I mean if you insist 2021-05-24 04:19:58 @larsr_h I'm also good at empathy, so often if someone forgets to say something important early on, but says it later, I catch on that they just accidentally omitted it and can figure out how moving it up would change the paper 2021-05-24 04:17:11 @larsr_h I'm very good, though, at understanding the contributions of a paper on a larger scale, how it fits into everything else going on, zeroing in on the right questions to ask about scope and limitations, and so on. So I have my place lol 2021-05-24 04:15:30 @larsr_h But, I am very strongly a "big picture" person, and very much not detail oriented. So I'm unlikely to catch small mistakes in proofs or syntax, unless the proofs are machine- checked. If there are a lot of human written proofs, I make sure one other reviewer is detail-oriented 2021-05-24 04:13:57 @larsr_h I can do about 5 per week with undivided attention, less time for papers I'm more familiar with. With that in mind, the lower the score, the more time I spend on the review making sure it is kind and constructive 2021-05-24 04:08:38 @larsr_h I think POPL has some outrageously high reviewer load IIRC, like 20. And an outrageously small PC. Never understood it. 11 seems like a very good number. 2021-05-24 04:06:14 Dinner after a week of hard work https://t.co/mPtUOYwURN 2021-05-24 03:56:03 I feel like I get more conscious every year. Is that a thing? I mean literally, in the "consciousness" sense of the word. I felt like I was not very present when I was younger, and every year I feel more and more aware of myself and my surroundings and my thoughts. 2021-05-24 03:52:21 I'm happy it worked out. I think you should treat my presence on the PC the same way. Rejections always suck, but the community is changing, and you can submit proof assistant work to PLDI. It will be a different paper, but you can do it. 2021-05-24 03:50:57 I know it gets super stressful thinking there is only one major PL conference per year for your work. I was so depressed when my POPL paper was rejected last year. I had no idea I could submit it to PLDI, but I saw some proof assistant people on the PC, so I decided to try. 2021-05-24 03:49:17 OK cool I'll be on PLDI PC next year, so if you don't make the POPL deadline with your proof assistant or applied type theory work, for sure don't feel like CPP/ITP are your only options! (Though CPP and ITP are also wonderful.) 2021-05-24 03:45:59 I mean 2021-05-24 03:45:50 CEO of a lab with a large formal methods contingent that can make the most of the Jewish population https://t.co/DgjBq3GjE6 2021-05-24 03:42:37 @brickware True yeah. I think first year I just teach a grad seminar in my research area, which seems not hard after writing a survey paper lol 2021-05-24 03:18:06 RT @PLDI: The PLDI program is now available! Short talks will be followed by posters / Q& 2021-05-24 03:13:17 Thank you PLDI for making Christmas not boring for once https://t.co/o1xZphjfSM 2021-05-24 03:08:32 Oh man I can do these over Christmas, OK I'm in, I hate Christmas as a Jew, it's so lonely and boring. Time to take a plane somewhere sunny and read PLDI submissions on the beach 2021-05-24 02:32:24 @xwangsd Depends when you start counting, since I served on CAV, but with a light load of 5 papers. Still 16 seems fine too, over 20 and I'd feel nervous 2021-05-24 02:30:13 @asajeffrey I've had a lot already, but I think being a woman proof assistant expert contributes a lot to that 2021-05-24 02:29:20 @JAldrichPL Feeling like POPL is your only option is really hard, when you get rejected you feel like the world is ending. So I have a strong reason to want to signal it is OK to send this work to PLDI 2021-05-24 02:28:33 @JAldrichPL If I serve on any PC, I think PLDI is the place I can best give back to the community, and help bring proof assistants more into mainstream PL (assuming the papers clear the bar) just by signaling it is OK to submit that work to PLDI 2021-05-24 02:27:30 @JAldrichPL So I already served on CAV but that was in January or something. Is that this year or last year? I'm happy to say yes to PLDI and then no to everything after that 2021-05-24 02:25:15 This is really, really sad. Glad @seL4Foundation exists. Definitely support them if you value a world of robust, secure systems with strong guarantees https://t.co/PW6etjk3i2 2021-05-24 02:24:18 RT @GernotHeiser: @CSIRO @Data61news This would be a desaster for #seL4, had we not set up the seL4 Foundation to minimise dependency on @C… 2021-05-24 02:20:41 @arachnocapital2 Probably not, I do tons of service work, but I'm not optimizing just for getting tenure 2021-05-24 02:20:08 I think I'll say yes, there is often a lack of proof assistant expertise on major PL conference PCs 2021-05-24 02:15:21 11 papers seems appropriately light, I did 5 for CAV and that work was much less familiar to me than PLDI work 2021-05-24 02:14:35 Should I serve on PLDI PC next year? On one hand, people warn against excess service during your first year. On the other hand, review load is only 11 papers, I'm not teaching in the fall, I want more proof assistant work in PLDI, and I'm OK saying no to all other PCs next year 2021-05-24 02:09:00 RT @ruthbenghiat: Agree. Authoritarians are continually testing boundaries and often are delighted they can commit crimes with little or no… 2021-05-24 02:08:30 RT @JamesFourM: Look, there are easier ways to kill someone, and Putin / Lukashenko have both used them before. If that's all this was, the… 2021-05-24 01:54:26 All of you also finishing a PhD during a pandemic are so strong and amazing, even though it would be great if we lived in a world where we didn't have to be 2021-05-24 01:24:30 @ChrSzegedy Thanks for sharing 2021-05-24 01:23:11 Wow OK just one thing left before I'm done with the whole PhD thing, need to make the defense talk and give it in two weeks!!!! 2021-05-24 01:17:20 RT @ChrSzegedy: Great video. Incrementally fixing proof of correctness after changing the program. https://t.co/AsNtyUc1qE 2021-05-24 01:08:47 @alpha_convert Paper: https://t.co/ot12Kr0wwP Or, for an extra adventure, thesis draft :https://t.co/XlBfGkP8QZ 2021-05-24 01:05:49 @ChrSzegedy Just self deprecating humor. It's positive energy I'm just also very hyperactive today haha 2021-05-24 01:05:24 Also, please join for the Q& 2021-05-24 01:05:01 Paper: https://t.co/ot12Kr0wwP 2021-05-24 01:04:39 @luna_lun_a And I'll have a Q& 2021-05-24 01:04:23 @luna_lun_a Check out the paper: https://t.co/ot12Kr0wwP 2021-05-24 01:03:51 @ChrSzegedy Also thank you 2021-05-24 01:03:38 @ChrSzegedy https://t.co/DXlJOiEaM6 2021-05-24 01:02:42 @alpha_convert Haha there won't be one, I have two weeks until my defense it's too much for me to handle. But I'm glad you enjoyed it! I'll try to make the defense talk good enough to share 2021-05-24 00:53:05 Serious ADHD energy in this also 2021-05-24 00:50:00 I won't write a 28:00 talk because my defense is in two weeks. Also, I mix "I" with "we" here because of all of the job talk and PhD defense stuff 2021-05-24 00:49:08 5:00 PLDI lightning talk video for Proof Repair across Type Equivalences https://t.co/p3P8i1vSNU 2021-05-23 23:58:34 @kamatsu8 Regrowth of the Israeli left would be the best possible outcome of this awful shitshow 2021-05-23 23:47:02 I guess that's one thing about trying to pressure Israel into doing things, it's strange that people view Israel as a coherent enough entity to be pressured into anything, forming a government has been impossible for years now 2021-05-23 23:45:24 Sincerely hoping for some representation from the Arab political parties in the majority government though, if anyone manages to form a government 2021-05-23 23:44:20 I think I just know too many academics https://t.co/q0VxlaTLiw 2021-05-23 23:40:02 I guess I'll find out during the 15000th election 2021-05-23 23:37:17 Is there actually an Israeli left resurfacing, or do I just know too many academics? 2021-05-23 23:16:44 RT @omdimbeyachad: מוסיפים עיגול סגול לשם המשתמש בטוויטר نضيف دائرة ليلكية لاسم مستخدم تويتر סגול זה הצבע של המאבק המשותף היהודי-ערבי נג… 2021-05-23 22:41:44 @etrolleybus Solution: paste the entire paper onto the poster board 2021-05-23 22:23:59 @Ragazaga5 Thanks 2021-05-23 22:19:48 "Also, everything is due a month in advance. We are optimizing for people attending the conference and not at all for student authors" basically 2021-05-23 22:15:42 @Ragazaga5 lol I hate it so much, but I also need to move on and do other things 2021-05-23 22:10:21 Almost got away with being terrible at posters, too, until the pandemic hit and PLDI was like "OK you also need a poster not just a talk" 2021-05-23 22:08:08 I'm good at talks, good at writing, good at programming, good at cool new research ideas. But posters. Posters, I will never be good at. https://t.co/vXyQKm2UU5 2021-05-23 18:41:58 RT @matvelloso: Eeeeeeeeeeee https://t.co/2720DgmO5t 2021-05-23 18:09:57 Different cicadas arise after different numbers of years, but have evolved such that those years minimally coincide. The result? Prime numbers. https://t.co/h7p9cJJaeY 2021-05-23 18:03:11 @CheeToS_ I'm sorry, that sounds really awful https://t.co/CQzlmHOBTi 2021-05-23 18:00:25 RT @TorrensJonathan: Hi! I’m a love courier. Just here to drop off some love. No payment necessary. Just please accept this love. For you.… 2021-05-23 17:56:37 @nbouscal Ugh 2021-05-23 17:54:14 But my bed is so comfortable 2021-05-23 17:37:00 RT @BriannaWu: Gotta say, @potus set a goal of getting at least one vaccine shot into 70 percent of Americans buy July 4th, and it looks li… 2021-05-23 17:34:36 RT @WholesomeMeme: https://t.co/vTAe6JTABh 2021-05-23 17:25:22 RT @yuvalpi: לא מקריפ כהוא זה https://t.co/1VzIgcsmEp 2021-05-23 17:18:49 RT @nancylcoleman: of the many layers of bad in this story, this particular botched photoshop job has gotta be the worst https://t.co/QibD1… 2021-05-23 17:11:32 RT @CJHandmer: 3D printed 3e7:1 scale bacteria flagella electric (not magnetic) motor. 48000 carbon atoms. From https://t.co/AwAOp7jQBY. ht… 2021-05-23 17:06:08 RT @AnAngryOpossum: Fucking finally https://t.co/Po5e5Uv4go 2021-05-23 16:51:06 RT @PaleoTesla: Unpacking some boxes & 2021-05-23 16:50:14 RT @holly: "So... you gonna eat all of that???" #HollysMankyMoggies https://t.co/QtCCIPq70W 2021-05-23 16:27:45 RT @annehelen: Here we go: https://t.co/fdL1CvYo6a 2021-05-23 16:25:42 RT @NerdPyle: The week we brought Darby home #dailynocrop https://t.co/mvPk99LwNG 2021-05-23 16:19:59 @mohomran @yoavgo @h_jackson_ I appreciate the correction. Just please don't be a jerk about this. I just didn't want an undergraduate to get harassed, that is all. 2021-05-23 16:17:59 @mohomran @yoavgo @h_jackson_ Probably reading the book would have been a good call, but I was exhausted from literally having just finished and submitted my thesis that night, so I just decided to protect her preemptively whether or not it was accurate. 2021-05-23 16:13:19 @mohomran @yoavgo @h_jackson_ I saw this play out with Timnit and Meg and, by proxy, with me when I stood up for them at all. I was terrified that negative comments about a paper by an undergraduate woman would be taken out of context and used to attack her. 2021-05-23 16:11:30 @mohomran @yoavgo @h_jackson_ The reason I didn't blame the undergrad is because of the power dynamic involved, as incoming faculty, of blaming an undergrad for basically anything, especially publicly on a website where harassment is common, and especially for a woman in CS. 2021-05-23 16:07:36 @mipcodes @yoavgo @ejay_arr I think that only works because Israel has control over the Golan 2021-05-23 07:39:49 That also gives some Palestinian numbers higher than what I had heard, and it's super likely my sources of Palestinian death numbers were in fact biased. So including those for balance https://t.co/kKltTgFSHF 2021-05-23 07:37:46 Ah OK this helps make sense of it, thankfully! Hopefully an easy fix or clarification in the article https://t.co/kVDAIO7o7k 2021-05-23 07:29:12 @mohomran @yoavgo @h_jackson_ Ah, thanks! That gives useful context for how this could have been misinterpreted! 2021-05-23 07:25:33 @QuiteVague For... but fur is an amazing typo 2021-05-23 07:25:21 @QuiteVague Oh fur sure dogpiling is real and terrifying 2021-05-23 07:24:45 I don't have answers, but if you feel down, or on edge, or traumatized, I'm here for you and can talk directly about the things that are making you feel that way 2021-05-23 07:20:57 So I think we should ask ourselves right now why we have run out of trust and energy? What do we need to do so that we can heal, and trust the people who are actually trustworthy? How do we recharge after a four year shitshow that culminated in mass death and unemployment? 2021-05-23 07:17:09 But like it's OK to slightly disagree with people and still try to approach things from a collaborative frame when you have reason and energy to trust them. It seems like people have run out of all reason and energy to trust each other. Big red flag, needs immediate healing. 2021-05-23 07:16:11 If someone really wants you dead, or thinks you don't deserve rights, or is a malicious troll, for sure don't talk to them. Those people exist and they're extremely scary. If you need a break from emotional labor, for sure don't do it. You don't owe anyone anything. 2021-05-23 07:14:16 For whom are you willing to give up some of your own goals and work toward something that the other person wants? 2021-05-23 07:13:29 Which school do you want to work at, after those interviews are done? 2021-05-23 07:13:10 It's like entering two research 1:1s at two universities. In the first one, the professor you're talking to spends the whole time trying to convince you your work is garbage. In the second one, the professor you're talking to tries to find common ground for a collaboration. 2021-05-23 07:11:15 The distinguishing characteristic of these conversations is that people approached them in an attempt to find common ground and move toward better shared goals, rather than whatever it is Americans use political discussion for (I don't know, I never felt normal in this country) 2021-05-23 07:09:34 I said "civil" actually but I think that is a bad word. Civility is overrated. I've seen conversations that don't end in accusations of other people being terrible people. I think it's perfectly fine to get heated when the topic is something human life depends on 2021-05-23 07:07:59 I think some of this is a manifestation of trauma both from Trump and from the pandemic. I also recommend finding healthy ways to cope with trauma if you feel on edge all the time these days. As someone with PTSD I am happy to help you with that if you wish 2021-05-23 07:04:31 Some people are really malicious or really want you dead, but probably those aren't your friends who agree with you 99% of the time and disagree 1%. Probably those are people you should talk to 2021-05-23 07:03:50 Also this dynamic of the left fighting itself to death is what killed the left in many other countries and let the right thrive, so I think this should be a huge red flag if we don't want 50 years of Trump or whatever 2021-05-23 07:00:23 Americans who believe in a single democratic state, that is. Like even disagreeing with the means to achieve it was an opportunity to yell at me or scold me. I don't know why, but we are not friends anymore. And they do not even have any skin in the game, which confuses me more 2021-05-23 06:59:15 But Americans have yelled at me and called me a bad person for saying "hey, I actually want a single democratic state eventually too, which is a pretty fringe opinion for an Israeli-American, but given Jewish history I think BDS is going to have the exact opposite effect" 2021-05-23 06:57:53 As an example, I've seen absolutely civil and wonderful conversations this past month between---get this---Israelis and Palestinians. Like ones that gave me actual hope 2021-05-23 06:55:35 If you can't talk to people who align with you 99% politically, things are going to be very bad 2021-05-23 06:54:57 It feels like if I disagree even slightly about the means to achieving the same goal as other friends who politically align with me, I sometimes get yelled at and lose friends. That's weird and should be extremely worrying to everyone 2021-05-23 06:54:12 I think the increasing amount with which Americans cannot talk to each other about politics is terrifying to me. And I am not just talking about left versus right, or far left versus moderate left, or whatever 2021-05-23 06:49:03 The future is going to be weird, if we survive long enough to see it https://t.co/SoNEZ951dx 2021-05-23 06:26:41 @FelixGoldberg1 @yoavgo @ejay_arr https://t.co/bchtPjPKUE Just a good read for seeing details skimmed over completely. National narratives are often told by omissions 2021-05-23 06:25:08 @FelixGoldberg1 @yoavgo @ejay_arr TBH my main takeaway from that book was "wow the UK really knows how to fuck things up" but it is a good read. I think it is in Hebrew and Arabic too though I read the English one 2021-05-23 06:23:42 @FelixGoldberg1 @yoavgo @ejay_arr Yeah though I've read Side by Side and the Israeli textbooks also omit a lot of important details, which just lets both national narratives spin out of control for average people (academics seem to know more) 2021-05-23 06:20:28 @FelixGoldberg1 @yoavgo @ejay_arr I don't think the Palestinians are really in a position to offer anything, or have a coherent enough government to do that even they want to 2021-05-23 06:19:28 @yoavgo @ejay_arr Get a good government and maybe it will work, though 2021-05-23 06:18:53 @yoavgo @ejay_arr Depends who is in power, but if it's more people like Netanyahu, I really don't think peace is the goal. I think he legitimately wants everything, hence the high tolerance for settlements slowly choking the WB 2021-05-23 06:16:01 @yoavgo @FelixGoldberg1 @ejay_arr Yeah but it's hard to get good will when you offer them the shit end of the stick 2021-05-23 06:14:05 @yoavgo @ejay_arr Because Israel would basically be able to boss them around as desired. And I don't trust the Israeli right wing with that power at all. Maybe if you manage a reasonable government at some point but seems hard on election number 5000 lol 2021-05-23 06:11:35 @yoavgo @ejay_arr You can ask for neutral third party intervention, but who and how? 2021-05-23 06:10:43 @yoavgo @ejay_arr I think it might be hard for Palestinians to accept a deal that cedes control of any of that to Israel, or even trusts Israel on that, also given recent history 2021-05-23 06:09:38 @yoavgo @ejay_arr What resources do they have? Will there be any reparation to help people harmed by war and displacement? Will Israel block aid efforts that might be used by bad actors? Will they have access to the whole fishing zone? An airport? Will they get to control the crossing to Jordan? 2021-05-23 06:03:13 @yoavgo @ejay_arr It would backfire in about 20 years, I think, unless you create a very clear path to autonomy and economic prosperity 2021-05-23 06:02:32 @yoavgo @ejay_arr Yeah it just sounds a lot like US "peace" with Russia, and it turns out people get really sick of living with ridiculous puppet leaders like Yeltsin in countries with little hope, and friends who only like them so much as they are weak 2021-05-23 06:00:41 Hoodie https://t.co/i2mSfL9bWW 2021-05-23 05:56:32 @yoavgo @ejay_arr Why does Israel get to be militarized and not its immediate neighbor? It really feels like a proposal not for two states but for a second Israeli controlled state, so I'm not surprised Palestinians are largely against it these days 2021-05-23 05:55:13 @mipcodes @yoavgo @ejay_arr Like if you trust people to keep promises, sure. I don't think the history here is indicative that anyone is going to be up for keeping promises 2021-05-23 05:54:41 @mipcodes @yoavgo @ejay_arr This didn't work well for Israel when the water was completely diverted last century and Israel had to annex the Golan 2021-05-23 05:44:49 @yoavgo @ejay_arr I think it's better than the status quo, especially if it means getting the settlers out of the WB, so I'd accept it if it happens. But I think it would just lead to the same exact dynamic, except that it'd be called a state and not a territory 2021-05-23 05:43:53 @yoavgo @ejay_arr Is secure travel free travel? How would a Palestinian state in WB and Gaza have any hope at, say, controlling their own water supply? Border security and autonomy? Freedom of movement? 2021-05-23 05:39:53 RT @bariskasikci: We (Michigan CSE) have a lot of reflection, fixing, and healing to do. The onslaught of these events has been draining fo… 2021-05-23 05:06:30 @jeffbigham @LocalElite Cool I'm so excited to run Boston this October despite the whole pandemic thing killing a year of training 2021-05-23 05:04:39 RT @aaron_turon: This one is so hard for me. 2021-05-23 05:04:00 Me when I hear internal biphobia https://t.co/nN6VOg6W4n 2021-05-23 04:58:18 @akravc My secret plan to keep everyone's hair silky smooth 2021-05-23 04:57:43 RT @akravc: sharp need for PLGoogle. searched for "proof repair" on regular Google and all that came up was shampoo 2021-05-23 04:50:13 RT @mathcolorstrees: I kept getting the same results. And if anyone asked me, I’d have said his solution wasn’t going to change abuser beha… 2021-05-23 04:49:57 RT @mathcolorstrees: His idea was that if victims did not see the abuse and did not respond to trolls, the trolls would stop the harassment… 2021-05-23 04:46:21 @yoavgo @ejay_arr Sorry that's long, just something I think about very often these days 2021-05-23 04:32:17 @yoavgo @ejay_arr (Which is not to say there is no way to do that, just that I really don't know and it's hard to find good information on this. So if you have information, I'm happy to hear it) 2021-05-23 04:29:50 @yoavgo @ejay_arr I also wish I could in the meantime do something to improve the living situation in Gaza, because I think that benefits everyone except Hamas and the Israeli right. But I don't even know where I could send aid and not worry about it being used to fire rockets at my family 2021-05-23 04:24:40 @yoavgo @ejay_arr It feels like every real solution ends in civil war, and everything that feels like a solution isn't a real solution. 2021-05-23 04:24:12 @yoavgo @ejay_arr But I also think that would take like, 30 years of intense peace and education efforts, and no politician actually has incentive to do something like that. 2021-05-23 04:23:34 @yoavgo @ejay_arr I used to like the two-state solution until I actually thought about logistics. I am wondering lately if there is a one-state solution that actually gives some autonomy and special protections to all of the indigenous ethnic groups, while recognizing the geopolitical reality. 2021-05-23 04:21:13 @yoavgo @ejay_arr I still don't understand how you can put two states there, though, like logistically. Like where do you draw the borders? Geopolitically it seems impossible for one state to not control, for example, the water supply of the other. 2021-05-23 03:49:26 @SMukherjee89 Important to keep your face young during your PhD 2021-05-23 03:36:21 @SMukherjee89 It's for my face, keeps my skin nice and young now that I get wrinkles 2021-05-23 01:31:58 Need to pay and moisturize my students 2021-05-23 00:48:59 Talia things no context https://t.co/w2L1Ui6gf7 2021-05-22 23:20:57 That is for when I get past moderates and into the right usually. It is a game and I'll play it. Doesn't mean it's fun, but sometimes I have to play it, and I think I can play well, though I mess up sometimes 2021-05-22 23:19:10 I don't like compromising with people who, for example, might even secretly want me dead or want me to not have rights, but I'm willing to do it because I think I can outplay them late in the game 2021-05-22 23:18:10 I have trouble relating to people who would rather not get there at all if it means compromising on the way there for a while. I am far left by also extremely practicalist, and when I talk about politics it is always from a practicalist lens 2021-05-22 23:16:13 I'd rather get 50% to leftist ideals than not there at all by not compromising. That gets me closer to what I want, and I can then push by other means for the difference later 2021-05-22 23:15:16 I kind of think political spectra need to distinguish between ideals and practical views. Like I have far left ideals, but ultimately am willing to put those aside because I think compromise and dialogue with people who are moderate is necessary to achieve leftist goals 2021-05-22 11:22:01 Another thing I forgot: I didn't write the proportion of Palestinian deaths who were civilians because I don't know the number 2021-05-22 11:14:19 Oh I should note also the outrageous lie does not come from the undergrad author. It's in published work that is cited in the paper. I want to make sure people don't get upset with her for that. The published work that is cited is where it went wrong 2021-05-22 11:06:11 @h_jackson_ @MIT @nytimes I know this is not your fault because the paper you cited is the one with the incorrect numbers. It should be easy to fix, though 2021-05-22 11:05:08 @h_jackson_ @MIT @nytimes Notably, if you fix this to include the real number of deaths, it doesn't really hurt your paper, since it's still disproportionate. But it does avoid perpetuating untruths about Israeli deaths, which is important for respect, credibility, and academic integrity 2021-05-22 11:03:07 @h_jackson_ @MIT @nytimes No comment on the rest of the work, and it seems important to study. But the cited number of Israeli deaths during the second Intifada (87, mostly military) is much lower than the real number of Israeli deaths (about 1000, over half civilians) 2021-05-22 10:53:34 RT @yuvalpi: I am THRILLED to announce that starting October, I am joining the CS department at Ben-Gurion University as a Senior Lecturer!… 2021-05-22 10:34:14 Good timing for this gift from my mom to arrive https://t.co/hSEXQnwDpe 2021-05-22 09:50:28 aspell is calling me out https://t.co/G41p2dsLoO 2021-05-22 09:40:49 A good thesis is a done* thesis A good Talia is an asleep Talia * mostly, pending committee feedback and feedback from friends these next two weeks https://t.co/0OmgIPaiUM 2021-05-22 09:38:08 OK, it's stable now. I'm not going to touch this except with external feedback. 2021-05-22 09:03:25 @Adel_Bibi Like I trust the numbers you're saying, just any number I put there would have been me making it up because I didn't have a good source on hand to use 2021-05-22 09:02:47 @Adel_Bibi I legitimately just don't have a good source of Palestinian civilian casualties 2021-05-22 08:56:14 There are already cute QTs from people who don't know me and haven't seen my takes on Israel/Palestine recently aiming for a single democratic state one day so like maybe just try not assuming that someone who doesn't like people lying about Jewish death hates Palestinians? 2021-05-22 08:54:56 Some sources of Palestinian numbers say 4000. I don't know, if that's true, my bad! I was just correcting the outrageous lie that only 87 Israelis died and basically all of them were military. 87 is pretty far from 1000 2021-05-22 08:53:14 @Adel_Bibi The paper was pretty much correct on the Palestinian numbers which is why I didn't bother correcting them? But it's cool tell yourself I don't care about Palestinian lives just because I don't like people lying about Jews 2021-05-22 08:51:43 @Adel_Bibi Except I was correcting a research paper that told an outrageous lie saying that 87 Israelis were killed and almost all of them were involved in offensive operations. Go back in the history of my Tweets for my views on Israel/Palestine, they are pretty far left lmao 2021-05-22 08:22:53 There's still a disparity! Just 3-to-1 and not 43-to-1. 2021-05-22 08:20:58 For people without enough context to detect Yoav's sarcasm, the second intifada killed about 1000 Israelis and 3000 Palestinians. Most of the Israelis were civilians. https://t.co/ii8od6yEcp 2021-05-22 08:03:01 @yoavgo TFW going out to eat at a restaurant is an offensive operation. What exactly are you all doing at those restaurants? 2021-05-22 06:30:50 @banazir But the results are super weak and it's hard to fit into a thesis. Chapter 4 is like "OK now I know a lot more, here's something really fucking badass" and I don't have to change it much 2021-05-22 06:30:08 @banazir I'm supposed to finish the draft today lol. Chapter 4 looks easy though. Chapter 3 was hard to fix because it was a vision paper saying "guys I bet this is possible" which everyone thought was wild to even try at the time IIRC 2021-05-22 06:14:43 @banazir 3 more chapters to revise, but only 1 difficult one. Chapters 5 and 6 are just related and future work really 2021-05-22 06:10:56 Gosh all of Chapter 3 is updated. That was the worst chapter at least 2021-05-22 05:45:10 3.5. This is slow I want to sleep lol 2021-05-22 05:08:28 3.4 updated~ 2021-05-22 04:41:42 Up to 3.3 is updated now. 3.4 onward still changing 2021-05-22 04:39:28 I can tell I'm tired because I just had a really, really serious thought while totally full of awe and the thought was exactly this: "Wow! Has anyone ever thought about soy milk?" 2021-05-22 03:46:16 @nikitab @supersat Writing PL papers in word sounds like a nightmare 2021-05-22 03:44:08 @elfprince13 I actually don't even track visitors to my website. Feels invasive 2021-05-22 03:43:51 @elfprince13 Lol why would I have? 2021-05-22 02:18:01 RT @mai_elsherief: I am delighted I will be joining @ucsd_cse as an Assistant Teaching Professor this summer! I will be building a research… 2021-05-22 01:48:37 If you find typos and so on, or if you're confused about anything, I'd be really happy to hear about it at this point in the process. I have the worst headache so it's really hard for me to do that kind of proofreading anyways. Definitely don't feel bad reporting typos lol 2021-05-22 01:47:10 I am just going to keep this as a live working copy for now 2021-05-22 01:46:38 Hey, I dropped a draft of my thesis on my website: https://t.co/XlBfGkP8QZ Chapters 3 onward will change more today, and the whole thing will change a bit over the next two weeks. But if you feel like getting an early preview, I'd super love feedback! 2021-05-22 01:19:20 @supersat Did you really? I can't imagine doing that 2021-05-22 00:51:17 Go to LaTeX jail Do not pass go Do not collect $200 2021-05-22 00:49:51 Broke: updating your LaTeX version so you can write \textbf{\textsc{REPLica}} and have it do both Woke: \contour[2]{violet}{\kl{\textsc{REPLica}}} 2021-05-21 18:46:14 RT @WomenInLogic1: NEW PAPER ALERT "What is the Meaning of Proofs? A Fregean Distinction in Proof-Theoretic Semantics" by Sara Ayhan, Jo… 2021-05-21 18:39:37 @BodunHu I think this is about my limit, I'll crash tomorrow and sleep a lot 2021-05-21 18:29:34 @theshawwn Also a conference poster and talk are due today but I already bugged the chair to let me send them over the weekend instead 2021-05-21 18:28:55 @theshawwn Thesis is due today, defense is in 2 weeks 2021-05-21 18:23:05 20 hours of sleep in the last 4 days. Ouch 2021-05-21 18:06:30 RT @allen_ai: AI2 is actively researching the potential for toxicity in these models, and the possible mitigations – learn more about how q… 2021-05-21 17:49:31 @palvaro Nice 2021-05-21 17:49:21 RT @palvaro: that last one was too subtle, sorry. I got tenure you trifling bitches 2021-05-21 17:48:46 RT @GlennF: US is now back to June 22/April 25, 2020 levels of COVID cases being reported. And the drop in cases keeps on going. https://t.… 2021-05-21 17:42:45 Like why not Amazon Pi Google IOTa 2021-05-21 17:41:40 People in other fields naming everything after lambdas but meanwhile we've moved on to all of the other Greek letters smh 2021-05-21 17:38:51 RT @nelhage: How to build LLVM from scratch in 90 seconds, using Amazon Lambda: https://t.co/Ljzk5GPFXx 2021-05-21 17:19:18 Ugh https://t.co/hCR494jqXw 2021-05-21 09:50:44 @BodunHu @kamatsu8 Man I wonder how many of my friends are spies at this point in my career 2021-05-21 09:46:54 @BodunHu @kamatsu8 https://t.co/GrE2MDezR0 2021-05-21 09:46:25 @BodunHu @kamatsu8 If you're lucky, it's the CIA and not just a random spammer! 2021-05-21 09:28:46 It's funny how many people on Twitter are like "I always thought I was NT but this describes me am I ADHD too??" Like yes you're on your dopamine flashy social light box at 3AM impulsively sharing all of your thoughts publicly you probably are 2021-05-21 09:25:50 I don't even know if they're standard in ML TBH. But they're much less clunky than "human-in-the-loop" and much more specific so maybe we should have something cute and simple like this for proof automation too 2021-05-21 09:24:15 BTW in the proof automation world we don't use these terms because we like to pretend we're not doing AI, we're strictly PL & 2021-05-21 09:18:40 Even better is if the machine can communicate why it's struggling, or figure out what it needs to ask for!! 2021-05-21 09:14:11 It's funny, but nowadays I'm getting approached by a lot of people who do proof automation with constraint solvers. And they talk to me because they actually want a way to insert humans back into the loop when the machine gets stuck or does something weird 2021-05-21 09:11:26 Applies for anyone doing automation, whether or not ML is the technique for doing so. I personally like always having the human in the loop, even when I have strong formal guarantees about my tool's behavior. Humans and machines make great teams https://t.co/5RoyBf7P4K 2021-05-21 09:02:05 RT @scarlett_jazmin: Just say your racist and don’t want a Black woman to be tenured. 2021-05-21 08:56:36 RT @Dr_LPrendergast: Using the term 'resilience training' as part of #MentalHealth and #Wellbeing points to greater systemic issues. One sh… 2021-05-21 08:52:15 @RoxanaDaneshjou Out of genuine curiosity, what's the distribution of the world population for skin tones? 2021-05-21 08:06:26 @paulg He could be successful in the UK https://t.co/XBJovWuElY 2021-05-21 08:05:30 RT @paulg: "I could be a chef in an all-toast restaurant." — 12 yo on his cooking skills 2021-05-21 07:18:34 Then I'll have two weeks to write and practice my defense talk. I hope it won't be too hard just reframing the job talk and adding some technical meat 2021-05-21 07:17:53 It looks like my thesis will be on time tomorrow despite the infection in both eyes and so on, which made staring at a computer screen this week kind of hard. Looks like I'll spend all day Saturday cleaning, and all day Sunday turning in my PLDI talk and poster (late) alas 2021-05-21 06:38:10 RT @ZJAyres: Go about your day and do all of the things, but also have horrendous period pain, but push on through the pain and act entirel… 2021-05-21 06:31:39 Oh and my coauthors are in the intro proper, in the reading guide subsection, since I thought they belonged in main text. 2021-05-21 06:30:00 RT @algebraicyclist: STOP constructing galois representations -GROUPS WERE NOT MEANT TO BE UNDERSTOOD -DECADES of normalizations yet NO REA… 2021-05-21 06:21:33 @va2lam "Impulsivity" is a symptom and people consider it bad societally, but it's that very impulsivity that pushes me toward problems people are sometimes afraid of 2021-05-21 06:20:08 @va2lam I think this makes sense as an NT vs. ADHD thing honestly. People say ADHD makes us more creative, but I think that just means less inhibited. I think NT people have a lot of good ideas but they worry too much about consequences 2021-05-21 06:19:16 @va2lam Related work, on the other hand, came with huge motivation overhead until I wrote a survey paper I could draw from lol 2021-05-21 06:17:40 @va2lam Oh I honestly like future work stuff more, if I get to talk about fun vision it's like I'm suddenly writing a manifesto. Not even for the planning part honestly, I like outlining exactly how to do the thing. I generally just like inspiring research directions and leading 2021-05-21 06:14:59 @va2lam Ah I have no motivation overhead for creative things that I'm excited about. Like writing, for example. I just sit down and words come out 2021-05-21 05:38:42 https://t.co/HVJCVhFPj0 2021-05-21 05:37:47 Spent this whole afternoon writing my acknowledgments section of my thesis, which is long https://t.co/hqUVuNTPml 2021-05-21 05:25:39 @kamatsu8 I'm pretty sure even for things they don't enjoy doing, they can often just kind of do them without spending hours convincing themselves to do them 2021-05-21 05:03:58 @_snek93 Yeah I am 2021-05-21 05:02:05 Can I embed Hebrew into my thesis easily? Savta doesn't read or speak English but I'd like to acknowledge her 2021-05-21 04:40:24 yesssssss ADHD feels https://t.co/1G1yuJL3B5 2021-05-21 04:39:45 @nikitab @__apf__ What is that? 2021-05-21 04:28:12 I have really high motivation overhead for getting ready to go to sleep, and honestly sometimes stay up late because I really hate getting ready to sleep (washing up, taking out my contact lenses, and so on), not because I want to stay up (I love sleep) 2021-05-21 04:26:43 Why am I tweeting if I'm late to meet a friend? Because the motivation overhead to leave and navigate somewhere, which kind of logistically difficult, is so, so, so high unless I know the friend really wants me to be on time 2021-05-21 04:25:42 And I keep getting into situations where neurotypicals are like "why do you have so much time for X and not Y if Y only takes n minutes" and it just finally occurs to me they don't have the motivation overhead thing 2021-05-21 04:24:39 Likewise, sweeping and mopping the floor, three minutes for my housemates and maybe like 2 hours for me, but still only three of the minutes are sweeping, just the 2 hours are staring at a wall pep talking myself into sweeping 2021-05-21 04:23:27 So like a 1 minute response to an email I'm not interested in, or that involves organizing things, "takes longer" than the day to day tasks of running an international mentorship program---because it's so boring it's really hard to talk myself into it 2021-05-21 04:20:39 I think one thing that is fascinating to me about neurotypical people is that they don't seem to have a lot of motivation overhead involved in tasks. So like if something takes 5 minutes, it often actually takes 5 minutes. Not 5 minutes plus 4 hours of motivation overhead 2021-05-21 04:12:22 @sreecharan93 @amyjko @nlpnoah @LTIatCMU @ShriramKMurthi The conference ones seem to got to spam, but the student ones don't 2021-05-21 03:57:50 @nlpnoah @amyjko @sreecharan93 @LTIatCMU @ShriramKMurthi @SigplanM I do a hell of a lot for the community, but responding to every random auto-generated email I get with even a single word or a URL sounds about as painful as doing taxes https://t.co/wOqyxcl3FR 2021-05-21 03:52:51 @nlpnoah @amyjko @sreecharan93 @LTIatCMU @ShriramKMurthi @SigplanM Glad it's easy for you, though. Not everyone is like that 2021-05-21 03:52:33 @nlpnoah @amyjko @sreecharan93 @LTIatCMU @ShriramKMurthi @SigplanM Yeah no, I have ADHD, email is literally the hardest thing, I will do everything possible to minimize the extra emails I have to send. I'll spend my labor doing things I enjoy like running mentorship programs. I'm getting admin help for my email I hate it so much 2021-05-21 03:50:46 @amyjko @nlpnoah @sreecharan93 @LTIatCMU @ShriramKMurthi @SigplanM So I'm just too tired for this kind of work right now. Hopefully soon! Just two weeks left of grad school lmao 2021-05-21 03:50:25 @amyjko @nlpnoah @sreecharan93 @LTIatCMU @ShriramKMurthi @SigplanM I get a lot of NLP, data science, and ML applicants for example, and have to kindly turn them away. For a while I was hoping to build a list of other programs to get them in touch with in other communities, but then I had to get a job, write a thesis, and make a conference talk 2021-05-21 03:49:25 @amyjko @nlpnoah @sreecharan93 @LTIatCMU @ShriramKMurthi Part of the purpose of @SigplanM is helping people build long-term international connections, so that they can ask their mentors how to find an advisor and so on. I think one of the hard parts is that a lot of people don't even know what "programming languages" is to start 2021-05-21 03:47:51 @etrolleybus @fbanados wholesome 2021-05-21 03:42:50 @nlpnoah @sreecharan93 @amyjko @LTIatCMU @ShriramKMurthi Like, "Dear Prof. TALIA RINGER," no I'm not responding to that, that is clearly a template 2021-05-21 03:42:21 @__apf__ Mine started as soon as I tweeted that I am faculty 2021-05-21 03:40:02 @sreecharan93 @amyjko @nlpnoah @LTIatCMU @ShriramKMurthi Was yours automatically generated by a template, or personally sent? I feel like there's a huge difference to me. I'd never engage with a template email, I find it disrespectful to spam people 2021-05-21 03:38:36 @nlpnoah @sreecharan93 @amyjko @LTIatCMU @ShriramKMurthi Yeah, I don't know, I'm convinced we need more people who don't know how it works and send emails that are accidentally silly. I'm not convinced we need more people who register for or employ spam databases to contact people---that is quite worrying behavior in general 2021-05-21 02:40:30 Man I wonder what happens if you use GPT-3 to generate spam emails asking to work with faculty 2021-05-21 02:39:56 can they at least sort the database https://t.co/W0ezBFxYdy 2021-05-21 02:29:47 RT @fbanados: @etrolleybus @TaliaRinger “I liked your paper on Bidirectional Type Checking, I’d like to work on Machine Learning” 2021-05-21 02:29:10 @danrkports @etrolleybus https://t.co/ve57TQKaff 2021-05-21 02:03:13 They all seem like my name was just filled in automatically. The personal emails are really nice but these are different, they look like they're generated and sent from a database 2021-05-21 02:01:29 Am I on a list or something for generic emails about getting a PhD in CS following some template without any mention of my work, or do all faculty get like 1 of these emails every day? 2021-05-21 01:49:27 time for boba 2021-05-20 21:25:36 @Blaisorblade Outside of speculative sections, I think it's good to justify beliefs, but it's OK if they don't turn out to be true if it's clear that they're beliefs 2021-05-20 21:24:46 @Blaisorblade I think beliefs in papers are fine, especially in sections that are pretty much always implicitly speculative like future work sections and related work sections 2021-05-20 20:34:07 @OsherL @samfbiddle @halbfinger Cool, yep. And that sort of thing is not what my Tweet was about 2021-05-20 20:31:33 @OsherL We are a huge target of hate crimes here, and have had multiple anti-Jewish mass shootings in just the last few years. So yes it makes sense even from an American Jewish perspective to ask people not to invoke antisemitism tropes 2021-05-20 20:29:03 @OsherL I got stalked home for "looking Jewish" by the same guy twice, a serial stalker in Seattle who stalks Jewish women 2021-05-20 20:28:39 @OsherL I wonder what position of profound privilege that author must come from to have never experienced a threat to safety for being Jewish in the US 2021-05-20 20:12:18 @OsherL Except extremely targeted ones toward for example specific leaders 2021-05-20 20:11:47 @OsherL Better yet, criticize the US. I don't really believe in sanctions in general 2021-05-20 20:11:11 @OsherL Criticize Venmo, don't suggest a conspiracy, easy. The Tweets about Facebook that you posted seemed to do that perfectly fine 2021-05-20 20:07:47 @OsherL The quote Tweet I originally posted uses that language, so that is what this Tweet is about, nothing more and nothing less. Your argument (including from that awful Jewish Currents quote) seems to be that we should ignore actual antisemitic tropes, and I won't do that 2021-05-20 20:02:53 @OsherL I agree, but you originally responded to my Tweet specifically decrying conspiratorial implications as antisemitic, and I'm not really down for denying actual antisemitism 2021-05-20 19:47:29 @OsherL In this case it is quite easy to do any reading and realize this is not a conspiracy, but a bad technical response to a bad US policy. Jumping immediately to it being a conspiracy says a looooootttt 2021-05-20 19:46:23 @OsherL Yeah no I know enough about antisemitism to recognize dogwhistles about conspiracies controlling money and/or the media, and that some people who don't hate Jews believe them doesn't make a difference 2021-05-20 19:42:55 @OsherL I'm pretty sure "don't use racist language that puts people at risk of imminent harm" is a form of consideration I am quite comfortable asking for in all contexts 2021-05-20 19:40:01 @OsherL But nobody is saying not to criticize Venmo? Just to please try not to evoke antisemitic tropes in doing so 2021-05-20 19:37:19 @OsherL But I'm not opposing critiques, I'm opposing antisemitic critiques that imply the whole thing is a (((conspiracy))). Just avoid conspiratorial implications and you're good 2021-05-20 19:35:51 Please don't come here assuming weird things about my politics that I didn't say. Suggesting "vast conspiratorial forces" are what make these Tweets antisemitic https://t.co/OAWyHUUbCN 2021-05-20 19:33:45 @OsherL Because if it's the latter I'm going to mute you, and if it's the former I'm going to explain why even if you haven't experienced this yourself, this _specific approach_ to discussing Israel/Palestine puts Jews worlwide at risk of physical harm 2021-05-20 19:32:54 @OsherL Can you tell me whether you're Jewish so I know whether I'm talking to someone denying antisemitism from within or outside? 2021-05-20 19:26:04 @OsherL @swodinsky @JewishCurrents Even if the people making the statements are not antisemitic, evoking antisemitic imagery and tropes is inherently antisemitic 2021-05-20 19:24:45 @OsherL @swodinsky @JewishCurrents Yes, it absolutely does 2021-05-20 19:22:33 @OsherL @swodinsky I also recognize antisemitic undertones when I see them, though, and a lot of the Tweets about this alleging some Big Conspiracy are profoundly antisemitic 2021-05-20 19:21:41 @OsherL @swodinsky Nah, you're projecting weird views onto me that I don't have. Recall for example that I actually am in favor of a single democratic state 2021-05-20 19:20:36 @OsherL @swodinsky It's not a conspiracy because it's a completely consistent but overly broad application of US law. A bad one, but the Tweets in question are kind of asserting there is a conspiracy related to blocking aid to Palestinians---only insofar as that conspiracy is liability to US policy 2021-05-20 19:17:35 @OsherL @swodinsky https://t.co/UmmN6jwDDn 2021-05-20 19:17:22 @OsherL @swodinsky That seems like a completely bizarre twist. Did you read the Tweet that I quoted? 2021-05-20 19:16:53 @OsherL @swodinsky https://t.co/lTQOp0ecRW 2021-05-20 19:15:05 @OsherL @swodinsky It doesn't 2021-05-20 19:13:48 @OsherL @swodinsky It's not good, but it is consistent and also not a conspiracy 2021-05-20 19:13:01 @OsherL @swodinsky Didn't read this exact article, was talking about the discourse. The antisemitic thing is the implication that it's a conspiracy when actually this is just what Venmo has always historically done for, for example, all Cuba-related transactions 2021-05-20 19:07:54 Specifically it was a spy joke implying that both of us had major involvement in world events by way of paying him back for a "Cuba trip" after a recent significant event. Venmo held that transaction for months, as was policy for all things involving Cuba 2021-05-20 19:06:03 Yeah these have intense antisemitic undertones. Venmo blocked my transaction once when I made a joke about Cuba in it https://t.co/EwaaUrUCCk 2021-05-20 19:02:30 RT @AkshithaSriram1: Tangential to the main story: perhaps most forms of post-conference networking interactions do not have to center arou… 2021-05-20 19:00:02 Oh cool, faculty will be blue check eligible soon https://t.co/QfajQk12A2 2021-05-20 18:59:19 @verified What if we were maliciously reported and locked out? Seems unfortunate 2021-05-20 18:44:03 @IjeomaOo @Prodigalson2025 Lots of good points about money, I'd also note that for me and a lot of other ADHD people the actual process of getting paid back is so painful that we often just swallow large costs because the alternative is too difficult 2021-05-20 18:40:37 @_pdarragh He's not wrong 2021-05-20 18:40:30 RT @_pdarragh: me: "I think I'm almost done debugging this code! only two tests are failing, and I think they're caused by the same bug..."… 2021-05-20 18:30:50 RT @ColorOfChange: The @TrevorProject's Guide to Being an Ally to Transgender and Nonbinary Youth covers a range of topics on how to suppor… 2021-05-20 18:30:26 RT @ColorOfChange: Transgender and nonbinary youth who reported having their pronouns respected by all or most of the people in their lives… 2021-05-20 18:30:15 RT @ColorOfChange: #LGBTQ youth continue to face unique mental health challenges and risk factors for suicide, but new data released by The… 2021-05-20 18:14:33 RT @LEBassett: Just walked past a woman in Manhattan cradling four ferrets as she walked down the street. No leashes, no bag, just a pile o… 2021-05-20 18:11:08 RT @blackeuler: https://t.co/6sJphP0YQD 2021-05-20 18:04:10 @yoavgo @_snek93 True. A performative attack only works if there's some threat of an actual one, though, no? 2021-05-20 18:01:28 @yoavgo @_snek93 Yeah I've heard this, but also think the Israeli military could figure something out with enough incentive to do so. They have the best startup tech and the "best" targeted assassination abilities in general, I'm extremely skeptical there isn't a more targeted form of deterrence 2021-05-20 17:58:53 @yoavgo @_snek93 Especially with Iron Dome. It is the large barrages that overwhelm it anyways. I think some of my Israeli friends are quite worried about what the surrounding states will do in response if escalations are ignored 2021-05-20 17:55:51 @yoavgo @_snek93 But if you want to argue against deterrence entirely I'd buy it. Like maybe just ignoring rockets while you work toward peace is just fine, like you mentioned with southern Israel 2021-05-20 17:54:50 @yoavgo @_snek93 Interesting, I mean when you get a rocket barrage you take out only Hamas militant leaders as deterrence e.g. by very targeted drone strike. I don't think air strikes in such a densely populated area can ever be justified honestly 2021-05-20 11:26:54 @_snek93 I am about to sleep by the way but always appreciate the engagement 2021-05-20 11:26:05 @_snek93 So I'd like to think of something that either comes from the inside, or looks like it does 2021-05-20 11:24:53 @_snek93 Both good points. To be honest, in Israel, I think even 4 is a barrier right now. This is what, the fifth election? I think the problem to me is that given Jewish & 2021-05-20 11:20:17 @EliSennesh I can't understand still why anyone would ever want to kill another person 2021-05-20 11:16:26 @EliSennesh Why would anyone do that 2021-05-20 11:14:05 Thank god (or Egypt, really) for this https://t.co/AIMMot31YP 2021-05-20 11:08:48 Oh and can we also talk at some point about the Egypt crossing? What is the status of that? I legitimately don't remember if it reopened, but it should be open right now to give civilians who wish to flee immediate a place to go if Israel won't 2021-05-20 11:04:55 It just feels like it must take such a profound amount of privilege and detachment to be able to talk about this all in terms of ideals and not worry about whether those ideals are realizable 2021-05-20 11:01:59 Leader assassination in response to rocket fire to be clear, this is usually the preferred strategy but Netanyahu benefits from shit policies that kill people IMO 2021-05-20 11:00:40 7 can come in two state solution form if need be but I am yet to see a proposal with realistic borders. My ideal is 1 w/ codified protections. I think if you neglect the steps before, all attempts at reconciliation will lead to civil war and much, much worse outcomes for everyone 2021-05-20 10:59:03 For example, I would like, in order 1) ceasefire, 2) temporary return to strategy of leader assassination in near future, 3) Netanyahu in prison, 4) a government, 5) withdrawl from settlements, 6) education & 2021-05-20 10:52:00 Honestly I don't even know what the words "Zionist" and "anti-Zionist" mean these days. I don't know what they are pushing for. The same way I feel about "pro-Palestinian" and "pro-Israel." What is your actual stance? Like, what policy are you advocating for? 2021-05-20 10:48:33 Likewise for the national narrative, like can we please not do the whole "X has no history in Y" thing (it's wrong), or "there never was a state called Z" thing (who cares), or "Q national identity is new" (who cares) thing? 2021-05-20 10:46:14 I think I get really frustrated by politics that are not extremely practically oriented when human lives are at stake. I have strong ideals but what I push for practically often diverge from those ideals dramatically, due to recognition my ideals are irrelevant 2021-05-20 10:43:14 Seems net harmful to joust about ideals of what states are or are not legitimate in our minds. None of that will make any difference for international policy, but it will make people hate and attack each other, and give the actual hateful people cover with nice words 2021-05-20 10:35:36 Why on earth would I care about whether any other person believes a state has a right to exist or not? Popular opinion is of little consequence to geopolitics 2021-05-20 10:34:42 I think another thing that confuses me is the framing of "Zionist" versus "anti-Zionist." That's a debate that made sense in my grandparents' generation. Why are we talking about ideals when the world looks like this? Shouldn't it now be "OK, now what?" 2021-05-20 10:19:25 @EliSennesh @ZachLewis3187 @EylonALevy Pleaaaaaase ASAP 2021-05-20 09:47:51 hahahaha "technologies technologies" go to bed talia 2021-05-20 09:47:19 Oh and I have this obsessive thing where the last sentence of all of my papers has to be cute and actually read like a last sentence, so: https://t.co/GvtHayPXdL 2021-05-20 09:39:58 The text here will need a bit of work tomorrow since it's redundant relative to the first page I posted, which is actually well written but whatever it's late 2021-05-20 09:37:23 more~ https://t.co/2NpNq49YD1 2021-05-20 09:37:08 weeeee https://t.co/C6Y1qyvwHW 2021-05-20 09:36:55 OK I'm like high from sleep deprivation so no guarantees this is good yet but here's the rest of the future work section after a revision pass. I don't know if I'll get to keep all of this but I really want to, and I feel like after 124 pages, 5-6 pages of future work is fine https://t.co/xPIZB9EFvc 2021-05-20 09:25:22 @bblipp I mean thank you 2021-05-20 09:25:14 @bblipp sorry I'm like high from sleep deprivation right now haha 2021-05-20 09:24:56 @bblipp dat datta 2021-05-20 09:22:30 I've slept 7 hours in the past 2 days it's late it's late 2021-05-20 08:03:41 Friend says the economy thing is mostly dying you might be able to get exemptions these days if you really want them and still get good jobs in Israel so that's cool. I don't know how that plays out in practice 2021-05-20 07:44:56 Fight back 2021-05-20 07:44:42 Because war, genocide, and what have you, all of these things only ever thrive when you dehumanize the "enemy" 2021-05-20 07:43:41 The way to fight back is to talk to each other, to let go of the luck, the coincidence of being born where you were born in the family you were born to and to just like actually talk to people and learn and view them as whole and human 2021-05-20 07:42:30 And it's not guilt. It's direct exposure to something that, until then, had been theoretical because that's what the borders and national narrative are there for, to make it possible for people to dehumanize each other. The way to fight back is to just never dehumanize anyone 2021-05-20 07:41:15 So instead most Israelis I know try to get roles that don't involve combat. I have a close friend though, who was a medic in Gaza in 2014. Let me tell you, he is the most pro-Palestinian Israeli I have ever met, now 2021-05-20 07:38:44 In Israel, unless you study abroad for a very long time, I believe it is basically impossible to get an exemption to serving in the military and still participate in the economy in any way when you're done. I think few in the world would be selfless enough to incur that cost 2021-05-20 07:35:51 But mandatory servi--- Hi, I'm one of the people who gets yelled at for being Israeli(-American). I actually didn't serve in the military at all, I was exempt because I lived here. But that was also luck! My parents moved here so my dad could get his engineering degree 2021-05-20 07:33:33 How about let's just not conflate people and governments at all. I sure as hell don't want the rest of the world holding me accountable for Trump's actions 2021-05-20 07:31:28 (The probability is actually more like 40% US, 40% Israel, 20% elsewhere, IIRC. But for Americans I think the point still stands: equally likely to have been born in Israel---so what is the statement you are making, really? You're a better person because of luck?) 2021-05-20 07:23:23 I get it that there are a lot of antisemitic losers who think every Jew needs to answer for the crime of Israel. But it's also a matter of luck that you ended up in the US and not Israel, so maybe also don't dehumanize Israeli Jews 2021-05-20 07:22:00 Fun fact: You can't control where you're born, or where your parents or born for that matter. Jewish? Pick a side, flip a coin. That's roughly the probability you would have been born in Israel. Much higher if your family includes recent Jewish refugees https://t.co/K68W443Veu 2021-05-20 07:18:40 @_snek93 It's my dissertation, so I hope so . I was a software engineer for a few years, and when I did this kind of thing I would always have to go back and rebase before pushing 2021-05-20 07:15:38 Please let me keep this in my thesis https://t.co/sIYycCp6vk 2021-05-20 06:40:54 @_snek93 or when every time you go to write a commit message, you forget what you were even doing the last 2 hours fo your life lol 2021-05-20 06:21:27 yeah I'm doing fine https://t.co/gwB1r1NbnA 2021-05-20 05:55:45 RT @josephimperial_: Money Is All You Need https://t.co/J8SodBum3U 2021-05-20 05:27:28 @hillelogram Ahh thanks! I appreciate even the sentiment 2021-05-20 05:09:02 @Fish_CTO you mean you don't have two days left to finish your dissertation and a PLDI talk and poster with two eye infections 2021-05-20 04:57:06 God I feel awful 2021-05-20 04:37:58 @lreyzin That's all, I guess. You're a reasonable person signaling that you're unreasonable, and I don't really understand why 2021-05-20 04:37:22 @lreyzin But I wouldn't go writing about "cancel culture" because that is a phrase a lot of malicious people use as a dogwhistle, and I wouldn't want to signal to other people that I am malicious---not for my wellbeing, but for theirs 2021-05-20 04:36:11 @lreyzin I don't think you're doing yourself any favors by using the same language and repeating the same points. I agree, for example, that it should be off limits to contact someone else's employer when you don't share an employer because of something they say online 2021-05-20 04:35:17 @lreyzin So, I would be unlikely to invite someone who uses coded words like "woke" and "cancel culture" because I'd be worried they were actually harmful in a way I wasn't catching. It's more about signaling than politics, though 2021-05-20 04:33:55 @lreyzin After that, I decided it was worth sometimes giving people who use that language a chance. But it's also scary, because sometimes engaging with people who use the same exact language and repeat the same points ends in me being subject to a harassment campaign 2021-05-20 04:32:48 @lreyzin After one Twitter interaction, I realized you are actually pretty open minded, you just seem to have fallen into the unfortunate trap of repeating points and language that really hateful people often use. I applied though my app was late (it's all good, happy with UIUC) 2021-05-20 04:31:21 @lreyzin Israeli honesty: I found your expressed views very worrying early on and was actually too nervous to apply to UIC after they contacted me, because a lot of people using the same language you use to discuss this are also actually hateful people 2021-05-20 04:24:20 @NuritPeleg Stay safe! I hope there is a ceasefire soon 2021-05-20 03:06:43 @CheeToS_ @wyoumans @BruceTennant Yeah I got a Venmo transaction blocked and two accounts flagged for jokingly sending someone a Venmo transaction that said "Cuba trip" in it 2021-05-20 03:02:57 @wilton_quinn Of course 2021-05-20 02:46:58 https://t.co/SFDSpqKRcg 2021-05-20 02:46:08 Some wonderful hashtags I don't feel comfortable phrasing as is, or even typing out, praising Shitler for the Holocaust 2021-05-20 02:40:25 It's not about shutting down conversation, it's about finding ways to signal safety that enable Jews to safely participate in the conversation, or even just safely go about our daily lives https://t.co/fRPKal1fIJ 2021-05-20 02:28:19 This is fun too: https://t.co/xSuwQ5oKxP 2021-05-20 02:22:49 This happened in 2014 Europe (especially France) by the way, and the result was a large group of Jews who moved to Israel, so I think that's also important to think about 2021-05-20 02:21:42 https://t.co/02lBYbzZOr 2021-05-20 02:21:07 This is what I meant when I said it's a really reasonable ask to have the ~90% of people who don't hate and want to hurt Jews try to use language that distinguishes them from the ~10% who do when raising important points so I know whether I'm safe https://t.co/NCmZQp43pk 2021-05-20 02:16:00 @EliSennesh I can't find news about NY, what is happening? I know about LA 2021-05-20 02:13:51 @EliSennesh ugh are they uggghhh 2021-05-20 02:13:15 But the problem with that coping mechanism is that those issues impact other people. I've noticed people doing this with issues that directly impacted me, and I really did not appreciate it 2021-05-20 02:12:35 Plus, I will be totally honest, sometimes when the world is really bad, I like having some other issue to focus on that doesn't impact me directly, and I like being able to pour some of my feelings into that 2021-05-20 02:10:44 I really think this is why online conversations feel like they're getting worse and worse. Did anyone else lose friends this year, just due to fighting over politics and so on? That usually doesn't happen to me, but happened a few times this past year https://t.co/MMjNgnE2EP 2021-05-20 02:08:56 @Fish_CTO @BowToChris @plt_abbie I just don't really feel motivated to eat when eating isn't social 2021-05-20 02:08:11 @kaflurbaleen It was so bad even indoors because my house is so poorly insulated 2021-05-20 02:07:39 @kaflurbaleen Oh gosh I forgot about the wildfires. I have bad asthma so I had to flee when the air suddenly got too bad for me to breathe in WA. I ended up impulsively driving to Utah 2021-05-20 02:04:38 @xl772 Yeah I still feel a little irrationally afraid of some activities I know rationally are safe, like hanging out unmasked indoors with other vaccinated people 2021-05-19 23:49:47 @Fish_CTO I don't eat lunch most days now 2021-05-19 23:46:42 But yeah, what an awful year and a half of suffering, no wonder everyone is so high strung about everything right now 2021-05-19 23:45:40 You might find yourself depressed, for example, and not understand why. If you do, this is normal and a physical phenomenon. You might need medication to get your brain making those connections again so you can even start to have the feelings you used to have 2021-05-19 23:44:30 When things get back to normal, a lot of you are not going to feel OK even once your external situation improves dramatically: friends, food, a job, and so on. Prolonged distressing situations like this can actually do physical damage to your brain 2021-05-19 23:42:46 The total collapse of my social support system overnight. And I think above all, the weird insistence by everyone that we call this "the new normal," as if there was ever anything remotely normal about it 2021-05-19 23:41:43 And then there was like, the national gaslighting sessions, the insurrection attempt, the online harassment, the hunger and job loss my friends faced, the fear my family might get sick and die, all while I'm trying to navigate a job market with 2/3 as many jobs 2021-05-19 23:39:36 It made me feel really sad and kind of sick, like it was actually hard to think of those memories of a lab full of people, there was a sense of great, sudden loss. I won't get those grad school days back because I'll graduate in 2.5 weeks 2021-05-19 23:38:31 Has anyone else found this last year and a half pretty traumatic? Last night I was thinking about what it was like to go into the office and see people, and get lunch with my lab, and what it was like when suddenly everything was canceled and people started dying and stuff 2021-05-19 23:08:29 @astahfrom have no fear https://t.co/06jvhKvbYT 2021-05-19 23:06:52 Why am I like this? Why do I write papers about user studies for proof assistant and type theory conferences, and papers about proof assistants and type theory for PLDI? 2021-05-19 23:02:52 why not https://t.co/awDJ85sIoQ 2021-05-19 23:01:24 Like I'm just going to own it https://t.co/0AX0H4oX1I 2021-05-19 23:00:52 I just went with a self-deprecating approach making fun of myself for writing this paper, while still selling people on "no, but seriously, you should read it, here's a bunch of really wild and useful things we do" 2021-05-19 23:00:12 I've got to say, writing a 5 minute PLDI talk about a tool for automatically fixing broken proofs in a proof assistant by combining search procedures for equivalences, a proof term transformation that externalizes automatic transport, and a proof term to tactic decompiler is hard 2021-05-19 21:00:20 Talia slides no context https://t.co/OCM1F4wsY6 2021-05-19 19:21:11 @DrDeeGlaze Often questions are valuable specifically for their ambiguity 2021-05-19 16:20:36 @yoavgo Learning both at the same time really gave me a poor sense of whether things should go left to right or right to left. For example I made a הגדה for class growing up, and I wrote the plagues in Hebrew, on the cover of the book (rightmost page), but 1st top-left, 10th bottom-right 2021-05-19 16:15:53 @yoavgo I learned to write and read both at the same time, and "?" looks forwards regardless of text direction, which I'm only now realizing is weird 2021-05-19 16:07:34 @kjw_chiu I think it's going to be a modified job talk 2021-05-19 15:58:07 @kjw_chiu Yeah but I haven't started writing my defense yet, which is 2 weeks after haha 2021-05-19 15:53:22 So many good people https://t.co/CcIYmRms4f 2021-05-19 15:51:15 Lmao gmail https://t.co/ZZE0hlociu 2021-05-19 15:49:02 @jonmsterling It was private first and then he tried to send it to the list, but I don't think it worked. It was a work of art 2021-05-19 15:48:14 https://t.co/pgUnZoxbp4 2021-05-19 15:46:58 https://t.co/vn8jSdddSR 2021-05-19 15:46:44 Subthread going on, are there really axioms you can't assign computational content to? What does that actually mean? What counts as computational content? https://t.co/MoDwOW3oxI 2021-05-19 15:43:53 It's still going, and it keeps getting more interesting. Of course this question does not have a "right" answer, but asking the question leads to really interesting thoughts, especially at the boundaries of what only some people view as programs. I guess it's philosophy then https://t.co/Zsf0kANKRP 2021-05-19 15:41:59 @DavitSoselia_ Have not made it out of bed. But managed to take 600 mg ibuprofen lol 2021-05-19 15:30:34 Now I have two infected eyes, not just one. Also, cramps from hell, slept 3 hours. Thesis deadline Friday, along with PLDI talk and poster deadline on the same day. 2021-05-19 15:18:13 RT @_snek93: @yoavgo @omershapira Sorry I have no idea what this says but are Hebrew speakers aware that this question mark exists: ؟ Is i… 2021-05-19 15:13:06 @_snek93 @yoavgo @omershapira Woah 2021-05-19 15:00:01 RT @andrejbauer: @HerrDreyer @TaliaRinger Technically, I'd call it a program if it is a closed expression (of a type that bares some resemb… 2021-05-19 14:37:18 Yes. Especially any software https://t.co/Be5mDWrBdA 2021-05-19 14:20:25 RT @Yair_Rosenberg: The thing about anti-Semites using Israel as an excuse to assault random Jews around the world is that making Jews feel… 2021-05-19 14:19:43 RT @chrisamaphone: games (and game design) can make a lot more sense when you think of them as attempts to technologically scaffold existin… 2021-05-19 09:55:22 Good night, hope this is an understandable take 2021-05-19 09:50:53 And please listen when I tell you that you're doing that. This isn't a joke, my grandfather was in a death camp, I have had nightmares about getting kidnapped by Nazis once per month for my entire life, I was afraid of showers growing up because I thought I'd get gassed 2021-05-19 09:49:42 I should sleep. Really, please do discuss how to build a world of true autonomy, peace, and freedom of movement for Palestinians. I want it. Just please do this without repeating language that puts my life at risk 2021-05-19 09:48:03 - viewing us all as privileged, rich people - implying that we control too much of the media narrative - implying the Israel lobby in the US is Jewish (it's evangelical) - implying that we control politics - implying that wars are always our fault 2021-05-19 09:45:54 - implications that we are all selfish people who only care about ourselves - erasing the _British_ history of the colonization when you call Israel settler-colonial 2021-05-19 09:42:28 - claiming that Jews aren't "the real Jews" or Israeli Jews are actually just Polish/Kazhar/whatever - scolding Holocaust survivors for living in what is now a fascist country, when no other place was taking refugees, and Holocaust survivors were oppressed in Israel too 2021-05-19 09:38:57 - implying that Israel does this because it's fun and we are bloodthirsty and enjoy harming children - erasing Jewish history - erasing Jewish ethnicity - erasing Mizrahi Jews 2021-05-19 09:37:06 Antisemitic anti-zionist rhetoric I'm sure you've all seen recently: - Nazi comparisons - inability to view any Israeli Jews as humans you can talk to and reason with, instead viewing all of us as evil monsters - expecting every Jew to state our opinions about Israel/Palestine 2021-05-19 09:34:22 It would be nice if, like other marginalized groups, people basically ever believed Jews when we talk about antisemitism. And not every Jew will experience it, so it is super tokenizing to hold up Jews who haven't as examples, and disrespectful to then ignore those who have 2021-05-19 09:32:07 When 10% of the rhetoric about a topic is calling for your family's death, and 90% is calling for the thing you literally also want with the same language, it's reasonable to ask the 90% to maybe use different language so that we know we can safely exist in a space 2021-05-19 09:30:23 The "Jews shut down all discussion of anti-zionism as antisemitism" is, ironically, an antisemitic trope, even if one a lot of American Jews have come to somehow believe 2021-05-19 09:28:57 Basically no Israeli is going to shut you down for saying something even actually antisemitic. They'll tell you exactly what they think, but in that culture it is a sign of respect and trust to do that 2021-05-19 09:27:12 1. a lot of anti-zionist rhetoric actually does parrot antisemitic tropes, even if inadvertently, and 2. most people pointing that out just want you to make your point without using dangerous and traumatic tropes. 2021-05-19 09:26:57 (Preface: Israeli-American Jew in favor of single democratic Israel/Palestine in the ideal, but also hopeless that will ever happen) I often seen "anti-zionism isn't antisemitism" and honestly this take is one of those technically true but also awful ones, because: 2021-05-19 09:15:11 @ZoeSchiffer @alexhanna But almost nobody is careful enough to do it 2021-05-19 09:14:27 @ZoeSchiffer @alexhanna This is possible to do without erasure, too 2021-05-19 09:13:48 @ZoeSchiffer @alexhanna It would be really nice if people would actually listen and just reframe things to avoid tropes historically used to justify murdering my family? 2021-05-19 09:12:53 @ZoeSchiffer @alexhanna Sometimes Jews who have been lucky enough to not have grandparents in death camps, or to not have their own lives threatened for their Jewishness at some point, don't even notice it and that's just a subtle form of privilege honestly 2021-05-19 09:11:31 @ZoeSchiffer @alexhanna As a Jew who wants a single democratic Israel/Palestine one day, like 90% of anti-Zionist rhetoric I see actually is antisemitic because people inadvertedly parrot antisemitic tropes that actual antisemites use 2021-05-19 08:42:55 @smileitsjoy For years I convinced myself I didn't have it because if I were sensitive to rejection, I was scared everyone would think I was weak and reject me 2021-05-19 08:42:13 @smileitsjoy It sucks 2021-05-19 08:41:24 RT @goulcher: know your worth https://t.co/F9yRDeVlWy 2021-05-19 08:29:38 I liked this thread a lot: https://t.co/1WRCGVcOg9 2021-05-19 08:28:04 I'm not sure why Bob Harper's answer isn't showing up, though---it was really thoughtful and fascinating. Maybe it hasn't gone through the moderation queue yet for some reason? https://t.co/zcyks2asJ2 2021-05-19 08:23:12 In case you wonder what happens when you ask a bunch of wonderful type theorists what a program is, here you go: https://t.co/TWPtT0griR 2021-05-19 07:59:41 Plus students are the best and I am sure will always teach me new things 2021-05-19 07:58:44 I'm really excited to start operating at a bit of a higher level with my research, because vision and ideas and plans for realizing them and social connections to find the right people to implement those ideas and bring them together are actually kind of my jam 2021-05-19 07:37:12 @emilymbender I almost attended Google IO in case there was any discussion about this, so if there is I want to be fully clued in 2021-05-19 07:35:36 @emilymbender This interview was before Google IO, just with the reporter? (curious) 2021-05-19 07:04:39 @matzo If it's linguistics, proximally by way of logic, but the meanings have diverged significantly 2021-05-19 07:04:03 @matzo If it's logic, it is related 2021-05-19 07:00:45 Seems like the CDC is really having an identity crisis these days! 2021-05-19 06:54:05 @matzo It is a type theory pun: https://t.co/DymNsmhS22 An intensional one 2021-05-19 06:47:55 @nbhansen It would be impossible to decide! 2021-05-19 06:44:26 Oh man! So CDC just put out a theory that fully vaccinated people don't need to worry about the distinction between definitional and propositional equality anymore. I doubt this theory was intensional! 2021-05-19 06:20:53 @jix_ ohhh I see what you mean, haha 2021-05-19 06:20:27 @jix_ what is that 2021-05-19 06:16:15 Just sent an email with the title "PLDI paper, external transport, Lambek's" and um 2021-05-19 05:23:41 @normaldeviate @djg98115 Turns out you can transfer an eye infection from one eye to the other via your keyboard 2021-05-19 05:21:25 @normaldeviate @djg98115 I have an eye infection in my left eye, and while editing my thesis today managed to also infect my right eye, so 2021-05-19 05:17:59 @normaldeviate This is @djg98115's hobby with his graduating students 2021-05-19 05:17:08 @kamatsu8 I agree though, especially now that I've made them \footnotesize in my thesis 2021-05-19 05:16:41 @kamatsu8 To be clear, the faces are in the PLDI paper despite adequate horizontal space because I thought they were fun and I was sick of taking everything so seriously 2021-05-19 05:15:10 RT @Fish_CTO: @TaliaRinger ~~you can save extra space by renaming "π (part)" to "τ"~~ 2021-05-19 05:15:05 @Fish_CTO hahahahahahaha 2021-05-19 05:14:49 Missed a "we" there 2021-05-19 05:13:32 Love correlates with happiness 2021-05-19 05:12:03 Good enough https://t.co/ppv4rgm3cD 2021-05-19 05:09:35 Automatic, mixed, and manual configuration are now going to be , , and in my table, and nobody can stop me 2021-05-19 05:07:29 "Finally you'll have as much space as you need to do whatever you want," they told me about my thesis, while totally neglecting to consider the entire concept of horizontal space 2021-05-19 03:54:10 Conor if you're awake, don't worry, you'll get an ack too 2021-05-19 03:51:17 And of course I still use it for its stated purpose https://t.co/uSyZmUDitm 2021-05-19 03:48:27 Non-authors too https://t.co/tBsC98yrXR 2021-05-19 03:47:11 Fun hack using the knowledge package to get an index into parts of my thesis where in fact I did not do the thing, but rather my coauthor or friend or something did the thing. That way, I can use "I" honestly everywhere else https://t.co/XNHgiLzZJA 2021-05-19 03:45:14 @johnregehr Oh haha I have a complete draft modulo formatting and tweaks and so on. But I ought to, it's due on Friday 2021-05-19 03:44:49 @johnregehr https://t.co/E3Ciohy8Nf 2021-05-19 03:42:24 @johnregehr Also, all names of people are indexed using the knowledge package, so their names link back to the acknowledgments section, and at the end in the index acknowledgements are sorted and all occurrences of names in the thesis are listed 2021-05-19 03:38:36 @johnregehr I use "I" for myself, "we" for mathematical handholding, "she" for imaginary proof engineers, "they" for anonymous people, and the preferred pronouns of people I know and refer to explicitly ("RanDair implemented a decompiler.") I have a "pronouns" subsection in my reading guide 2021-05-19 01:56:43 Let's play a type game, play a type game Do you want LEM or you want fame? Are you in the game? Doin' the type game 2021-05-19 01:51:25 I wanna take a ride on your proof tactic 2021-05-19 01:51:09 Don't think too much, just paste that trick 2021-05-19 01:48:52 @EdNutting @alpha_convert It's pronounced "quock." It's French for the sound roosters make 2021-05-19 01:48:23 I wanna take a ride on your proof tactic 2021-05-19 01:46:14 @alpha_convert New solution to the Coq naming problem: pronounce it "sock" 2021-05-19 01:45:17 @alpha_convert You'll get bored enough at some point in your PhD to need to sing yourself songs about type theory though, I promise 2021-05-19 01:43:17 @YVigfusson it's good to correct me now, before this is published :) 2021-05-19 01:41:32 Let's get down this beat is CIC$_{\omega}$ 2021-05-18 23:59:22 RT @samth: @agdakx @TaliaRinger @taooftypes I just realized that this actually comes up a lot! The standard proof of the undecideablity of… 2021-05-18 23:27:34 brazen 2021-05-18 23:24:41 I'm just unabashedly using exclamation points in my thesis like why not https://t.co/JGwsOVxPoA 2021-05-18 22:28:01 How are there so many cool professors 2021-05-18 22:27:40 It brings me such immense joy to talk to my future colleagues and potential colleagues 2021-05-18 22:22:06 What do we have against just helping people stay alive though https://t.co/M8LyIF6e6W 2021-05-18 21:29:19 RT @ionathanch: What is a program? A miserable little pile of terms 2021-05-18 21:06:47 @luna_lun_a We all study .... something 2021-05-18 21:05:57 RT @AJLUnited: The harm in what happened with @60minutes yesterday isn’t about which researcher put out what research first. It’s about ho… 2021-05-18 21:05:41 RT @AJLUnited: On Sunday night, @60Minutes aired an episode on facial recognition and #algorithmicbias, and excluded the groundbreaking wor… 2021-05-18 20:59:58 I love type theorists. I email out this question and they respond with "I don't have answers, but I'd like to add to the confusion" hahaha 2021-05-18 19:57:41 @samth Is...is everything a program 2021-05-18 19:51:23 Context: https://t.co/VDF0mWoRZe 2021-05-18 19:45:05 I've heard getting your PhD is basically unlearning things https://t.co/oQMATUMEvV 2021-05-18 19:28:44 RT @peter_boyland: @TaliaRinger This reminds me of the (possibly apocryphal) story of a linguistics conference where someone asked a speake… 2021-05-18 19:28:35 @jameskjx I think we use "program" to mean things-that-could-be-programs if only we knew more or fixed some mistake. Likewise, the "broken proofs" my tools fix are not proofs until they are fixed 2021-05-18 19:26:39 @rakamaric I do think most Israelis still support two-state by the way. My opinions are kind of fringe leftist by Israeli standards, although probably not that weird to Israeli academics 2021-05-18 19:24:38 @peter_boyland I'd really love some modern papers about what programs actually are! 2021-05-18 17:51:55 Could be a wrong assumption though! 2021-05-18 17:51:46 Assuming mostly PL experts answered this simply because of the wording 2021-05-18 17:51:00 Programming languages experts do not agree on what a "program" actually is https://t.co/ooc8HsxgO5 2021-05-18 16:36:22 RT @aaronclauset: Here's the full @ScienceMagazine Perspective piece, on the limits of prediction in the science of science: https://t.co/O… 2021-05-18 16:36:01 RT @aaronclauset: Back in 2017, @DanLarremore @robysinatra and I warned in @ScienceMagazine about the coming use and abuse of machine learn… 2021-05-18 16:31:33 RT @weinberz: The goal: the metrics: https://t.co/4QcrncHlvg 2021-05-18 16:21:22 Oh and @_primes_ that's an aestehtically pleasing one 2021-05-18 16:20:35 How do you tell which people following you are bots and block botnets? I don't want bots except @IntuitionisticB and @mathslogicbot and @mathstableaubot and maybe @EmojiMashupPlus 2021-05-18 16:01:56 RT @ChBE_Illinois: A team led by Zhao Huimin invents new COVID-19 test that is portable, affordable, accurate, and fast! The new test ca… 2021-05-18 16:01:01 RT @GirlsWhoCode: Sorry to burst your bubble, but if you can open a PDF, you can learn how to code. No fancy degrees, hacker hoodies, or ex… 2021-05-18 15:54:28 I have pink eye with 4 days left to finish my thesis and PLDI talk and poster. It hurts to open my left eye, and I can't put in my contacts lenses (my rx is like -6.00). Cool cool https://t.co/Fp5L8fbMsq 2021-05-18 15:41:42 RT @littmath: Recently realized that in our faculty meetings, a tie vote on a motion results in that motion failing. Now I will forever str… 2021-05-18 15:19:23 RT @TheGoldenRatio4: You are doing a great job today! https://t.co/Axwjc3Nmre 2021-05-18 15:15:26 RT @moorehn: I was running errands and passed an old antique shop and then JUST LIKE THE NOVELS TELL YOU, it felt like being transported ba… 2021-05-18 15:03:59 RT @ejimenez_ruiz: Very glad to announce the 15th International (Virtual) Workshop on Neural-Symbolic Learning and Reasoning (NeSy@IJCLR):… 2021-05-18 09:22:04 @yoavgo הכל בסדר https://t.co/j4ZKDhM27V 2021-05-18 08:12:40 You can blame the people still there all you want, but all of this is by design by the British to break up the Ottoman Empire. Zionism to the British was just a convenient tool for that. This is why I'm adamant the UK needs to repair damage 2021-05-18 08:11:12 Oh one more thing, anyone who reads this book will come away so angry at the British colonial powers it will be difficult to contain yourselves. That this is yet another British colonial project ending in fleeing consequences is clear from reading both histories together 2021-05-18 07:21:02 It also gave me a little more faith: if people literally just don't know the history, maybe knowledge is empowering enough, knowledge that counters bullshit selective national narratives meant to sustain barriers between people 2021-05-18 07:19:29 This book did the work of shifting me left on Israel/Palestine despite being sold a hard Israeli narrative growing up. But I also recognize the reality that many have never had the chance to understand the difference in narratives and there will be no peace and justice that way 2021-05-18 07:15:15 That diff is the most informative thing you will ever learn about Israel/Palestine 2021-05-18 07:14:48 It was put together by a group of Israeli and Palestinian teachers who were stunned at the difference in narratives, and wanted to close that information gaps. By presenting them chronologically next to each other, the reader can see the "diff" between them 2021-05-18 07:09:09 Some middle eastern friends responding to this by DM asked for reading. This is my favorite book about Israel/Palestine: https://t.co/bchtPjPKUE It shows how each national narrative relies on erasing the history of the other. Each history is true, but the omissions are telling https://t.co/2FHsimHt8U 2021-05-18 06:38:30 @rakamaric https://t.co/MsD8DgGRsP. 2021-05-18 06:34:16 @rakamaric https://t.co/OWQhXX4cJk 2021-05-18 06:32:55 @rakamaric If you have two "states" but one controls the water supply, what do you really have? That's just one example 2021-05-18 06:32:28 @rakamaric Mostly the locations of the borders, and the geopolitical reality. For example, most of the water supply in the region comes from the Golan Heights, so whoever controls the Golan controls water. This is the real reason Israel annexed the Golan, after the water supply was cut off 2021-05-18 04:05:44 @kamatsu8 @GuiltyOfSloth @yawaramin @wilbowma Oh, fair, these days I think it's a form of "knowing" 2021-05-18 04:04:44 @kamatsu8 @GuiltyOfSloth @yawaramin @wilbowma It's me my family is from Israel I was born with dual citizenship, why do you think I get so annoyed about this 2021-05-18 04:03:37 @kamatsu8 @GuiltyOfSloth @yawaramin @wilbowma Hi 2021-05-18 03:56:01 @kamatsu8 @GuiltyOfSloth @yawaramin @wilbowma That's actually unusual? It should be hard to exist in the community without meeting Israelis 2021-05-18 02:40:34 RT @geomblog: It's time for a transition. 1. Moving to @WHOSTP to work with @AlondraNelson46 on bias/fairness. 2. Moving to @BrownUnive… 2021-05-18 02:38:13 @geomblog @WHOSTP @AlondraNelson46 @BrownUniversity @BrownCSDept @Brown_DSI @senykam @UtahSoC Also, soooo happy the administration has good experts involved in this 2021-05-18 02:37:39 @geomblog @WHOSTP @AlondraNelson46 @BrownUniversity @BrownCSDept @Brown_DSI @senykam @UtahSoC Nice, Providence is my hometown 2021-05-18 01:43:20 OK any middle eastern friends can also join by DM---I think it is the American/European/etc. perspective I cannot handle https://t.co/5Otdky07Pl 2021-05-18 01:33:25 @MichaelDoronII That's fair 2021-05-18 01:30:14 @hillelogram But I've had so many good conversations with actual Palestinians 2021-05-18 01:18:11 OK I am done talking about Israel/Palestine to anyone except Israelis and Palestinians, sorry, I can't do it, it's too hard. It's like explaining geopolitics and history to a brick wall that yells at you. Any Palestinians and Israelis can DM me anyone else I'm just done 2021-05-18 00:55:48 @kamatsu8 @GuiltyOfSloth @yawaramin @wilbowma It's a pretty natural reaction to everyone hating you and trying to kill you no matter what you do, which is all of Jewish history. You know leftists Israelis because you know academics 2021-05-18 00:41:00 @Yara81639888 @GuiltyOfSloth @yawaramin @wilbowma Very, in fact it is the only path to change. Isolate Israel and they will align themselves with right wing countries like Russia 2021-05-18 00:39:42 @GuiltyOfSloth @Yara81639888 @yawaramin @wilbowma It would be strengthened significantly, actually. This is the problem with trying to dictate the actions of another country without understanding its internal politics. 2021-05-18 00:38:10 @kamatsu8 @GuiltyOfSloth @yawaramin @wilbowma The left has been dying and dying and dying, killing any hope there was for peace and reparation, it is time to ask why and fix that 2021-05-18 00:37:01 @kamatsu8 @GuiltyOfSloth @yawaramin @wilbowma Yes, actually. Yes. 2021-05-18 00:36:05 @GuiltyOfSloth @yawaramin @wilbowma "both-sidesm." It's a catchy phrase, but it totally ignores the actual politics of the region and the fact that you need serious internal, not external change in Israel in order for right of return to not immediately cause a second civil war between Jews and Arabs 2021-05-18 00:34:10 @Yara81639888 @GuiltyOfSloth @yawaramin @wilbowma BDS will push Israelis way further right I can promise you that 2021-05-18 00:32:41 @GuiltyOfSloth @yawaramin @wilbowma It is extremely disproportionate I will give you that. But yes this is politicians sustaining the situation 2021-05-18 00:31:42 @GuiltyOfSloth @yawaramin @wilbowma Nah you know literally nothing about this 2021-05-18 00:30:56 @GuiltyOfSloth @yawaramin @wilbowma Need a way to get Netanyahu out of power and into jail. Need a government with Arab representation. Both were supposed to happen until this started and now Netanyahu gets to stay in power longer 2021-05-18 00:29:39 @GuiltyOfSloth @yawaramin @wilbowma Need a way to teach accurate history to both groups and not erase each other's histories 2021-05-18 00:28:43 @GuiltyOfSloth @yawaramin @wilbowma Need a way to open dialogue between Israelis and Palestinians, the people not the governments 2021-05-18 00:27:58 @GuiltyOfSloth @yawaramin @wilbowma Need a way to regrow the Israeli left 2021-05-18 00:22:58 @kamatsu8 @yawaramin @wilbowma It's not though they use the same language as reasonable people until really pushed, that's why I'm too scared to go to any in person event pushing for my actual political views 2021-05-18 00:21:32 @yawaramin @wilbowma I think both. I also hear Israel frequently referred to as a "white European settler colonial state" and it's like people just completely deny and erase Jewish history. (It is settler colonial but only insofar as the British are the colonizers and the Israelis the settlers) 2021-05-18 00:19:59 @GuiltyOfSloth @yawaramin @wilbowma Because that's how this whole thing started, the same problems are there and in fact exacerbated, you cannot create a state of true coexistence without first reeducating people to coexist. Populations are like 50/50 2021-05-18 00:18:43 @yawaramin @wilbowma I don't know if it's really fringe, though. I hear "the Jews should go home" like all the time when talking about this issue 2021-05-18 00:17:26 @Yara81639888 @GuiltyOfSloth @yawaramin @wilbowma SA wasn't formed by a group of indigenous refugees who had been oppressed and genocided for 2000 years. Using the same approach is never going to work 2021-05-18 00:16:23 @GuiltyOfSloth @yawaramin @wilbowma I don't oppose it on principle, but yes it will absolutely lead to civil war, and that's the point 2021-05-18 00:14:52 @yawaramin @wilbowma FWIW that is my hope too but I wouldn't ever march in something like that because I know there are people there who also mean "and kick out the Jews" and I can't tell who means what with the current framing 2021-05-18 00:13:43 @yawaramin @wilbowma I don't think Israelis and Palestinians are very interconnected though 2021-05-18 00:12:35 @GuiltyOfSloth @yawaramin @wilbowma It would be nice, but it just won't happen 2021-05-18 00:12:13 @GuiltyOfSloth @yawaramin @wilbowma Cessation of settlements should have been done forever ago. Right of return will lead to immediate civil war it won't happen 2021-05-18 00:10:48 @yawaramin @wilbowma A dream would be one democratic state with peace and constitutional protections for all of the indigenous ethnic groups including Palestinians and Jews It will never happen though, you would need to get people to stop hating and fearing each other first 2021-05-18 00:09:25 @yawaramin @wilbowma Or genocide, that seems on the table for a lot of people's "solutions" where they want "one state, but my state" 2021-05-18 00:08:40 @yawaramin @wilbowma Two state you get Israel controlled Palestine with like, nominal self determination but not much more, divided over three territories. One state you get one of apartheid or civil war converging back to two or three de facto states, take your pick. 2021-05-18 00:06:55 @yawaramin @wilbowma I just don't even know what that means. I don't know what people are matching for or want. I want peace and freedom of movement and prosperity and reparation but the question nobody seems to be willing to answer is how? What's the solution? 2021-05-18 00:05:05 @kamatsu8 If I found out he knew the history, sure, but I'd assume most Americans don't know that this is yet another British colonial disaster 2021-05-18 00:03:37 @yawaramin @wilbowma This happened in 2014 too but my friends weren't politically active enough to notice 2021-05-18 00:03:04 @yawaramin @wilbowma Probably I get angry because when people give flat flashy takes about Israel without discussing the history, Jews get attacked, flee to Israel, and vote right wing feeding into a cycle of violence that perpetuates forever 2021-05-18 00:00:36 Everyone discussing Israel/Palestine needs to discuss the UK's role in it and ask the UK to do something about it. Just most people are ignorant enough not to know the history and if you do then what on earth are you doing not spending your energy there first 2021-05-17 23:59:45 Another one that keeps coming up "but he's a US citizen now why should he have to talk about the UK actions" Because that is the root of all of this, not discussing it is ahistorical and counterproductive, they have serious unpaid debt in this issue, and he is clearly aware 2021-05-17 23:57:38 @kamatsu8 When he's discussing an issue literally started by the UK, it's disingenuous to not target the root he is clearly aware of 2021-05-17 23:56:16 @RunDaltonRun Because it's ridiculous to ignore the root of all of this which he is clearly aware of. Want to pressure Americans to do something productive? How about pressure us to pressure the UK to pay reparations and take in refugees 2021-05-17 23:53:44 @kernelmethod For starters, he can tell everyone we should ask the UK to pay reparations and take in more Palestinian refugees 2021-05-17 23:50:49 @RunDaltonRun OK I've lived in the US for 30 years and am an American citizen, but you bet I'm going to get involved in the politics of my other country of citizenship, Israel 2021-05-17 23:50:07 @kamatsu8 I'm extremely involved in Israeli politics and I've lived in the US my whole life 2021-05-17 23:48:26 @kamatsu8 @ToddTheLinguist @wilbowma US relationship is just less complicated because of the Iran/Russia allyship 2021-05-17 23:47:25 @kamatsu8 @ToddTheLinguist @wilbowma This is another weird western take, the idea that the US has a lot of influence over Israel or Israel benefits so strongly from the relationship. Israel would be perfectly fine without the US, this is why Netanyahu has been sucking up to Putin so much 2021-05-17 23:43:35 @kamatsu8 It kinda does if I don't use my political capital to pressure the US 2021-05-17 23:39:02 @vzakhary Yeah I just wish we could all focus on getting Netanyahu out of power and into prison, and not replacing him with a Netanyahu Junior like Gantz 2021-05-17 23:37:54 @kernelmethod @noidd Even for Jews 2021-05-17 23:37:47 @kernelmethod @noidd My main problem with conflating Isrealis with the Israeli government is that the Isreali government has shown itself in the last few years to not be a democracy 2021-05-17 23:36:45 @kernelmethod @noidd I guess he is also a US citizen and I don't know if he denounced UK citizenship though 2021-05-17 23:36:11 @kernelmethod @noidd It is perfectly fair: if your country did a shitty thing, ask your country to repair the shitty thing, don't just yell at other countries 2021-05-17 23:32:26 @vzakhary Sorry for yelling I just am getting annoyed by the amount of caring everyone has about this issue without doing the corresponding amount of reading to understand the political and historical landscape and plan solutions 2021-05-17 23:31:27 @vzakhary So you can yell that "it's simple" on TV but only insofar as dictators gonna dictate. How do we get Netanyahu out of power? Who knows my family has been trying for years 2021-05-17 23:29:53 @vzakhary In any case IMO that is happening because Netanyahu wants to maintain power so he doesn't go to prison. A government was supposed to replace him but now won't because there of this. So Israel goes into its 5th or 50th election or whatever now instead 2021-05-17 23:27:54 @vzakhary Focusing on this latest wave of violence is a distraction, if things just go back to normal Gaza will continue to be unlivable and people will slowly die but the world won't care because it won't be flashy and exciting news 2021-05-17 23:26:27 @vzakhary PS https://t.co/BQXyfzAr43 2021-05-17 23:26:09 @vzakhary It's not "oh the UK starts so many problems too" It's "the UK literally started this problem why not start by getting them to change things?" 2021-05-17 23:24:36 "This is < No, if you're from the UK, ask your country to stop being awful and actually repair the damage they did to directly cause the issue, for example taking in refugees, paying much more in aid, etc 2021-05-17 23:23:02 @vzakhary Not quite, just annoyed he misses the obvious opportunity to ASK THE UK FOR REPARATIONS 2021-05-17 23:21:19 https://t.co/a3q92jQgef 2021-05-17 23:21:06 @ToddTheLinguist @wilbowma The only results of this kind of rhetoric will be 1) hate crimes against Jews, 2) resulting new waves of Jews immigrating to Israel for safety (see France 2014), 3) annoyed Israelis who vote further right 2021-05-17 23:17:58 Conflict benefits the Israeli right, isolation benefits the Israeli right, yelling a lot benefits the Israeli right, BDS benefits the Israeli right, spend like 10 minutes talking to an Israeli about politics 2021-05-17 23:17:02 Also yelling at people doesn't work and is what will keep pushing Israelis further right because Israel is a reaction to oppression of Jews, but have fun I guess if it makes you feel good 2021-05-17 23:15:27 @deckeresq You're fine 2021-05-17 23:14:45 @noidd Yeah, which drives me crazy because why does the UK, who started this conflict, get to have an opinion on this? What are they doing to actually make anything better 2021-05-17 23:14:03 I'm sure people will come in making weird assumptions about my political beliefs so https://t.co/5jb1Gls7we 2021-05-17 23:13:21 @ToddTheLinguist @wilbowma Me, that is. Democratic Israel/Palestine down with partitions. We need to talk about how to get there one day and still keep all ethnic groups safe 2021-05-17 23:12:05 @ToddTheLinguist @wilbowma And remember this is coming from someone who quite literally wants a single, democratic state 2021-05-17 23:11:35 @ToddTheLinguist @wilbowma Yeah I mean the things he's yelling are not helping they are just leading to more simplistic views 2021-05-17 23:09:47 @ToddTheLinguist Yeah but obviously what they mean by friends is "strategic allies" that's how geopolitics works 2021-05-17 23:09:01 I'll probably just interact a little less with you in the future 2021-05-17 23:08:10 Diverging from John Oliver here since he obviously has other political takes, moving on to Twitter recently. If you never Tweet international politics but are suddenly obsessed with Israel/Palestine, and aren't Palestinian or Isreali, I'm super suspicious 2021-05-17 23:06:24 So forgive me if I'm a bit suspicious if the only time people get this angry and think everything is this simple is when the country with half of the world's Jews is involved 2021-05-17 23:05:19 I also just never see westerners get this heated about other international catastrophes like the Rohingya genocide, the Uyghur genocide, or any of the times the US marches in and overthrows entire governments 2021-05-17 23:00:45 https://t.co/JlaOdxZqys 2021-05-17 23:00:33 @wilbowma It's a bad take and also an obvious and just way to make things better is to pressure the UK government into actually repairing any of the damage it has done by, for example, taking in any refugees who come from Palestine. But it's easier to yell on TV I guess 2021-05-17 22:58:14 @wilbowma Super looking forward to the anti-Jewish hate crimes that will result from everyone yelling a lot about how Actually This Is Simple though 2021-05-17 22:57:28 @wilbowma I don't know, probably spend his money on direct aid to Palestinians? Seems more reasonable. He makes a lot of money, right? Also https://t.co/6eVoMGMRAU 2021-05-17 22:55:45 The whole segment is absurd 2021-05-17 22:53:57 Imagine if the UK actually helped repair any of the problems it started 2021-05-17 22:53:20 @wilbowma Just like, what if the UK actually did anything to help with the problems it started, instead of just yelling at the people still there 2021-05-17 22:52:41 Though they wanted to colonize the the alone, so they spent some time attacking, tear gassing, and turning back ships of Holocaust refugees so they could keep them in camps in Cyprus for a while first 2021-05-17 22:48:14 Dude you're from the UK, the country that literally started this https://t.co/3ke5IuJDoM 2021-05-17 21:49:35 @hillelogram @johnregehr IIRC donations just go back to the sports program, though 2021-05-17 21:39:31 @johnregehr @hillelogram https://t.co/YP1L7vRyei 2021-05-17 21:38:28 @johnregehr @hillelogram Most universities lose money on football 2021-05-17 21:33:44 @hillelogram @johnregehr A lot 2021-05-17 21:30:43 @johnregehr @hillelogram Obvious low hanging fruit here https://t.co/9OUHUTpfe4 2021-05-17 21:28:49 @johnregehr @hillelogram How much do those people make? There aren't that many of them 2021-05-17 21:25:24 @hillelogram Also like the skills that go with being a good researcher are radically different from the skills that go with being able to manage grant budgets, organize things, hire and pay people, and so on. A university with faculty doing both would be such a mess 2021-05-17 21:23:05 @hillelogram Yeah like pay your college sports coaches less or whatever, but if the university is a research university, admins are essential to faculty and student productivity. Also accessibility for ADHD etc 2021-05-17 21:19:55 Seems like a lot of the "top 5" schools distinguish themselves in admin support that they offer 2021-05-17 21:18:49 I maintain that admins are the most underrated university employees and the whole system would collapse without them. Actually I think investing in more admins even for non-grant stuff like email and scheduling probably has huge returns on faculty research productivity 2021-05-17 21:17:06 I think I'll get the hang of the professor thing? Just sent an email looping in grant admins instead of forwarding to my advisor and asking "what does any of this mean" 2021-05-17 18:59:46 @taooftypes Thanks!!! Yeah I confuse myself more and more. The representation of the transformation was super syntactic which is why I've been trying to use Carlo's categorical intuition, but I want to do this in a way that doesn't just confuse people more 2021-05-17 18:58:21 @taooftypes Univalence helps~ in that it's the metatheory in which you can state correctness of the transformation, which preserves equality up to transport 2021-05-17 18:57:47 @taooftypes PLDI paper is here: https://t.co/ot12Kr0wwP But note that the categorical representation isn't in there, I'm trying to use it to give some intuition in my thesis, but it makes me realize I don't understand how to reason about the details that matter to me in my type theory 2021-05-17 18:56:20 @taooftypes That's nice because it's clean, but it also sucks because I can't point to a picture and be like, yes, this thing here, we call this iota, because nobody ever calls the proof that the square commutes anything. How do I reify it? 2021-05-17 18:55:43 @taooftypes You can do this for any equivalence by Lambek's, but you end up needing to represent eta and iota explicitly, and port definitional eta/iota to propositional eta/iota. But in the categorical representation, those show up as "there exists unique" and "this square commutes" 2021-05-17 18:54:53 @taooftypes My tool does this in a type theory that doesn't have univalence, and what it does is replace the old eliminator with the new one. This breaks definitional equality, which is why I need to port definitional iota over nat to propositional iota over bin (which I do) 2021-05-17 18:53:54 @taooftypes e.g., consider porting a proof about unary natural numbers to a proof about binary natural numbers, using the eliminator "shaped like" the eliminator for nat (similarly for constructors). Carlo figured out you can view this as following the map from the nat initial algebra, but: 2021-05-17 18:50:29 @taooftypes My work usually ends up externalizing things from HoTT in non-univalent type theories, using HoTT as a metatheory, but if I want to state correctness for finite cases in the type theory itself, I need to be very explicit about things that HoTT takes care of internally 2021-05-17 18:48:13 @taooftypes What is "modern type theory"? HoTT? It's just one type theory, there are so many type theories to choose from 2021-05-17 18:47:21 @taooftypes So the categorical abstractions always seem nice, but they're nice precisely because they give me something I can't easily actually implement. Are there ways around this? 2021-05-17 18:46:54 @taooftypes But sometimes I actually need to, like, think about definitional equality. I need to take things that were iota definitionally and move them to iota propositionally. And I feel like category theory doesn't give me a good way of talking about that 2021-05-17 18:46:19 @taooftypes Likewise the def eq question: Say I represent an inductive type as an initial algebra. But then in my type theory I actually _need_ to reason about eta and iota. Do you know where iota shows up? It's the proof that the diagram commutes---something that's just implicit :| 2021-05-17 18:45:09 @taooftypes It just feels like the necessary details are abstracted away, when in category theory two isomorphic things are treated as "equal." 2021-05-17 18:43:14 @taooftypes I think you're misunderstanding me. I'm saying, say I want to reason about a type theory that isn't univalent, like vanilla CoC. But I want to do so categorically. How am I able to represent the nitty-gritty details that matter? 2021-05-17 18:41:36 RT @geoffreyirving: @platypii @TaliaRinger Curry-Howard works fine for nonconstructive proofs, it’s just that the resulting programs take i… 2021-05-17 18:41:13 @MonniauxD Oh, sure, that's fine to me, I'm asking about other people's opinions 2021-05-17 18:35:17 @MonniauxD IMO yes 2021-05-17 18:24:59 @MonniauxD What if one exists, but isn't known yet? 2021-05-17 18:16:27 @MonniauxD if you had an interpretation of the axiom, you'd be able to compute your programs completely. there are subprograms that do compute 2021-05-17 18:15:32 *propositional 2021-05-17 17:57:24 @agdakx @taooftypes And if so, are those axioms finitely realizable even if not infinitely so? 2021-05-17 17:56:50 @agdakx @taooftypes Are there interesting axioms for which provably no computational interpretation is possible? 2021-05-17 17:53:41 @agdakx @taooftypes Yeah, I tend to agree. I'd call it a programming languages, and I'd still call programs that can't be run "programs." I think it's more desirable if programs run, but I don't think it's necessary to be able to call it a program. But I'm just now realizing this is weird 2021-05-17 17:50:00 Final month of a programming languages PhD is when you realize you're not even sure what a "program" is https://t.co/ooc8HsxgO5 2021-05-17 17:44:41 I'd love to hear any thoughts people have here in comments, too 2021-05-17 17:40:56 If you answered no, would you call a term that does fully compute in the same language a "program"? 2021-05-17 17:39:29 If you're working in a language in which not all terms compute (say, HoTT without a computational interpretation of univalence, so not cubical), would you still call terms that mostly compute but rely on axioms "programs"? 2021-05-17 17:37:22 @XenaProject There are a lot of PL/SE ideas that don't fundamentally depend on being able to compute---like program transformations, program refactoring, program repair, design principles, style guides, etc etc 2021-05-17 17:34:12 @XenaProject People might disagree with me here, so I wouldn't go around quoting me. But certainly the mathematical objects that we reason about as programming languages researchers, which I like to think of as programs, are sometimes in theories without computational interpretations 2021-05-17 17:30:20 RT @samth: @XenaProject @TaliaRinger I don't fully understand this line of work, but these ideas have been extended to more powerful axioms… 2021-05-17 17:17:40 @XenaProject Can only find a few unvetted results with type theories with DC/AC though, would be curious to read more 2021-05-17 17:15:36 @XenaProject It's just nicer to be able to have all of your programs compute. But they're still programs if they don't 2021-05-17 17:14:48 @XenaProject Whether you can interpret an axiom seems orthogonal. Like HoTT was still a type theory before cubical existed 2021-05-17 17:01:36 RT @samth: @JAldrichPL @TaliaRinger @XenaProject This is a good explanation: https://t.co/5QRGg82p71 Basically, take (P -> 2021-05-17 16:38:02 @platypii https://t.co/cCoRpSQAnc 2021-05-17 16:34:24 RT @smarr: We are hiring a Senior Lecturer/Reader (also known as Associate Professor) here at Kent. https://t.co/Hqpl3580oC Fields of int… 2021-05-17 16:29:30 @platypii Actually all people can 2021-05-17 16:27:42 Paging @jonmsterling, @carloangiuli, @admitscut etc this has always confused me a lot 2021-05-17 16:27:05 Like how do you represent definitional versus prepositional equality? And what when you don't have equality up to isomorphism as a form of equality at all? 2021-05-17 16:26:18 @BowToChris Category theory still seems to assume equality up to ismorphism as a primitive, no? Many languages don't have that, and even when they do, it's usually just propositional (must be proven explicitly) 2021-05-17 16:24:04 It's a nice abstraction to forget these things, but when you try to concretize sometimes it is too hard to do anything useful. I often get frustrated by these differences 2021-05-17 16:23:26 When reasoning about type theory categorically, how do people usually deal with differences in equality? For example, in real life you may sometimes need an explicit representation of a proof that the square commutes, or that a morphism is unique 2021-05-17 16:18:55 This link will be confusing to most people without category theory background but basically full is one direction of a thing, faithful is the opposite direction, this is a math pun. You can sub in "onto" and "one-to-one" for "full" and "faithful" basically 2021-05-17 16:16:31 Inside jokes can alienate people so here's the explanation of the joke https://t.co/StpxYpHhVR 2021-05-17 16:12:24 Talia moods: - math/logic jokes - let me explain this cool thing - deep and incoherent questions - intense political discussions - time to break all taboos - food 2021-05-17 16:10:57 All of this talk about people who are fully vaccinated, but where are the people who are faithfully vaccinated? 2021-05-17 16:10:18 @fullyabstract Fully and faithfully vaccinated 2021-05-17 16:05:04 Wow!! CDC just said vaccinated people can use the law of the excluded middle 2021-05-17 15:55:26 @HarmonicesMvndi Oh, interesting---I'm not sure off the top of my head, but it would be really interesting to look at examples if you have any 2021-05-17 15:51:19 @HarmonicesMvndi Induction for sure has a representation (by way of initial algebras) but it's kind of hard to work with IMO 2021-05-17 15:49:10 @RaphMim Wild to arrive at this conclusion from where you're coming from. Isolating and penalizing Israelis just pushes them further right 2021-05-17 15:43:43 RT @JAldrichPL: @samth @TaliaRinger @XenaProject Here's a non-paywalled version: https://t.co/gkWiXXI15a 2021-05-17 15:42:48 RT @samth: @TaliaRinger @JAldrichPL @XenaProject Griffin's paper is here (it's a great paper): https://t.co/9Mcg4awzeM The type of C (also… 2021-05-17 15:38:45 @JAldrichPL @XenaProject Sounds like a POPL 1990 paper, just waiting for the link But for real the most general form of CH holds, think very big about it 2021-05-17 15:35:15 @JAldrichPL @XenaProject OK so encode LEM directly. It was just an example 2021-05-17 15:34:37 @samth @JAldrichPL @XenaProject Can you point me to this? I'd love an easier example to show 2021-05-17 15:34:19 RT @samth: @JAldrichPL @TaliaRinger @XenaProject Pierce's Law is indeed inconvenient but the point stands even looking at more conventional… 2021-05-17 15:33:32 @JAldrichPL @XenaProject Peirce is just one example, if you want to directly encode LEM that should be feasible, the language will probably just not have desirable properties for programmers (maybe it will though!) 2021-05-17 15:32:03 @HarmonicesMvndi Yeah the correspondence extends to this thing called category theory which is basically a clever way of interpreting directed graphs 2021-05-17 15:29:06 @JAldrichPL @XenaProject OK but that's missing the broader point which is that you can encode logics as type systems and type systems as logics regardless of which logic or type system you fix 2021-05-17 15:25:42 @catnaroek It's the job of PL experts to abstract away confusing details from domain experts 2021-05-17 15:20:22 @ChrSzegedy @XenaProject Category theory does make some assumptions about equality that make this erase useful information often, though. I'm genuinely not sure what people usually do to resolve that discrepancy 2021-05-17 15:17:13 RT @EmmaSManning: Apparently (h/t @emilymbender ) this article quotes me as not being 'unduly concerned' about the idea. I want to clarify… 2021-05-17 15:07:15 @ChrSzegedy @XenaProject Wait until you hear about Curry-Howard-Lambek 2021-05-17 14:53:13 This does not fix a particular logical system, by the way, if you are coming here as a math person: https://t.co/DPly5ighaG 2021-05-17 14:52:04 Curry-Howard works for classical logics too, and it's important to understand this. It's a fundamental, deep relationship between programs and proofs that, at this point, is not at all limited by your choice in language or logic! https://t.co/o9f3LzhhjM 2021-05-17 14:49:56 @XenaProject The original Curry-Howard was about constructive logics, but if you change your logical basis, you just change your language / type system. The relationship is much deeper and more general than that! 2021-05-17 14:48:17 @XenaProject Common misconception, but it's not true! It works for classical too! See, for example: https://t.co/ojUEzyheBd 2021-05-17 14:43:23 @akravc The PLDI paper is my latest 2021-05-17 07:27:30 @noneuclideangrl People think we are just making up our history completely. In reality we recorded combinations of mythology and history and the contracts all at once and it's Canon 2021-05-17 07:22:35 If I see other nonsense I'll come back here. I'm continually surprised how little my very politically aware friends know about Jews. 2021-05-17 07:21:38 - "they should just go home" (that's literally why Israel is a thing to begin with) 2021-05-17 07:21:11 - "it's just a religion not an ethnicity they're just Polish people" (you can convert to Judaism, but it's an ethnoreligion, and ethnic Jews by and large share basically no genetic material with Poles or whatever) 2021-05-17 07:17:54 See: https://t.co/AXxXz4RDHq 2021-05-17 07:15:45 @hussains78 Except Israel isn't European 2021-05-17 07:15:26 - "Jews have no history in Israel before modern times" (even your "European" Jewish friends like me are like half middle eastern half Italian by ancestry actually 2021-05-17 07:13:15 Other actually antisemitic tropes to avoid while criticizing Israel, to make your criticisms better and also not hurt people in the process: - "Israel is just a bunch of white Europeans" (https://t.co/bHda1OfXWF, https://t.co/XMofkoOdIL) 2021-05-17 07:08:11 @DomSteinhoefel During the pandemic, not a lot, since I crave social interaction and end up trying to get it from Twitter lol 2021-05-17 07:03:50 @roger_witte Being underrepresented in tech is always like this you never quite feel like you were supposed to end up being a professor and stuff 2021-05-17 07:00:49 @DomSteinhoefel Yeah human-in-the-loop is one possible workaround to this challenge. The related work chapter discussion of my thesis will go into much more detail on the comparison 2021-05-17 06:56:28 RT @itamarok: @stubbornella We've had 4 elections in 2 years. His base is so strong and he manages to lasso just enough extreme right parti… 2021-05-17 06:54:49 My family has been asking this for years and yet https://t.co/xj1erakdPX 2021-05-17 06:54:20 @itamarok @stubbornella Well he knows as soon as he loses power he'll go to prison 2021-05-17 06:51:19 @stubbornella In this instance, because for as long as this latest iteration of bombing goes on, the government that was supposed to replace him won't replace him In general, because whenever he is about to lose power, he finds a way to keep it. He's basically Putin 2021-05-17 06:44:58 @deliprao https://t.co/WSPBEUh5nl 2021-05-17 06:37:40 I decided I could have chicken nuggets iff I also ate vegetables 2021-05-17 06:14:10 Seriously this is so accurate that I worry someone is following my life closely and using it to write Reductress articles 2021-05-17 06:05:40 RT @jshiplett: @TaliaRinger Reminds me of this old gem https://t.co/Ihw8C7KW0I 2021-05-17 06:03:28 Literally this in 2018-2019 but almost did not survive the mental health effects of the retaliation https://t.co/tth6tqcIUe 2021-05-17 05:52:22 Internal dialogue, just now: - I want chicken nuggets, but it's almost 11 PM :( - You can DO that! You're an adult!! 2021-05-17 05:20:31 Super regret not citing Frank Pfenning's thesis work in this talk, but also the intro about "inventing calculus" in quarantine and the true history of that is my favorite intro I've ever given. It's pretty funny https://t.co/7YnlvH7pFD 2021-05-17 05:14:36 1K people were like "oh maybe this weirdo who wears her heart on her sleeve and tells everyone everything to break down all taboos and gets weirdly excited about type theory actually knows what she is doing, or is good enough to trick people into thinking she does" 2021-05-17 05:13:16 lol I gained like 1K followers by becoming a professor. wild 2021-05-17 05:02:01 @NonMurkyConsqnc I regret not citing Frank Pfenning's thesis on proof transformation in this talk, though 2021-05-17 05:00:40 @NonMurkyConsqnc The intro to this talk I gave as part of the logic supergroup I think is helpful: https://t.co/L9tA7HKarY 2021-05-17 04:57:28 The orange links (I may change the color later) are indexed links to earlier definitions of those concepts so that all important uses of them are tracked and indexed and easy to find 2021-05-17 04:53:46 Still very pleased with this talk-to-paper part of the thesis, too https://t.co/qU5KZ6p9yL 2021-05-17 04:50:09 @theshawwn You could run off and dive into it! There's so much room for improvement here 2021-05-17 04:39:26 Though the thesis is an even weirder exercise in paper-to-talk-to-paper-to-talk 2021-05-17 04:38:06 OK it's late and I should go home, but I'm also really, really pleased with this piece of my thesis. This piece also comes from my job talk. I should really try the talk-to-paper order more often. https://t.co/B3jayzbE4K 2021-05-17 03:58:26 @joeyginorio @lreyzin At mine it showed up in an advanced optional PL course for late undergrads. It was like a one-day lecture but the kind that I got super obsessed with and never stopped thinking about 2021-05-17 03:51:11 @lreyzin Just wrote to him asking for coffee once I'm in the area (will obviously drive up to Chicago). Very cool! There are so many interesting people over there 2021-05-17 03:37:45 @lreyzin Oh, wow! Once I get out to UIUC, I'm hoping to spend some time in Chicago collaborating with folks over there. It'd be cool to try to end up there when he's around 2021-05-17 03:01:57 @csgordon @wilbowma "Guarded recursion" and "guarded corecursion" should both be agnostic to whether the guard is syntactic or not, no? 2021-05-17 03:01:07 @csgordon @wilbowma These seem like two orthogonal axes though: 1. recursive vs. corecursive 2. syntactic guard vs. internal guard 2021-05-17 02:54:00 @csgordon @wilbowma I'm so lost, especially given that the paper also mention's the syntactic guard in Coq 2021-05-17 02:47:50 @wilbowma @csgordon Oh OK I just use it to refer to recursion with a guard condition. What does it usually mean? 2021-05-17 02:47:06 @wilbowma I'll cite both, seems like the right move 2021-05-17 02:46:58 @wilbowma Ah it looks like it's a more limited notion of elimination though 2021-05-17 02:43:02 @wilbowma So what's the difference between this and the transformation used in Jesper's thesis? I'll cite both for now 2021-05-17 02:37:34 @csgordon I think it's a transformation defined in Jesper's thesis: https://t.co/SMQqppzBPa 2021-05-17 02:36:27 @csgordon Coq's guarded recursion is equivalent to elimination you just need to be clever about how to define the eliminator, and in particular need to generalize the motive through another type 2021-05-17 02:35:20 Maybe this @agdax thesis? https://t.co/SMQqppzBPa 2021-05-17 02:27:03 I also feel like someone mentions @agdakx every time I mention this 2021-05-17 02:25:46 I feel like it had something to do with Agda 2021-05-17 02:25:13 Yo type theory friends what's the paper again that defines the transformation showing the equivalence between guarded recursion and induction? 2021-05-17 02:19:47 @EmmaJuned @LucyMason217 They are not addicted to Twitter so we don't see them much 2021-05-17 01:58:02 @ChrSzegedy There are politicians who support one state solution on either side but both mean "but MY state" 2021-05-17 01:57:12 @ChrSzegedy Yeah exactly, it's the moral ideal and also a total pipe dream 2021-05-17 01:56:24 To Americans this is always about their moral positions or whatever but to those of us with skin in the game we need practical answers and everyone keeps failing at that, corrupt politicians who benefit from war maintain power and conflict, it's a total balagan 2021-05-17 01:55:24 BTW my democratic Israel/Palestine dream would have free and easy citizenship for all indigenous groups including Jews and Palestinians, and explicit recognition and protection of our languages and cultures. But it's nothing but a pipe dream, and that's the problem 2021-05-17 01:53:50 Like yeah it's complex and it is weirdly narcissistic to imagine that you have all of the answers 2021-05-17 01:52:40 And then the whole thing is really unprecedented. When in history has an indigenous group of refugees oppressed everywhere we have gone revived our mostly dead language and taken "go home" literally, displacing another indigenous group in the process, creating more refugees? 2021-05-17 01:51:14 Even then, for as long as Israel maintains military power, which they will do because states act in their own best interest, a Palestinian state is only nominally a state for as long as Israel still controls it 2021-05-17 01:48:53 The "two state solution" is kind of a joke because there are three territories and three governments, for one, and you can't have two states like that. With enough land swaps you could define two states but I don't see a way to draw reasonable borders 2021-05-17 01:47:49 When I say civil war here I mean you could theoretically just declare a single democratic state one day, governments willing, given some miracle, but people will immediately just fight again and converge toward multiple terroritories/de facto states 2021-05-17 01:46:25 And this response is fairly understandable given that historically, the motivation for most behavior isolating us was antisemitism, so even when it's not it just kind of triggers the same "OK, we'll take care of ourselves then, you know nothing" response 2021-05-17 01:44:32 And further because the histories are really messed up, and Israelis tend to react to isolating rhetoric and outside pressure by shifting further right, so I'm not sure anyone on the outside can really do anything about it 2021-05-17 01:43:34 "Complex" doesn't mean "symmetric." I say it is a "complex issue" because my ideal---a single democratic state---is not one that it seems possible to achieve without one of a miracle or a civil war https://t.co/US8U3ysHm3 2021-05-17 01:34:30 @kamatsu8 No you even have to resupply a bunch of info that has already been reported to the IRS separately by your employer 2021-05-17 01:31:42 What on earth is wrong with this country https://t.co/CNkDLJM5hI 2021-05-17 01:31:29 @MattAlhonte @abysmallytall Omg I hate this country sometimes and this is one of those times 2021-05-17 00:09:48 @PTOOP Point taken 2021-05-17 00:07:51 @PTOOP Yeah I'm not a doctor but I'd put a lot of money on you having ADHD hahaha. How do you deal with rejection? Ever get consumed by impulsive, intense, in the moment emotions? Do you have trouble stopping yourself from expressing them immediately? 2021-05-17 00:05:32 I think my favorite complaint I get often is that I manage to spend so much time on Twitter. That makes sense. Twitter is a dopamine machine!!! Designed to game the brain's reward system 2021-05-17 00:03:47 @tobycmurray Yeah, I don't know if you have ADHD or not, but there are a lot of things that intersect extra badly with gender roles. Like, societally I'm kind of a failure of a person if I can't motivate myself to do chores or cook 2021-05-17 00:01:58 @PTOOP I think a lot of researchers probably have undiagnosed ADHD. Do you ever lose like 16 hours of your life and forget to eat because the thing you're doing is so interesting you can't stop? 2021-05-17 00:01:02 @tobycmurray I hope one day I live in a society where I can like easily get accommodations for the things I need the most help with, but it's hard and ADHD accommodations that are "reasonable" to most employers are woefully outdated and the IRS for sure doesn't care 2021-05-16 23:59:58 @tobycmurray Yeah but ADHD regularly costs me thousands of dollars in failing to do things I'm supposed to do. Like file for reimbursements. It really must be nice to be able to easily choose to do things you don't like doing! OTOH I know ADHD is why I do such creative work so intensely 2021-05-16 23:58:20 @tobycmurray No, I'm just saying that "must be nice" is a perfectly reasonable reaction for anyone with any disability, because disabilities are socially defined to begin with 2021-05-16 23:56:48 @tobycmurray That's the ideal but society often doesn't allow for this, I imagine even more so for blindness! 2021-05-16 23:55:27 @tobycmurray If there weren’t advantages to being NT, then ADHD wouldn't be defined 2021-05-16 23:49:33 @abysmallytall How do you remember which ones are paperless or not paperless? How do you remember what accounts you have if they're paperless? How do you know any of your passwords? Or even usernames? 2021-05-16 23:47:56 @abysmallytall How do you remember you put it in the folder? How do you get yourself to put it in the folder when it arrives? 2021-05-16 23:45:28 @abysmallytall The hard part is convincing myself I'm even capable of finding the paperwork 2021-05-16 23:44:37 Specifically ~dopamine~ "individuals with ADHD tend to chronically have insufficient release and reloading of the neurotransmitter dopamine at synaptic junctions of neurons in the networks that manage executive functions" 2021-05-16 23:42:57 NTs like "everyone has this problem" no, you all have this weird brain thing that lets you choose to do unpleasant things 2021-05-16 23:41:14 @tobycmurray Like we can't just say "OK this will suck but let's do the thing" our brains don't do that 2021-05-16 23:40:44 @tobycmurray I'm sure everyone hates it but these things are way harder for ADHD people 2021-05-16 23:32:58 @Fish_CTO I just do the thing most interesting to me so I've built my work life around things most interesting to me 2021-05-16 23:30:30 @SMT_Solvers Forget arithmetic. Imagine finding documents! 2021-05-16 23:21:18 This for chores, taxes, buying groceries, and so on 2021-05-16 23:20:42 https://t.co/9ygeboKOO2 2021-05-16 23:18:35 Imagine being able to easily choose what you do or don't do. Must be wild 2021-05-16 23:17:25 3 hours and then brain is done existing for a while 2021-05-16 23:16:47 Neurotypicals are like "it only takes you 3 hours to do your taxes why not just do them" and it's like "no no no you don't get it, that's 3 hours plus two months straight of trying to convince myself every weekend it's a thing I'm actually capable of doing and should do" 2021-05-16 19:21:41 RT @kamatsu8: @TaliaRinger @DomSteinhoefel I wonder if anyone has tried to make a classical proof assistant with continuations? 2021-05-16 19:13:23 @kamatsu8 @michele_spo I think the confusion here is that "Curry-Howard" originally referred to a correspondence between a particular type system and a particular logic. But as we learned more, we generalized it, and now we use it to mean the correspondence between type systems and logics generally 2021-05-16 19:07:39 RT @DomSteinhoefel: Peirce's law is a version of the Law of Excluded Middle (does not hold in intuitionistic logic) and corresponds to the… 2021-05-16 19:07:24 @kamatsu8 @michele_spo I didn't catch it but I get it now 2021-05-16 19:07:12 @kamatsu8 @michele_spo Haha I hope so!! 2021-05-16 19:04:52 @kamatsu8 @michele_spo Yeah we are pendants this is why nobody likes us 2021-05-16 19:03:34 @kamatsu8 @michele_spo To be fair, type theorists can't agree on whether equivalent things ought to be equal 2021-05-16 19:02:02 @kamatsu8 Sure! And there are plenty of those but I think the general beauty of CH is the ability to switch between languages and logics, not any reliance on particular properties of those languages and logics. MP is just one most people seem to assume in most languages and logics 2021-05-16 19:00:50 @DomSteinhoefel Pierce's law, for example, corresponds to continuations. https://t.co/ojUEzyheBd 2021-05-16 18:59:35 @DomSteinhoefel So when people say "Curry-Howard" these days I think we usually mean "type systems and logical systems correspond to one another" in a general sense 2021-05-16 18:59:03 @DomSteinhoefel It's not wrong, it's just the original correspondence that was proven. Later on people drew correspondences between other languages and logics 2021-05-16 18:55:57 @kamatsu8 Aside from this, you can always build static types on top of dynamic ones. The correspondence still exists even when nothing is checking it. The knowledge that you'll get back an int doesn't need to come from an implemented type checker IMO 2021-05-16 18:53:25 @kamatsu8 I'd do it in parallel in a course but that's the norm for explaining CH and it leaves most people totally confused, so I wanted a complementary explanation without lingo and symbols and reliance on PL knowledge 2021-05-16 18:52:32 @kamatsu8 That part I think is actually a distraction for most people and scary for those who don't do well with formalisms 2021-05-16 18:52:07 @kamatsu8 Sure yeah, I just don't think most people even think about what the type checker is doing when they write programs. In a course it seems important to mention, in a quick summary I'm not convinced, I will be when someone outside the field gets confused by it 2021-05-16 18:50:50 @DomSteinhoefel Curry-Howard applies more generally! You just have to change the language and logic 2021-05-16 18:50:10 @kamatsu8 Importantly they're theorems that normal people have proven. And people can see how this corresponds to implication & 2021-05-16 18:48:51 @kamatsu8 Why? They are theorems. Just boring ones. 2021-05-16 18:48:09 @kamatsu8 Like similarly in an unsound logic, modus ponens might not hold. But I'm not comparing to an unsound logic, because I think most people IRL assume their languages and logics are sound. Even for modern unsound languages, you often have to try hard to break the type system 2021-05-16 18:45:11 @kamatsu8 I mean it depends on the point. But the point I was making was that this corresponds to modus ponens, they are two views of the same thing 2021-05-16 18:43:49 @kamatsu8 I guess you can also have an unsound type system that corresponds to an unsound logic, but then the translation doesn't buy you much in general, though instances of it do! Like there are unsound type systems where you'd have to try hard to force a mistake 2021-05-16 18:40:15 @kamatsu8 It's just a different frame on the same thing 2021-05-16 18:39:39 @kamatsu8 But when you have type soundness, you get the benefits of CH, so I'm not sure the benefits of soundness can ever exist in a vacuum apart from CH? 2021-05-16 18:39:04 @kamatsu8 Except for late in a PL course. But I'm going for folks who know little about PL and have just programmed in some languages with these things called types before 2021-05-16 18:38:01 @kamatsu8 Yes, but I think explaining this a distraction from giving the intuition 2021-05-16 18:37:04 @kamatsu8 And that is an illustration of the intuition behind CH. If you're frustrated that it's not rigorous enough, you're not in the target audience 2021-05-16 18:36:27 @kamatsu8 It's not explaining this as a fact derived from CH, it's explaining the relationship between this intuitive thing about programs in most reasonable languages, and a logical rule most people learn in undergrad discrete math 2021-05-16 18:35:17 @kamatsu8 The argument from CH is that this is really an instance of modus ponens 2021-05-16 18:34:40 @kamatsu8 Next time I'll make it explicit, I guess 2021-05-16 18:34:11 @kamatsu8 Oh sure I assumed a language with types, I thought that was implicit 2021-05-16 18:31:29 @kamatsu8 No that is limiting Curry-Howard to a particular logic which is just not a good thing to do IMO. Every type system is a logic. Some are inconsistent, some are consistent but boring, but the relationship is deep and omnipresent 2021-05-16 18:29:18 @kamatsu8 Type safety has a relationship to Curry-Howard, just like anything with types does. Limiting Curry-Howard to a particular type system or logic is a thing of the past 2021-05-16 18:14:37 RT @NewYorker: “The time has come for people to understand that what we recognize as reality is not necessarily what other animal species r… 2021-05-16 18:13:05 RT @AliVelshi: Vincent Chin was murdered on the night of his bachelor party for "looking Japanese." His murderers never spent a day in jail… 2021-05-16 18:07:34 I have to say, it's true there are a lot of things we don't get any training for before becoming faculty. BUT the interview process is really useful for learning to explain what you do to people who have never heard of your work before. 2021-05-16 17:58:34 RT @kapebeans: https://t.co/tsRE350pXE 2021-05-16 17:49:47 RT @jestom: a str8 woman friend said she would pay me for love life coaching but tbh my profesh advice is to be gay 2021-05-16 17:47:13 @monadic Yes it is basically an entire advanced programming languages course in a few Tweets haha 2021-05-16 17:23:51 @noneuclideangrl Equality is weirdly like one of the most complicated things in type theory. Experts spend lots of time arguing about it hahaha 2021-05-16 17:23:09 @noneuclideangrl Yes, I agree!! That's one of the details I omitted. It is incredibly counterintuitive IMO that you can define a type for equality the same way you can define numbers or lists. I hope to have an equality thread at some point. 2021-05-16 17:21:55 I did well with the classic, formal explanation with lots of Greek and lingo. My concern is that few people do, so I want to complement that with other explanations. 2021-05-16 17:21:07 BTW if a minority find it confusing but you are one of the people who does, this isn't a reflection of you---everyone understands things differently, and there are some confusing details I omitted that would have thrown off undergrad me, I think! 2021-05-16 17:16:48 Thread: https://t.co/m4PWAa9q2E 2021-05-16 17:16:12 You can give feedback on this thread here: https://t.co/E48lblB841 2021-05-16 17:15:27 What did you think about my attempt to explain Curry-Howard in a Twitter thread? I may do more of those for concepts that people in my field often complicate with lingo and Greek letters. 2021-05-16 17:09:52 @AJLUnited The tendency of people to blindly trust the decisions of programs in situations in which they would normally be skeptical of people making the same decisions. The view that programs are somehow capable of being objective and apolitical. 2021-05-16 16:58:19 RT @AJLUnited: What scares you the most about algorithmic or coded bias? 2021-05-16 08:35:54 Here is a cute thread if my research statement isn't accessible to you, but you want to know if you'd enjoy my kind of work: https://t.co/m4PWAa9q2E 2021-05-16 08:35:01 @indyjonas I'm not sure, honestly. When I write my first course on this, though, I'll keep it in mind. I took this approach explaining Curry-Howard to my Dad, who knows only Fortran and Cobol, and it worked, so I want to see how CS students deal with it. 2021-05-16 08:12:59 If this sounds cool to you, you'd probably enjoy being my student, and I'll need those soon 2021-05-16 08:12:24 Anyways that's just a taste. It's this rich, beautiful connection that lets you freely translate work between two fields. And also do cool, practical things with that power, like prove that your operating system is secure. Woahhhhhhh 2021-05-16 08:11:03 Wrote a nice survey paper on proof engineering, "QED at Large" a couple of years back. See my website. 2021-05-16 08:10:36 This is seriously underexplored---this side of Curry-Howard. There's this whole beautiful world of programming languages and software engineering. Take that, transfer it over, and we get proof languages and proof engineering. 2021-05-16 08:07:52 Actually, some things we do over programs are easier in these fancy type systems that let us write nice proofs. Because we can easily check we're correct in the end. And we get a LOT of hints along the way. That's super cool. 2021-05-16 08:06:23 You can take two proofs and look at the difference between them to determine what changed---like diff, but fancier. And from that infer something that helps you transform broken proofs into fixed ones. That's my thesis work, proof repair. 2021-05-16 08:04:59 You can do the same thing with proofs!!! You can write transformations that take as a proof, and return a new proof by transforming the original proof. A proof transformation. You can automatically optimize proofs. You can generalize them. You can go inside and pull out lemmas 2021-05-16 08:03:04 This is my FAVORITE THING EVER because it means that you can take things you do to programs, and do them to your proofs, too. Like, inside of compilers there are these things called program transformations. They often do cute things like optimize your code for you. 2021-05-16 08:01:13 So maybe you can see how programs are proofs. But there's an underappreciated fact that falls out of Curry-Howard, and it's the opposite direction. Proofs are programs. 2021-05-16 08:00:23 No, all you're relying on is a tiny little program that checks the types of your programs. We call this the "kernel." 2021-05-16 07:59:46 So then the only thing you're relying on to make sure your proof is correct is not, like, your friend or your advisor or your reviewer or your mom or whatever. They're complicated, they probably mess up sometimes. 2021-05-16 07:58:40 At that point, though, you've written a proof. The absolutely wild thing is that your computer can CHECK YOUR PROOF FOR YOU. All it has to do is type check your program! 2021-05-16 07:57:39 But you can actually write a function with this type in sufficiently fancy languages. And there's this thing called "proof automation" that helps you write that function so you don't have to think too hard. (That's the stuff I like to build!) 2021-05-16 07:56:17 Split them into cases, basically. Though this proof turns out to be a little annoying if you're naive about how you do it---you have to be careful about what you do next, and you might actually need to back up a bit. I won't get into that, it's a rabbit hole. 2021-05-16 07:54:37 Actually let's keep going. Assume arguments of types length l1 = m and length l2 = n. Our goal is to write a term of type: length (append l1 l2) = m + n Weird, right? Absolutely. But our language is super cool---it lets us do cool things like induct over numbers & 2021-05-16 07:51:25 Fix l1, l2 arbitrary list arguments. We want to write a function of type: length l1 = m -> length l2 = n -> length (append l1 l2) = m + n 2021-05-16 07:47:49 OK, with that in mind, let's revisit the type of the function we want to write: forall (l1 l2 : list T), length l1 = m -> length l2 = n -> length (append l1 l2) = m + n What does it do? Well, it will take in two lists as arguments. So let's modus ponens that shit. 2021-05-16 07:46:00 Dependent types are when values ARE used in types. So then we have to name them . So that's what forall is doing. 2021-05-16 07:44:56 What the heck is "forall"? Actually we already used it earlier, even for our most boring function about adding integers. We could have written that: forall (i1 i2 : int), int But we didn't bother because the values i1 and i2 aren't used in the type in the body, int. 2021-05-16 07:43:06 They write plain old list append first. Then they write a separate function with this type: forall (l1 l2 : list T), length l1 = m -> length l2 = n -> length (append l1 l2) = m + n WOAH that's weird hold on a sec what's going on 2021-05-16 07:40:34 And write a fancy append function that implements this. But this is kind of hard usually (without some fancy automation like my thesis work). So more often people split this up. 2021-05-16 07:39:30 There are fancy type systems that let you do this kind of thing. These are called "dependent types." You could write this a couple ways. Given n and m, you might say: list < list < list < 2021-05-16 07:37:07 You take a list of length m and a list of length n, for numbers m and n, and you append them. What do you get? A list of length (m + n). 2021-05-16 07:36:14 OK but what if you could do something super, super weird. Hear me out. What if you could also track the length of a list like that? So you want to know more than the type of your list at the end. You want to know something about its length too. 2021-05-16 07:33:56 Now if you give it two lists of numbers, you know you'll get back a list of numbers. Cool. 2021-05-16 07:33:31 Take your run of the mill list append function, which appends two lists of the same type and gives you back a list of that type. This proves: list< Which is slightly more interesting! 2021-05-16 07:32:23 OK, but now let's make our type system fancier. In Java you can have polymorphic lists, list< 2021-05-16 07:31:11 This is modus ponens: if P implies Q, and P holds, then Q holds. 2021-05-16 07:30:25 Every time you have a function of type: P -> which takes a P and returns a Q, you can pass it a P. What you'll get back will be a Q. 2021-05-16 07:29:30 Even with such a boring theorem though, you can already start to see logical rules fall out if you look closely. Like what happens if you pass your addition function two integers? You get back an integer. 2021-05-16 07:27:53 Let's write that theorem as: int -> Addition is one way to prove that theorem. But I bet you can also think of lots of other ways: Return the first number. Return the second number. Multiply. Subtract. Take the minimum. Take the maximum. 2021-05-16 07:26:32 In languages you're used to, you write proofs all the time, just kind of boring ones because your type system is not that interesting, usually. Like take your run of the mill integer addition function. It's a proof that if you have two integers, you get back an integer in the end 2021-05-16 07:25:50 If a term has a given type, we can imagine the term as a "proof" that it's possible to write a program with that type---so call the type a "theorem." 2021-05-16 07:24:02 Curry-Howard says that there is a correspondence between programs and proofs, in that the rules for type checking programs correspond to logical rules. It's the coolest thing I learned in undergrad. I'll try to explain a little. 2021-05-16 07:00:59 RT @acweyand: Remember when you wanted what you currently have. 2021-05-16 06:53:58 @noneuclideangrl https://t.co/gmcQH4NNLG 2021-05-16 06:51:53 @emremjones I was so pleased with myself 2021-05-16 04:03:51 Legit just typed up the transcript of my job talk in this chapter of my thesis a month ago, and it's honestly my favorite part of the thesis https://t.co/pwNGPlZrDK 2021-05-16 02:48:16 RT @AnimalsWorId: mood throughout the day https://t.co/ikFX0gvEDx 2021-05-16 02:47:23 RT @AnimalsWorId: smooth panda https://t.co/KWSTHgNPHw 2021-05-16 02:37:31 Update, an exception: if you are Palestinian and in my research community I will talk to you and listen because dialogue needs to happen, and I may be misinterpreting, and you deserve benefit of the doubt 2021-05-16 02:05:16 @deckeresq I don't think that is offensive, since it isn't about a traumatic event in recent Jewish history, and it doesn't uphold antisemitic tropes 2021-05-16 01:51:39 @matheussber Same, though people always seem to converge toward creating borders 2021-05-16 01:50:27 RT @hisham_hm: @TaliaRinger At the same time, Palestine was just about to hold its own elections again for the first time in a *long* time,… 2021-05-16 01:50:20 @hisham_hm Yeah, I can agree on that 2021-05-16 01:47:12 @hisham_hm He belongs in prison for many, many reasons 2021-05-16 01:47:01 @hisham_hm IIRC Netanyahu even deliberately stopped the policy of strategic targeting of leaders minimizing casualties, something Israel is fully capable of, because he benefits so strongly from Hamas staying in control of Gaza, and from the conflict 2021-05-16 01:43:10 @hisham_hm Yeah that's a huge part of it too. He is the worst and this time it literally did stop the government from replacing him. A government that was supposed to even have some Arab representation for once!! 2021-05-16 01:41:47 @hisham_hm I feel like those feelings can be legitimate in spite of my agreement that Israel started this, and especially this latest iteration, IMO so Netanyahu could cling to power a bit longer. I'm angry at Netanyahu, not the Palestinians 2021-05-16 01:39:58 @hisham_hm Still, I was there in 2014 and there were hundreds of rockets over a few days. I was really grateful for the bomb shelter in the house because we had to go there often, every time a siren sounded. It was very jarring, and my elderly grandparents often couldn't get there in time 2021-05-16 01:37:52 @hisham_hm Now obviously the offensive and defensive infrastructure Israel has is extremely disproportionate, and that is largely because of the blockade, and really the main deliberate purpose of the blockade 2021-05-16 01:36:11 @hisham_hm Our friends' house was hit today, he was just inside of the bomb shelter that houses have built in because he had enough warning time to get there https://t.co/fjR7JB4lrq 2021-05-16 01:22:45 @hisham_hm If you assume I don't know the history myself maybe just ask, but I do, and I hate it 2021-05-16 01:21:40 @hisham_hm OK, just if I say my dad is from al-Majdal everyone who knows what that is will think I am Palestinian, and everyone else will be confused. I'm fine giving history here and the history is awful, just not in threads about worrying about family and friends 2021-05-16 01:18:33 @yoavgo Yeah I don't know they just show up when I saw "I'm worried about my family and friends." Maybe next time I will Tweet in Hebrew 2021-05-16 01:17:05 @hisham_hm It sounded like the response to "I'm worried about my friends" was to justify them being hurt 2021-05-16 01:14:37 OK new rule, if you post anything to the effect of Holocaust denialism, Holocaust comparisons, or implicit or explicit suggestions that my family and friends deserve to get hurt, I will instantly block you with no explanation (unless you are a student) for my wellbeing 2021-05-16 01:07:40 @hisham_hm Or are you just assuming I have political views that I don't have? I wonder why? 2021-05-16 01:07:14 @hisham_hm I wonder if I deserve to be shot too for living on native land in Seattle? 2021-05-16 01:05:37 @hisham_hm Like I can't worry about my family and friends at the same time? 2021-05-16 01:05:19 @hisham_hm Yep, what is your point? I don't even support the notion of a Jewish state 2021-05-16 01:01:38 Like it turns out that just isn't how things work. Every country needs to do their part and accept as many refugees as they can support or the cycle of people displacing and killing one another will continue until humanity destroys itself 2021-05-16 01:00:46 The worst part about this logic is that the entire creation of modern Israel is one of the world's oldest refugee groups taking "go back where you came from" literally after 2000 years of violence for our existence https://t.co/kepcuoWe7C 2021-05-16 00:42:38 Know your history. He had no family and no home. Many did not. There was no place to go home to. If you have a problem with it, advocate for all countries accepting more refugees 2021-05-16 00:39:54 Eventually he went to his old work camp and found someone still there that he knew, who hooked him up with a Zionist group, the only people willing to take my grandfather in. After feeding him and educating him, they helped him illegally immigrate to British Mandate Palestine 2021-05-16 00:39:00 So he squatted in abandoned German houses and stole their leftover food. When he got back to Poland he had nothing left there so he was homeless for a long time 2021-05-16 00:38:22 You know what happened when the Soviet Union liberated my grandfather's death camp? They just left him there with a tiny sum of money. He went by horse & 2021-05-16 00:35:35 Friends tell me when you get deported to a death camp does your house just stay there so you can go back to it 2021-05-16 00:34:58 "Your family can just go home" lmao what home Why do you think they're in Israel to begin with? https://t.co/uayEvsMcT9 2021-05-16 00:33:34 @Desoxirriisopo My homeless refugee grandparents? What home did they have to go back to? They were stolen in the Holocaust Or their children? They were born in Israel soooooo 2021-05-16 00:30:02 @xwangsd @wilbowma Yeah writing a survey paper helped a lot In general I've always been good at vision and planning how to achieve vision. I struggle more with lower-level research tasks. But I had a ton of help from an industrial partner 2021-05-15 21:25:55 @wilton_quinn My Dad was in the navy and spend the whole time grenade fishing, basically never had to do anything. I had a friend who was deployed as a medic in Gaza in 2014 and he is the biggest hippie I have ever met now, extreme left, it changed him completely 2021-05-15 21:24:55 @wilton_quinn Most programmers end up with office gigs in intelligence 2021-05-15 21:24:13 @MichaelDoronII https://t.co/8dMCLMfuid 2021-05-15 21:22:42 @wilton_quinn I was drafted and had to have my guidance counselors at high school sign a form saying I live and attend school in the US in order not to serve 2021-05-15 21:21:02 @wilton_quinn Yeah also there is mandatory service and to get an exemption you basically have to either move abroad or choose not to participate in the economy (hard to get a job without service) which is asking a lot 2021-05-15 21:18:52 @wilton_quinn I feel like I rarely get personally judged for the terrible actions of the US military abroad, but I frequently get judged for even saying my family is from Israel, and I don't understand that 2021-05-15 21:18:05 @wilton_quinn And also I wish people would be careful not to conflate people with their governments, especially when democracy there has slowly been eroding 2021-05-15 21:17:15 @wilton_quinn I just wish people could be cognizant of the fact that you can't choose where you are born or, in my grandparents' case, where you will have access to a home as a refugee 2021-05-15 21:08:21 @wilton_quinn Thank you. Yeah it is hard for me to separate the political and the personal because this is always personal to me 2021-05-15 21:05:56 @SouthernSummr The houses have shelters in them. The hard part is a lot of elderly people can't make it to the shelter in time, you have like 15 seconds. That is why my grandparents left Ashkelon 2021-05-15 21:05:09 @SouthernSummr Yes everyone is OK just property damage 2021-05-15 21:02:14 Our friends in Ashkelon are not doing so hot though, houses of friends have been hit 2021-05-15 20:33:17 I feel perfect today because of this. I'd just nap in the sun all day if I didn't need to do taxes and revise my thesis 2021-05-15 20:31:11 @DrKeithRamsay Yes I am starting in October for this reason! Delays tenure clock and gives me time to ease into the job without teaching until January 2021-05-15 20:21:05 RT @koronkebitch: if yall need ideas for what to do with conference t-shirts https://t.co/PZWScne4VH 2021-05-15 20:21:03 @koronkebitch Omg want 2021-05-15 20:10:14 @NuritPeleg Yes, thankfully my aunt and uncle and cousins all live across the street from her 2021-05-15 20:05:13 I'm not worried about her getting hurt---she deliberately moved away from Ashkelon, and also deliberately moved into a house where her bedroom doubles as the safe room/bomb shelter. Just worried about missing the chance to see her given her age 2021-05-15 20:03:33 I've been to Israel during rocket fire before, in 2014 in particular, but what is happening now seems even more serious 2021-05-15 20:02:47 I feel bad complaining about this given how much worse things are for Palestinians, but I didn't see my grandma in Israel all through the pandemic and she is old, so I always worry I will lose that chance. I was really excited to see her in July, but I'm not sure I'll be able to 2021-05-15 19:45:35 I'm so extroverted and social how did I survive this year 2021-05-15 19:42:10 Sunshine & 2021-05-15 18:14:31 Very incomplete list, of course. Genocide seems par for the course in human history unfortunately 2021-05-15 18:12:46 Here is a handy list of genocides in case you want to compare something to genocide but can't think of anything except the Nazis https://t.co/eu4b5YskD5 2021-05-15 17:59:41 @StealthButterf1 @sibinmohan He also listened though 2021-05-15 17:40:44 @sibinmohan @jacob_chrn Thanks, appreciate it 2021-05-15 17:37:32 @sibinmohan @jacob_chrn Yeah but that is extremely traumatic for a lot of Jews (these are our grandparents you are talking about) and plays into the antisemitic "Jews are hypocrites and only care about themselves" trope. Also the people in power in Israel are largely not Holocaust refugees & 2021-05-15 17:32:38 @sibinmohan @jacob_chrn I hate Bibi too, so does my whole family in Israel, but somehow he keeps managing to hang on to power even when people vote him out. He's like Putin 2021-05-15 17:31:17 @sibinmohan @jacob_chrn I think a mix---the actions of Israel are bad, but using the Holocaust history to criticize it is unnecessary and plays into common antisemitic tropes. I know you don't mean it that way, I just think it's best to stay away from that particular frame of what is happening 2021-05-15 15:46:10 RT @schemeprincess: @koronkebitch I had a necklace for each of my grad school papers: Sapphire, TAPIR, Diamond. Demikernel is probably goin… 2021-05-15 15:40:58 I'm about to see some of my coworkers for the first time in over a year!!!!! 2021-05-15 15:21:57 Thanks, I see the sparkle now! 2021-05-15 15:17:54 @jacob_chrn @sibinmohan https://t.co/9MBYvwdZgO 2021-05-15 15:16:00 RT @schemeprincess: This is what a systems hacker looks like. https://t.co/Z3KDXb0We7 2021-05-15 15:15:47 RT @QuantaMagazine: A conjecture posed at a 1972 tea party by mathematicians Paul Erdős, Vance Faber and László Lovász has finally been pro… 2021-05-15 15:13:52 I know how these tools work and without deliberate intervention they end up dividing friends up into echo chambers that reinforce biases. I don't trust any company to set the incentives right 2021-05-15 15:11:56 For a while I had turned off personalization and my timeline had been chronological. But now it appears to be learned again, and I can't find the option to disable that. Am I missing something? 2021-05-15 15:08:04 RT @littmath: Several people have explained to me that there is a better version of this joke There are two guards blocking… 2021-05-15 15:00:25 @hanalyst I tripped on a piece of my disassembled broken toilet and hurt my toe as I was reading this Tweet 2021-05-15 14:51:37 RT @jcreed: category theory Q: if I have a monoidal category where *every* object is uniformly a monoid (i.e. the multiplication m_A : A ⊗… 2021-05-15 06:16:04 Admins are the most underrated, S Tier https://t.co/6lwKqzMQO2 2021-05-15 06:14:18 @wilbowma Yeah faculty would be worthless without admins 2021-05-15 06:12:03 @Foone https://t.co/j2snj2ctod 2021-05-15 06:06:53 RT @sarah_zrf: happy belated birthday to this tweet 2021-05-15 05:58:30 @wilbowma Admin are so underappreciated 2021-05-15 05:57:14 @wilbowma So I mostly mean like the actual logistics of paying students once you find them lol 2021-05-15 05:55:04 @wilbowma I also want to pay my graduating undergrad coauthor this summer for a few hours per week of work but I don't actually know how to do that 2021-05-15 05:54:02 @wilbowma I'm in a similar situation actually, I have a grant with a kickoff in July, and trying to coordinate that I realized how little I know. Then I was comforted by how little my advisor also knew, other than "we need these 3 people to talk to each other and sort it out" 2021-05-15 05:49:26 @wilbowma How did you figure out how to like, recruit and pay students? Which people to contact about what? Like what happens between accepting the offer and the first day of your job 2021-05-15 05:48:00 @kjw_chiu I think I get 2 official ones from day one. Just not sure when and how I choose them 2021-05-15 05:40:21 So you just like, show up and you're a professor one day? There are so many things I don't know 2021-05-15 05:12:51 @ChrSzegedy Thankfully proof of work is dying 2021-05-15 05:10:24 @ChrSzegedy IIRC it's also a proof of work currency 2021-05-15 05:09:24 RT @IanColdwater: Did you know that geese have teeth on their tongues? Now you do. Sweet dreams 2021-05-15 05:09:00 RT @grassmannian: I forgot how awful commutative algebra is!!! 2021-05-15 05:08:44 RT @AlexJohnLondon: Holy Shit: CDC now saying that fully vaccinated people no longer need to switch doors in the Monty Hall Problem! 2021-05-15 05:05:51 @ChrSzegedy Dogecoin millionaires 2021-05-15 04:44:46 I can't believe in 2021 there are people getting rich off of environmentally unsustainable memes. Who wrote the script to this show it's unrealistic 2021-05-15 04:26:11 RT @KennKnowles: The CDC says fully vaccinated people can merge code that "looks ok" without waiting for tests to complete. 2021-05-15 04:10:52 RT @AnimalsWorId: I don’t know who needs to see this but https://t.co/91pkJxF4eM 2021-05-15 04:04:43 RT @SkippersWif: The AWS Rust team is looking for an Applied Scientist to reinvent programming language design. We want to meet developers… 2021-05-15 04:00:33 Will post more details when I actually plan the thing 2021-05-15 04:00:17 Hi friends in Seattle, for graduation since I don't get to walk and get hooded, I want to have an outdoor professional photoshoot with friends in June. Vaccine please, wear a gown if you have one, otherwise have as much fun as you want dressing up, be yourself, make it fun 2021-05-15 03:03:16 @wilcoxjay @pjreddie Did you just go back and find this Tweet to quote yourself 2021-05-15 03:01:56 @bhatele That sounds nice, but also potentially difficult to find things! 2021-05-15 03:01:22 @jaredweakly No I don't like using tools to write text, I just use gedit and compile by command line. I just never understood the point of having consistent keys until I needed to write a thesis 2021-05-15 02:55:49 @pjreddie please do this 2021-05-15 02:46:16 @SMT_Solvers oh yeah I'm not using any more complicated systems 2021-05-15 02:40:32 I would not wish this pain upon anyone. Use consistent keys in your .bib files across papers, friends 2021-05-15 02:37:01 After two of the most boring hours of my life, I now have deduped all of the citations in my thesis 2021-05-15 02:23:00 @samth Wait until you hear my views on the two-state "solution" 2021-05-15 02:11:54 @sarielhp Next time borders need to be fractals 2021-05-15 01:45:32 @thatinterlace Yeah that's a good summary 2021-05-15 01:44:37 @samth Yeah but also like look at it 2021-05-15 01:33:55 Like who was on your ad-hoc committee, the India/Pakistan guys? Come on 2021-05-15 01:31:32 Like that's not how states work 2021-05-15 01:30:52 Can we all agree though that the UN was high when it was like "oh yeah, this is going to work" https://t.co/B1KJonqroK 2021-05-15 00:32:40 @PTOOP Of course. But it's OK to answer "any" too :) 2021-05-15 00:32:03 https://t.co/aAQSZZvfwT 2021-05-15 00:29:40 If you want to get specific I'm a "grey agender woman." Better to just call me Talia and go with whatever pronouns you feel are right on any given day I think it's nicer that way, though I'd rather language not force these decisions 2021-05-15 00:28:00 I guess one more, 3) my gender identity is so weak that this gives me more fluidity in presentation and perception on days that I do feel like being masculine or feminine or whatever 2021-05-15 00:25:21 @hipsterelectron Also, it gives me a chance to change how I'm perceived and let other people infer pronouns they want to use, and so on. I just really genuinely prefer to leave it up to others lol 2021-05-15 00:24:31 @hipsterelectron If at some point my gender identity strengthens I want to know. The way I found out I was grey agender to begin with is when someone used he/him to try to insult me and I didn't notice until like 10 text exchanges, and when I did I felt nothing about it 2021-05-15 00:22:35 @PTOOP I think anyone who feels better leaving it up to others should do so, and anyone who doesn't should state it if they feel comfortable doing so, and not state it if they don't feel comfortable doing so 2021-05-15 00:18:43 @hipsterelectron https://t.co/aAQSZZvfwT 2021-05-15 00:17:56 @PTOOP Sure, and I say "anything is fine" 2021-05-15 00:03:12 2) It gives people freedom to use pronouns politically and usually people whose politics align with mine are the most willing to do that so I want to just let them do that 2021-05-15 00:02:04 1) Every time someone uses a pronoun, it gives me a sense of how I feel. If at some point I feel strongly about a given pronoun, I've learned something new about myself. I want to learn lots about myself always 2021-05-15 00:01:24 I don't have strong enough gender identity except politically (too many men in CS) to feel like it makes sense to give myself pronouns. I'd prefer people use whatever pronouns they want for two reasons 2021-05-14 23:59:51 I have no pronouns in bio because I prefer people assign whatever pronouns to me they want to use if asked I say "any" https://t.co/cmhT2SCXXz 2021-05-14 23:51:46 @polikarn The VEC has been formed to handle that 2021-05-14 23:43:43 RT @Anne_Spencer_: An important read on barriers disabled faculty face in academia https://t.co/832hpaoxjQ 2021-05-14 23:39:33 I still haven't gotten reimbursed for two of my visits because I somehow have to find my receipts and for one of them I even have to take a picture of a tax document? But I don't know where any of those documents are, and no I haven't done my taxes yet 2021-05-14 23:37:56 Like that quantifies the cost I'd pay to not have to do paperwork, and it's roughly my entire paycheck 2021-05-14 23:36:49 Honestly with ADHD getting reimbursed is such a nightmare that I often just swallow the cost, which I'm lucky to be able to do https://t.co/Fi4LIP0BnQ 2021-05-14 23:35:48 Me once I finish my thesis https://t.co/4GVFZHvVZF 2021-05-14 23:30:39 @asajeffrey Close! You win, though not much competition. 9:09 2021-05-14 23:30:00 RT @Reductress: How to Overthink Every Achievement Until It Becomes a Bad Thing: https://t.co/CeqYXQhrf4 https://t.co/cQNbj7pUMh 2021-05-14 23:01:53 I would categorize my position as "hopeless and disillusioned" 2021-05-14 23:01:18 If what I want is a single, democratic state with explicit protections for all indigenous groups including Jews and Palestinians (plus the Druze and so on), but I also know any attempt at that will lead to civil war and two territories again, what does that make me? 2021-05-14 23:00:18 I also find the labels "pro-Israel" and "pro-Palestinian" confusing. Like what do you actually support? 2021-05-14 22:35:11 @PTOOP Also I think your definition needs to be refined to adults unless you believe children should vote in democracies (I'd be willing to hear that out honestly) 2021-05-14 22:34:11 @PTOOP Are there any that don't disenfranchise people? Like in the US "democracy" felons can't vote 2021-05-14 22:30:57 @rakamaric Though I am very tired of every American having a strong opinion on this without studying the history at all, and have gotten some hot takes this week saying my family deserves violence 2021-05-14 22:29:59 @rakamaric Mine is basically all the Palestinian perspective. It is probably a function of who your social circle is. I follow a lot of other leftists who are sharing that perspective. I appreciate it because I don't get it from my family 2021-05-14 22:07:48 I know the answer now, I'll go for a run and you all have another hour or so to guess 2021-05-14 21:47:50 Just wrote a first draft of my 5 minute PLDI talk. Who wants to guess how long it actually is? (I don't know yet, but I'll practice and let you know. Closest without going over will win, Price is Right style.) 2021-05-14 21:46:45 RT @johnregehr: huge congratulations to @jubitaneja for successfully defending her PhD thesis today!!!!! 2021-05-14 21:46:42 @johnregehr @jubitaneja Congrats Jubi!!!!! 2021-05-14 21:41:44 @gcruchon The risk of transmission from vaccinated people is basically negligible 2021-05-14 17:19:32 @awesomeintheory A good takeaway would be, "we really need to take refugees in everywhere 2021-05-14 17:18:18 @awesomeintheory He didn't have an education past the age of 10 except for the education Zionist groups gave him, and it's not like the internet was a thing. I'm not sure how he was supposed to know anything about the internal situation in British Mandate Palestine 2021-05-14 17:17:23 @awesomeintheory Yeah someone said this to me and also said my family deserved the violence because of that. But Saba was homeless after death camp liberation and the Zionist groups were the only ones willing to take him in and help him. And once in he just kind of farmed the negev 2021-05-14 17:15:36 If basically all faculty feel like impostors, are the real impostors the ones who don't feel like impostors? 2021-05-14 15:50:35 Protips: - you can criticize Israel perfectly well without making Nazi comparisons - making Nazi comparisons about basically the only place that most Jewish Holocaust refugees had to go is _actually_ antisemitic 2021-05-14 15:31:32 CDC just said fully vaccinated people don't need to write proofs in their POPL papers 2021-05-14 15:23:57 RT @AlejandroPiad: CDC just said fully vaccinated people can train and test on the same data! 2021-05-14 15:08:16 @jonmsterling @bernpton Women with children, to be clear, because their partners usually don't put in equal work 2021-05-14 15:06:52 @jonmsterling @bernpton Among professionals, I've also seen lots of unemployment among those unable to work from home who had to quit jobs or got fired. Mostly women and people with ADHD or depression. Even grad students 2021-05-14 15:03:11 RT @jonmsterling: We've just released a revised version of our paper "A Cubical Language for Bishop Sets" about XTT. I have to say that one… 2021-05-14 15:01:53 RT @harriesadam: @TaliaRinger You did write the rest of the thesis so that the first letter of every paragraph forms a coherent abstract, d… 2021-05-14 14:52:47 @bernpton The goal is to incentivize people to get vaccinated. Not sure it will work. Community outreach via inside leaders is what worked in Israel 2021-05-14 14:48:07 @bernpton And we have cards. But I haven't seen it done yet 2021-05-14 14:47:45 @bernpton In WA, I think it mostly matters at the business level. Businesses are legally allowed to require masks, but also for unmasked people to inquire about vaccination status (and kick them out or require masks if they don't oblige) 2021-05-14 14:46:14 @jonmsterling The knowledge package is your friend 2021-05-14 14:43:23 @adamberaha That's ridiculous 2021-05-14 14:37:09 @yoavgo Yikes 2021-05-14 14:33:39 I'll probably still mask in public for a while (until numbers are close to zero here) because strangers have no way of knowing I'm fully vaccinated without asking, and I don't want them to feel anxious. I'm surprised that isn't coming up more 2021-05-14 14:24:34 @jonmsterling Makes sense, HoTT gave everyone such beautiful new ways to imagine types 2021-05-14 14:19:20 RT @provisionalidea: If you're using a tool you're familiar with, and using the tool properly (i.e. not grabbing a fork by the prongs), you… 2021-05-14 07:40:36 @HerrDreyer @jeanqasaur @jonmsterling Yeah I find Q& 2021-05-14 07:39:20 RT @plt_abbie: who called it the "very small type theory community" and not "several people are typing"? 2021-05-14 07:36:57 @jonmsterling What is the new generation of research in type theory? 2021-05-14 07:34:16 RT @jeanqasaur: For anyone who needs to hear it: it's way easier to ask a question that makes someone feel bad about their work than to ask… 2021-05-14 07:33:41 @HerrDreyer @jeanqasaur In a conversation, I love being asked about overlap with other problems, but not in a Q& 2021-05-14 07:32:22 @HerrDreyer @jeanqasaur In general I love being asked about future research plans, vision, the cool technical details I glanced over in a talk because I wasn't sure anyone would care ("how do you deal with the challenges of definitional equality?") 2021-05-14 07:30:00 @HerrDreyer @jeanqasaur The best question I ever got was from Carlo & 2021-05-14 07:28:03 @HerrDreyer @jeanqasaur I hate Q& 2021-05-14 07:18:09 @inihelene Sometimes it really can take years and that can be a normal reaction to a bad breakup 2021-05-14 07:11:07 RT @jeanqasaur: Taylor Swift as notable #AAPI computer scientists. Andrew Yao, 2000 Turing Award Winner. Known for proving Yao's Princip… 2021-05-14 03:44:20 Ok my whole thesis draft is done, save for half of my giant acknowledgments section, which I'll get back to on Monday. I have a week and a day to revise it before the initial deadline so that's nice. Just also somehow need to write a PLDI talk and a defense lol 2021-05-14 03:35:21 @ezyang Dude mine is already 4 pages long and I am maybe half done 2021-05-14 03:19:22 If you've helped me in any way through my grad school journey, including just by being there for me, can you please remind me so that I don't forget to acknowledge you explicitly? I have a long list, but it's hard to make sure I list everyone. 2021-05-14 01:19:05 @TheDavisJam Sure! I'll be submitting a draft in a week, and hopefully have it revised by my defense in June 2021-05-14 01:18:16 RT @mkulkarni: @TaliaRinger The Internet has made it very easy to make dumb/awesome things very quickly: https://t.co/NeM5YCeBaj 2021-05-14 01:17:59 @mkulkarni omg haha 2021-05-14 00:27:29 Some of the language is my existing papers that my thesis work builds on or from my research statement, if it sounds a bit too polished for a first attempt 2021-05-14 00:22:09 My first attempt https://t.co/qRd5aGJczb https://t.co/9Y5ULsTb1y 2021-05-13 23:33:10 I like to start with what's practical and work backwards to what's beautiful. Then I like to use the beautiful to drive new practical things I hadn't even thought to be possible. And then I like to iterate on the beautiful-practical cycle forever. Somehow I get paid for this 2021-05-13 23:31:08 @kamatsu8 Same but not snacking out of boredom, just unable to motivate myself to exercise enough when there are no track practices and no races to compete in 2021-05-13 23:10:20 Man, the only parts left of my thesis before I have a somewhat reasonable end-to-end draft are the abstract and acknowledgments. But how do I write an abstract for my thesis? Sounds overwhelming 2021-05-13 23:05:02 @alpha_convert I buy Soylent to add a little extra food during depressive episodes, usually stops me from losing weight 2021-05-13 23:03:47 And if your depression is that bad, also, I'm always down to talk and help you find resources and so on 2021-05-13 23:02:18 For real though I certainly hope judging grad students for attire and weight is not very common. Be comfortable, work your best, live your life 2021-05-13 23:01:04 Dark humor, but my solution to this problem was to go through such serious depression that I lost all of the gained weight in a couple of months. YMMV https://t.co/ZXfvnIbRZV 2021-05-13 21:56:27 RT @bariskasikci: If you are a grad student with close ones in Israel/Palestine, and if you are worried the current situation makes it impo… 2021-05-13 19:18:53 RT @plfmse: @plfmse faculty, Tianyin Xu (@tianyin_xu) is featured at @Facebook Academic of the Month. He shares his experience working at @… 2021-05-13 18:51:50 https://t.co/yoaMQCcoo8 2021-05-13 18:51:15 Things to talk about might be: - why you want a PhD - what kinds of problems you think are cool - what about my research interests you Don't worry too much about not knowing enough. It's fine to say that upfront and just let me know something that sounds cool at a high level 2021-05-13 18:49:08 If you contact me without those, it's not annoying or anything, so don't worry if you've done that already. It's just that I'm more likely to gloss over the email or DM 2021-05-13 18:47:30 https://t.co/Y2rBExILgE 2021-05-13 18:46:11 I'm getting a lot of messages about wanting to work with me and it's wonderful, but also really hard to tell which inquiries are serious. I recommend reading my research statement first (https://t.co/5lIGoNyXW2) and saying something about what your goals and interests are 2021-05-13 17:19:50 I'm glad you all want to be experts in Israel-Palestine now but I'd like to talk to you about the history and possible long term solutions when I have the energy 2021-05-13 17:18:03 @hillelogram That's cool but like IDK what the other possible moves were. The Zionist groups were the only ones helping the refugees 2021-05-13 17:17:22 @hillelogram I'm like hard left by Israeli standards by I've seen some takes today like "your family deserves this" for trying to find a place to safely live after fleeing a nazi death camp, and ending up on a Kibbutz in the negev after illegally immigrating to British mandate Palestine 2021-05-13 17:12:46 I wish people got this passionate about other forpol 2021-05-13 17:12:05 Very close to muting the words "Israel" and "Palestine" this week jfc 2021-05-13 17:03:25 @HilaCodes 2021-05-13 17:03:12 RT @HilaCodes: This is just a reminder that if you have students/colleagues/friends with family in Palestine or Israel, and especially in… 2021-05-13 16:58:21 @2BitSalute Also we can totally afford to help people who need it. Actually in many cases, not helping people is more expensive. For example, housing-first policies save money in the long run since housed people contribute back later. We basically pay a tax to keep people unhoused 2021-05-13 16:56:59 @2BitSalute Immigrants are super hardworking people grateful to have a better place to live 2021-05-13 16:49:02 @2BitSalute Countries are not companies 2021-05-13 16:45:06 @2BitSalute Like, it's not just about being careful---I shouldn't have the power to be careful or not careful here. The immigration status of another person should not depend on all of these things. Just let people immigrate, why do we police immigration? 2021-05-13 16:43:55 @2BitSalute I think it just points to the system being broken. A lot of people try to apply for the sake of immigrating, not for the sake of doing research in x, y, and z. Lots of messages and emails about this to all faculty. But it points at being a part of a broken system 2021-05-13 16:41:15 @muscovitebob How about we just let people who want to be American become American, as long as they pay their taxes 2021-05-13 16:35:10 Man you can tell our immigration system sucks when by choosing which international students to accept, faculty become the arbiters of who gets to immigrate. Not power we should have!! 2021-05-13 16:27:44 @JamelTheCreator Where were my Holocaust refugee grandparents were supposed to go, given that no country was accepting refugees? 2021-05-13 16:21:14 RT @moorehn: Imagine you are a woman at Apple working with Antonio Garcia Martinez. He has institutional power. He's your boss. He controls… 2021-05-13 16:21:07 RT @moorehn: Second: "Why should his personal views affect anything as long as he can do the job?" Consider: He cannot do the job. His vie… 2021-05-13 16:17:14 RT @sibinmohan: Bill Gates wants to get into Higher Ed. This article explains why this might be a disaster. @ShriramKMurthi you’re the ed… 2021-05-13 16:17:07 RT @sibinmohan: Ok please don’t do this! We already have such a system in India where a “university” only offers one type of degree, e.g. E… 2021-05-13 16:13:31 RT @mark_riedl: This is dangerous. AI false alarms are sending police into black and brown neighborhoods primed to expected armed confronta… 2021-05-13 16:09:12 @JediTigger @Yair_Rosenberg Nobody "is a race," race comes from being perceived as a race, which for some groups like Jews is only sometimes true 2021-05-13 16:00:00 RT @mboehme_: "Thrilled to announce.." -- how to welcome your new faculty? A few notes from someone who joined a research group with oth… 2021-05-13 15:48:24 @BresockCalculus Long periods of stressful situations actually can do physical damage to your brain that are, thankfully, reversible through treatment. In other words, that's normal and I hope you feel better soon. Hang in there 2021-05-13 15:31:04 @annolik Lmao 2021-05-13 15:30:56 RT @annolik: однажды в школе я ехала в метро и читала книжку с текстами песен группы сплин а мужик сидевший рядом сказал что приятно удивле… 2021-05-13 15:19:31 @roeschinc Omg hope you are ok!!! 2021-05-13 15:15:34 RT @roopshasamanta: My cousin in India is in critical condition with COVID-19. We need Tocilizumab 400 MG IV urgently in Faridabad, Haryan… 2021-05-13 15:15:03 @zhandayeva During a pandemic, at that! Congrats 2021-05-13 15:14:22 RT @moorehn: I have seen the "woke mob" discourse around a man who literally wrote a book saying repeatedly how women are conniving and use… 2021-05-13 07:24:24 @Natanael_L @prof1danilo Talia Ringer Ringer 2021-05-13 07:23:05 @SC_Griffith Donut start with the food puns 2021-05-13 07:18:01 @koronkebitch @ionathanch @wilbowma @rg9119 @mwillsey @ChandrakanaNaN Honored to be named after such a great research tool 2021-05-13 07:11:19 I've peaked 2021-05-13 06:29:13 @wilbowma Also should have said "ill-hyped" 2021-05-13 06:10:26 I'm so happy. My name is never in anything 2021-05-13 06:04:53 @jeanqasaur https://t.co/XGYKtTPVNZ 2021-05-13 06:04:18 RT @jeanqasaur: We the robots acknowledge @TaliaRinger's transition in life stage and have reclassified her accordingly. Congratulations… 2021-05-13 06:00:41 https://t.co/IgG7cpmuBP 2021-05-13 05:59:29 I'm finally in The Onion https://t.co/HON8wZob24 2021-05-13 05:59:15 @ionathanch @wilbowma @koronkebitch @rg9119 @mwillsey @ChandrakanaNaN Wait how is the second one Talia am I hallucinating 2021-05-13 05:58:36 @ionathanch @wilbowma @koronkebitch @rg9119 @mwillsey @ChandrakanaNaN 2021-05-13 05:56:58 @yoavgo What is it meant to do other than escalate 2021-05-13 05:54:45 @yoavgo Country with the best assassins and amazing intelligence and technology can't carry out targeted assassinations without destroying high-rises? Sounds absurd to me 2021-05-13 05:52:31 @yoavgo Ugh get rid of both of them and find someone reasonable 2021-05-13 05:51:30 @yoavgo Who thought a Jewish country was a good idea anyways? We can't agree on anything let alone how to run a country 2021-05-13 05:48:49 @yoavgo Didn't he turn on everyone and suck up to Bibi last time around or something? Equally awful 2021-05-13 05:41:18 RT @larry_author: @LauraSavino747 The other plane deployed its parachute to come down safely. https://t.co/7UFPuLy9TI 2021-05-13 05:39:51 RT @EmilyGorcenski: Buy your local aerospace engineer a drink 2021-05-13 05:34:12 RT @jeffbigham: LOL. 2021-05-13 05:24:33 @CharbelHage12 Israel has good enough intel and engineering to target leaders when they leave their apartments or whatever. They have the best team of assassins in the world IIRC 2021-05-13 05:18:48 Please Israel just form a competent government without this loser, how many tries has it been now? 2021-05-13 05:17:51 My honest analysis on Israel right now is that Bibi is looking for a way to hold on to power for a bit longer by provoking conflict, when Israel could very well carry out extremely targeted drone strikes to take out Hamas leaders rather than bomb high-rise apartments 2021-05-13 05:08:42 RT @amyjko: Check out Falx, our award-winning data viz by example system at CHI ’21. • Chenglong’s blog post: https://t.co/ZwequBPbIh • Jo… 2021-05-13 05:07:13 RT @ZJAyres: Some people think that mental health is too personal to be shared with others. That we need to put on a front to be profession… 2021-05-13 05:04:48 Neurodiversity is a form of diversity, and failure to accommodate it is a diversity failure like any other 2021-05-13 05:02:53 Everyone I know who dropped out of grad school this year or went on leave had ADHD, depression, or both. It's unjust to force them to stay home when the law isn't stopping you and there are safe ways to make it work, especially with vaccines in the US 2021-05-13 05:00:54 If your workplace isn't open this summer despite local laws allowing for it, super please consider letting your employees with ADHD and depression and so on come in because it can be the difference between productive, pleasant routine and directionless despair https://t.co/PTgmIBDaJ5 2021-05-13 04:56:37 @wilbowma @ShriramKMurthi Ugh actually I know 3 people who fit that description and it makes me so angry 2021-05-13 04:55:30 @wilbowma @ShriramKMurthi The lack of accommodations for mental illness and ADHD this year at a lot of other workplaces really irritated me, it's clear few people take it seriously, but I know people who have dropped out or gone on leave for similar reasons 2021-05-13 04:54:16 @wilbowma @ShriramKMurthi I would for sure not have a faculty job this year if I had to work from home all year. Actually was dangerously depressed when any semblance of routine and a place to work were ripped from me, and have been mostly fine ever since I got an office back 2021-05-13 04:52:52 @wilbowma @ShriramKMurthi The alternative is stewing in your own depression with nothing to look forward to, but constant reminders that you can't actually get work done at all even when you actually have motivation to try, which makes you feel worse about yourself and enforces the feedback loop 2021-05-13 04:51:52 @nbhansen @ShriramKMurthi I'm amazed that environment is "home" for some people and not the office. There is definitely not a quiet workspace in my house lol 2021-05-13 04:45:52 @wilbowma @ShriramKMurthi The ADHD/depression combo makes WFH torture 2021-05-13 04:42:50 @wilbowma @ShriramKMurthi I'm fine when it's used as a reason to provide physical space for employees who need it, not force them to use it 2021-05-13 04:42:18 @ShriramKMurthi I personally think hybrid is the future of everything though, it is the most just 2021-05-13 04:41:35 @ShriramKMurthi I still don't manage breaks without social help though, and I usually forget to eat almost all meals without social help. Shouldn't be used to stop people from WFH, but is a serious reason to provide those who benefit office space 2021-05-13 04:40:15 @ShriramKMurthi As someone with depression this quote is 100% true for me, I actually got medical permission to go into the empty office since last July 2021-05-13 04:34:17 RT @danrkports: I am absolutely certain that @schemeprincess can not only tear down a diesel engine and outshoot the average Bay Area tech… 2021-05-13 04:30:53 RT @wilbowma: @TaliaRinger I got some interesting book recommendations from the thread earlier: 1. Hello World: Being Human in the Age of… 2021-05-13 04:29:15 I'm weird and opinionated though 2021-05-13 04:27:38 @s5bug S T O 2021-05-13 04:26:35 Few shot learning is underhyped 2021-05-13 04:26:13 Neurosymbolic techniques are underhyped though 2021-05-13 04:25:19 @wilbowma I think that's where I am 2021-05-13 04:25:07 I'm not sure what my answer is FWIW, I waver between "overhyped" and "appropriately hyped" depending on the papers I read and the ideas I talk to people about. Some results are extremely impressive, some extremely worrying, some marginal improvements on bad benchmarks 2021-05-13 04:21:51 I'm honestly just curious about how my followers' opinions sway here. Is machine learning overhyped, appropriately hyped, or underhyped? 2021-05-13 03:59:52 @BayesForDays High speed rail between Urbana and Chicago would make me so happy 2021-05-13 03:51:17 "Congratulations, Talia!  It will be exciting.  You will soon discover how far it is to Chicago." - my running coach on my new job 2021-05-13 03:42:20 @marileezafari This lifestyle is meant for an outdated era when it was easy to find a partner and get married young and bring your partner with you everywhere so you only ever had to briefly be alone, if at all. But in the new era your family can be whoever you make it 2021-05-13 03:28:47 The pandemic taught me how much of a blessing it is to live with other people. Never going back 2021-05-13 03:25:09 @nikitab Hahaha fair 2021-05-13 03:22:00 @nikitab Does he still rent out rooms? Lol 2021-05-13 03:20:59 @roboticwrestler @syardi Yessss please, I'd love to hear about it and get ideas for where to look. I'm so surprised there aren't other faculty at UIUC with the same taste 2021-05-13 02:42:20 I guess I can start my own coop if I need to 2021-05-13 01:22:56 @syardi I think it's fine as long as they are not in my department. There is no power dynamic that way 2021-05-13 01:19:07 4+ people minimum, I really don't like 2-3 people houses, from experience 4+ is the right number for me. But I also feel a weird pressure to act "my age" which apparently means being lonely for no reason 2021-05-13 01:18:14 Coop life is basically about having a default family in the area to live with and share things with, and it feels like everyone who lives in one of these across the US is kind of similar in ways that make it really easy to go between them and just have more family 2021-05-13 01:17:15 One coop has an attic studio I could live in and then share dinners and so on, which seems like a nice balance. I also think it would be fine if the coop folks are students in other departments. Just don't want to make things awkward for CS students 2021-05-13 01:15:37 Any faculty live in shared housing during early years? I don't really have any desire to live alone. I really like the coop I live in right now. I'm looking at coops near UIUC but it's tricky to balance that with whether any students are in the coop and if that's awkward for them 2021-05-13 00:12:13 I need a beer with this guy https://t.co/6wEuVx92BI 2021-05-12 23:47:52 @hackedy I think it's the same where it shows just the constructors and not the lifting with the eliminators 2021-05-12 23:27:02 @PTOOP Conor are you drinking 2021-05-12 23:26:35 @PTOOP I'm just curious how to improve definitional equality without going extensional 2021-05-12 23:23:17 @PTOOP I honestly wish all of the reductions were always explicit, or at least there was a mode I could enable for that 2021-05-12 23:22:36 @PTOOP So when you go from something cute where that happened implicitly by definitional equality to something less cute where it doesn't, now things break and to normal humans it's not clear why 2021-05-12 23:21:12 @PTOOP So these categorical abstractions look pretty, but then suddenly I'm in a world where "exists unique" is something I need to show myself (eta) 2021-05-12 23:20:37 @PTOOP I always find it difficult to reason about things in type theory vs. category theory because of definitional equality. The reduction rules all seem to correspond to hidden things like "exists unique" and "commutes" 2021-05-12 23:17:10 @PTOOP Hmm, will you not end up extensional that way? I'm scared 2021-05-12 23:09:13 @PTOOP This is great, a whole bunch of things I'd heard about separately are starting to unify. I still really just want a pretty picture and I'm really bad at that, though 2021-05-12 23:08:05 @PTOOP Yes, but the implications of it in general didn't register to me at the time! 2021-05-12 23:05:09 @PTOOP The tool just replaces the eliminators, constructors, and implicit definitional proofs of "exists unique" and "the square commutes" with the explicit propositional versions. Fix the same structure and we get that those proofs stay definitional~ 2021-05-12 23:03:38 @PTOOP yesyesyes 2021-05-12 23:03:26 @PTOOP Basically for any two equivalent inductive types A and B, IIRC you can define the eliminator over B with the structure of A without bouncing back and forth to A (though you do as much work, and that's unavoidable). I've seen you do this for a few instances but it's quite general 2021-05-12 23:01:42 @PTOOP Carlo realized a while ago that what my tool is really doing is the transformation that corresponds to that, it's just that I need the "the square commutes" proof and the "exists unique" proof the be defined explicitly 2021-05-12 23:00:10 @PTOOP What is cute is that, say B is equivalent to A but has different inductive structure. Like fix A unary natural numbers, B binary natural numbers. Then think about what you get out of the initial algebra construction with the structure of A, it's cute 2021-05-12 22:56:04 @PTOOP ohhh I found a free version :) 2021-05-12 22:55:44 @PTOOP ahh it costs money haha 2021-05-12 22:52:37 @PTOOP Is there a citation for this somewhere? A lot of this is not explained very nicely in a lot of the reading I can find about initial algebras and inductive types. Somehow it's just expected that figuring out this lifting will be obvious 2021-05-12 22:51:12 @PTOOP ohh i see 2021-05-12 22:41:23 @PTOOP Like how do I define the universal predicate lifting of the functor? 2021-05-12 22:40:29 @PTOOP Ah I figured there might be two initial algebras here! What does that look like, though? 2021-05-12 22:35:06 I'm still confused by what the induction principle for the natural numbers looks like if I try to draw it using an initial algebra. Like I understand the constructors (1 + N ---[0, S]---> 2021-05-12 19:51:16 OK so it's not quite that this is the inductive structure, just if the two types have the same inductive structure then nice things about the corresponding initial algebras fall out of it by reflexivity, I need to think more I'm still confused 2021-05-12 19:45:37 Lambek's is freaking magic 2021-05-12 19:44:56 It takes me months and months to digest the things Carlo tells me lol 2021-05-12 19:44:22 @OwoTizusa I was thinking in the sense of ornaments. Basically nat has the same inductive structure as list but not as binary trees. nat and bin have different inductive structures even though they are equivalent 2021-05-12 19:42:38 And this is why the PUMPKIN Pi transformation works 2021-05-12 19:42:26 But you're basically going to stick in a proof that it's OK to do that in the middle of your eliminator which is why it will be inefficient 2021-05-12 19:41:44 This is why you can write eliminators over, say, binary numbers that are "shaped like" eliminators over unary natural numbers 2021-05-12 19:40:49 But since initial algebras are more general, you can kind of adjust any type to have the inductive structure of another equivalent type by defining the initial algebra that corresponds to the equivalence 2021-05-12 19:39:47 wowwwww so the "inductive structure" of an inductive type is legit the initial algebra that represents its semantics, no? 2021-05-12 18:54:32 @peztronic The campus is beautiful! I'll just miss that gorgeous mountain oh my gosh 2021-05-12 18:53:02 @Anne_Spencer_ Yeah I know it's possible. It looks like you're supposed to be able to contact them to get the official purple & 2021-05-12 18:51:58 @peztronic Oh I meant the mountain lol 2021-05-12 18:43:38 @Anne_Spencer_ Also any luck getting the purple & 2021-05-12 18:42:54 RT @Anne_Spencer_: @TaliaRinger I was thinking a "performed by actors" like a dramatic unsolved mysteries re-enactment and making my partne… 2021-05-12 18:42:51 @Anne_Spencer_ can we just do this for each other 2021-05-12 18:13:08 How does virtual PhD graduation work? Do we hood ourselves? 2021-05-12 17:28:50 Interestingly, letters and numbers both seem to have physical locations that correspond to their placements on the keyboard, so maybe this is from learning how to type really early? 2021-05-12 17:28:00 For me it's a point halfway between forward and the right of me, and halfway between forward and up of me, where I can sort of vaguely imagine there is a 7 2021-05-12 17:27:02 Imagine the number 7. What do you imagine? 2021-05-12 17:23:10 @rob_rix But numbers have always had physical locations around me. When I had to take an IQ test for ADHD stuff to rule out other things, they said my working memory was really good, but they tested it with numbers. I think I would have done miserably with letters 2021-05-12 17:21:54 @rob_rix I think it's a spectrum. I can sort of imagine images in my mind, but it's hard and they are often not very useful. Blurry, vague colors, only captures a small amount of the object and I have to "move" to see the rest 2021-05-12 17:20:25 @rob_rix I think so. 2016? Should be on the website. Didn't get far into categorical representations of dependent types, but still learned a lot. Good notes, too 2021-05-12 17:18:29 @Rucha_RPK @lambda_calculus It's on me lol. Or more accurately on @lambda_calculus 2021-05-12 17:18:14 @Rucha_RPK Oh man, @lambda_calculus got me a boba gift card for Kung Fu Tea so I'm down 2021-05-12 17:17:04 Which is how numbers have always felt to me, so it's rather interesting 2021-05-12 17:16:25 I still don't really "see" them when I think about proofs, but I do "feel" them. That is, I exist in a physical space, and I can feel terms and types around me and move them around and manipulate them and zoom in and operate over them and so on 2021-05-12 17:15:14 As someone who can barely think visually, staring at a million diagrams would have never been enough to just kind of get it. But learning the semantics made it instant, and now it's actually the way I naturally reason about proofs, since it's an elegant representation 2021-05-12 17:13:41 I used to not be able to look at categorical diagrams and understand what was going on. But then at OPLSS I was taught the semantics of the diagrams, and now I look at them and intuitively know what is going on. So OPLSS installed a category theory compiler in my brain 2021-05-12 17:09:02 @invictusqed In general I do badly when things aren't explained formally enough. I was also afraid of calculus until I took real analysis, which made much more sense 2021-05-12 17:08:22 @invictusqed Just bad experiences in high school. Also math is such a big area. Abstract algebra was more my speed 2021-05-12 17:05:37 This is an interesting mushroom I hadn’t seen before https://t.co/ewuCbGNXYL 2021-05-12 16:52:18 Being afraid enough of physics and statistics to sign up for CS as my "supporting sequence" for my math degree https://t.co/dFUUm44toc 2021-05-12 16:50:53 RT @heades: We are ending this semester with a great talk this week: "Mathematical Structures in Dependent Type Theory" by Assia Mahboubi.… 2021-05-12 16:45:16 So I only need to trick them into thinking I'm not an impostor until tenure, right? 2021-05-12 16:27:02 She made me tea and breakfast and we chatted for an hour and it was very nice 2021-05-12 16:26:37 For a few years I've been a friendly visitor and ESL tutor for an elderly refugee, which is basically like having an extra grandma. But I had been talking to her just over the phone for over a year. Both of us are fully vaccinated, so I just saw her for the first time since 2021-05-12 16:06:51 @peztronic Where? Send detailllsss I'm cool driving 2021-05-12 14:47:03 Haha what an email to get https://t.co/sONYCfx05I 2021-05-12 14:39:29 RT @ShriramKMurthi: Things I wasn't warned about before moving to #RhodeIsland. From @DoorsOpenRI. https://t.co/iRDoPcPxfM 2021-05-12 14:37:26 RT @typer_uma: textbf is good but what about textgf 2021-05-12 14:29:35 RT @uwcse: “We hope this study encourages others to critically examine surveillance and the risk of harm in carceral systems.” #UWAllen's @… 2021-05-12 14:27:20 RT @jonmsterling: ..... I just solved the regularity problem in cubical type theory. 2021-05-12 14:21:02 @Natanael_L @andrejbauer I can't really visualize much. It's more like existing inside of a physical space I can interact with and manipulate 2021-05-12 08:16:58 @alexhanna Well that answers my very naively optimistic remnant of hope "maybe the team asked for this as a form of repair and they listened?" 2021-05-12 07:52:42 RT @jix_: @TaliaRinger https://t.co/mNrLgyKYUp 2021-05-12 07:22:49 @ChrSzegedy @yoavgo @giffmana @theshawwn Maybe you're just smart :) 2021-05-12 07:21:49 "Is TensorFlow a python" lmao Google I'm losing it https://t.co/KozA79n5RN 2021-05-12 07:19:00 @ChrSzegedy @yoavgo @giffmana @theshawwn As an ML outsider this looks like I imagine type theory papers look