Message: One footnote on the J& The analysis for deaths and hospitalizations (H) was post-hoc (not pre-planned) and the prevention was 100% ≥28 days post-vaccine. Prior to that there were some H events as shown in the Table. No deaths in the vaccine arm (7 in placebo) https://t.co/aRmDcpFg5R
Message: @juliagalef @mindspillage It was very special. We went off the beaten path too - didn't see another human. Highly recommended! (6 days, IIRC, to be pedantic.)
Message: Bell's palsy, associated w/ vaccines (not just vs. covid), is rare, usually self-resolves "The observed incidence ... in the vaccine arms is between 3·5-times and 7-times higher than would be expected in the general population" https://t.co/IUPrDxuZvt Why? We don't know https://t.co/KBsLEXZCgF
Message: @mindspillage @juliagalef "noticeably happier and more creative" is very inadequate here. It's the problem with reducing to language something whose main benefits seem inarticulable to me.
Message: @mindspillage @juliagalef Many other experiences like this where immersion seems to me the key element. I've been to the bottom of the Grand Canyon a half dozen times - very similar. Indeed, I spent a week down there once, and that was different again. Ditto many other experiences.
Message: @mindspillage @juliagalef Yes. Eg: I sometimes (pre-pandemic) would spend a day working in an art museum, simply for the immersion. I noticed decades ago that simply _not thinking_ while in a gallery would make me noticeably happier and more creative for days or weeks after.
Message: @juliagalef As for point 2, I won't argue, but will note that (unsurprisingly) I have a very different opinion. Not necessarily of the "in principle" question. But as a practical matter many of the most important things are very hard to measure.
Message: @juliagalef Someone once told me "travel is the only education". It's too strong... but has a kernel of truth. [I thought it was @tylercowen, but asked him, and he said no.]
Message: @juliagalef (An amusing logical feature of this line of argument: in some sense it's making the case that the weaker the argument is, the stronger it is.)
Message: @juliagalef If it makes you feel any better: I think your original tweet is very wrong Though the most important concrete things you get from travel are inarticulable, and so hard to argue. It's very much a "what's it like to be a bat?" situation.
Message: @michael_guihot @chrmanning Thanks for sharing the article, I will check it out! I grew up on the Gulf Coast of Texas in a town with higher humidity &
Message: "For, if you keep silent in this crisis, relief and deliverance will come to the Jews from another quarter, while you and your father’s house will perish. And who knows, perhaps you have attained to royal position for just such a crisis.” I am using it often on high profile 3/n
Message: Friends. Tomorrow night, Feb. 25, we celebrate the Jewish holiday of Purim. Throughout my up bringing, Purim was a holiday of fun, joy, clowns and costumes, sort of Jewish Halloween, celebrating our deliverance from a genocidal plan in Persia, 580 BCE https://t.co/bjIi8XF5Xn 1/n
Message: Thread with detailed, informed, important information about how @Google fired @timnitGebru and @mmitchell_ai and what it means: https://t.co/rRjWWyEjmz
Message: For anyone whose reaction on reading this was "Well, I don't have enough information..." I've written a thread for you! (linked in next tweet) https://t.co/WcSVLxUuqj
Message: And in their open letter (the same one @le_science4all talks about in the videos above), which lays out what Google could have done (rather than doubling down on terrible decisions): https://t.co/wDiZsPBa6L
Message: @timnitGebru Likewise this interview by @kharijohnson where @timnitGebru lays out the inaccuracies (and *harms*) in Google's statements about the situation: https://t.co/u1STh7u4iS
Message: @timnitGebru @mmitchell_ai @Google Other valuable sources include @le_science4all 's videos (in French and English) French: https://t.co/6yzLJCwjlm >
Message: If you follow @timnitGebru and @mmitchell_ai, it may feel like everyone must know the story about how @Google fired them (& https://t.co/9rTMylGvUw >
Message: @ctongfei oh i have a lot to say about linguistics as well (c command? little v? v shells? ecm? theta roles? control?...) i just don't care as much. the FP stuff is at least useful.
Message: Encouraging data for suppression of asymptomatic infections (avoidance of carrier state), proxy as explained in legend, 90% after 2 doses https://t.co/eurpZeYeiz
Message: The Israel data from 597,000 vaccinated matched w/ same N of unvaccinated individuals is published @NEJM today, providing extraordinary real world evidence of preventing infections, hospitalizations and deaths https://t.co/J7EAeINxW2https://t.co/c8JA1x2WLh
Message: @zacharylipton So true... what is it about the interface that makes it so error prone compared to all the other messaging services? That pull down and intermingled messages I guess.
Message: Summary of the 5 Phase 3 vaccine trials. One theme: 100% protection from death and hospitalizations That is impressive! https://t.co/wAa62zcDh6
Message: @michael_guihot @math_rachel In particular, I was born in Bundaberg — which has the odd property that every Australian knows about it, but almost no one outside Australia. I owe my life to Bundaberg Hospital. So definitely read that article!
Message: @nbonacchi @drdevangm @gpereyrairujo @nicolas_guyon We are also likely to run open calls for editors in the near future, with a particular focus on geographical locations where we are underrepresented.
Message: hi PL people. I think one of the obstacles to ppl learning FP patterns is that the naming is opaque af. I am not talking about car/cdr. Curry, monad, pure, bind, left, right... nothing really makes sense. Its not trivial to come up with alternatives, though. So my Q: was it done?
Message: @nbonacchi @drdevangm @gpereyrairujo @nicolas_guyon It is really hard to get proper diversity. If you want people from Iran, Chile, Ghana etc. then you quickly get beyond people's sphere of knowledge. Check out the people on our eCAG: https://t.co/W4WOqB9rw3 It is really valuable for the journal to hear these perspectives.
Message: Idom looks interesting - it's basically a replacement for ipywidgets, which provides access to the full React ecosystem. Would be interested to hear experiences from anyone that's tried it. https://t.co/75Jjx2Y1Ma
Message: We're losing momentum on our vaccinations, last week weather but this week it's supply issues We need to go bigger and not just on vaccines https://t.co/W2EbxYBWVM @abettel @dariustahir rapid home tests now to all households scale genomic, digital, wastewater surveillance https://t.co/9fbINGdgys
Message: @drdevangm @nbonacchi @gpereyrairujo @nicolas_guyon Thanks Devang - @nbonacchi yes this role is compensated (although like all elife roles, not as much as we would like). If we invite we get people who all look the same because we have to know them to invite them. Open calls are really powerful for getting a diverse group.
Message: Very interesting! Because I have not met even a single convert among Bayesians. My theory: They fall in love with the power of probability calculus to capture prior scientific information and they refuse to see how this calculus excludes them from capturing most of science. https://t.co/ezUnYJPCKO
Message: These sorts of debates are plagued by a lack of agreed definitions and host of assumptions that are influenced by one's cultural background. Now, where did I put that 10 foot pole...? https://t.co/aKwr3ula7c
Message: Students of causality will find challenge in this variant: Z | v X----> Is it good or bad to control for Z to estimate P(y|do(x))? Z seems to be an "effect modifier" b/c it "clamps down" M, which is a mediator between X and Y. But is it? https://t.co/7CqcPrZVE1
Message: @PhilippBayer @drjohnflackett @_AiLab @math_rachel @PerthMLGroup We'd love to come some day, but might be a while. Would you be open to a virtual talk?
Message: @memosisland @GaryMarcus I would have liked to join your Colloquium, to refine (ie, falsify) my own theory of what it means to "understand" something: https://t.co/GCeYu5nEol. But I don't do facebook
Message: @Kelly_Clowers @rcalo Yes! Sundials in the Southern hemisphere go the other way! But sundials here in Seattle (when they work) go (what we already call) clockwise.
Message: Why do we see superspreader events from #SARSCoV2 far more often than influenza? https://t.co/B9Z33Xbq5J @eLife @fredhutch "SARS-CoV-2 superspreader events occur when an infected person is shedding at a very high viral load and has a high number of exposed contacts." https://t.co/N01PLqQdDm
Message: @trashcanmagic I feel you on this and sorry about these barriers. If I had literally any other interest outside of academia, I am sure I would have made the same decision.
Message: @mjcavaretta @larsjuhljensen @hmason The same model has to write ad copy, run AI Dungeon, and who knows what else people are thinking. The prompt determines what the application is, so it is vitally important to get the prompt right.
Message: @joe_hellerstein Yes we definitely have it easier than other fields on average. (Partially because our avg is rigged...) I do think tho cross department comparisons can at times be a trap. No one should be this underpaid regardless of what they did before and what they can do after.
Message: @joe_hellerstein In any case, CS departments tend to have more power within universities. This isn't always a good thing, but in cases like this I think we may be able to push to improve things for all departments and I hope we will!
Message: @joe_hellerstein Depends on which students, honestly. "Borrowing from the future" is not really an option for many people. A lot of people end up forgoing CS grad school b/c they can't afford to be so underpaid for so long. Even things like when the 1st month check arrives makes a big difference.
Message: Best the US has looked in many months. But far from containment, as defined by ≤10/100,000 new cases/day Only Hawaii, Puerto Rico, Oregon and Maine have achieved this goal so far @CovidActNow Best way to be ready for B.1.1.7: max vaccination + getting cases low as possible https://t.co/LDXzfGkVfT
Message: Grad students and contingent academics are severely underpaid. Our current system only works for people with few/no financial responsibilities and safety nets. It's only this year that I didn't need to be constantly worrying about how to make ends meet and let me tell you...
Message: I was surprised to hear that Andrew Gelman agrees with me on this daring statement (we normally agree on generalities, rarely on substance). I checked the blog, and I don't see any agreement. Bayesians find it harder to understand causality than frequentists. https://t.co/bKKaA2BsHx
Message: @roydanroy In 2000 people were held back by running Matlab on Windows. Try autodiff on Matlab.... Hard to imagine what you had to go through before Linux, Python and things like PyTorch and NumPy. You had to write your own (which some of us did).
Message: @dan_s_becker You got this wrong. Links to news articles on FB take you to the publisher's website. Publishers can then get revenue from ads or a paywall. There is no content on FB, except for the title of the article.
Message: @marianeo23 FB has called for government regulations to define what constitutes acceptable online content. In the absence of such regulation, it had to come up with its own content policies. The policy against hate speech is very clear. https://t.co/v55f8lfUvP
Message: @sethvneel @ben_golub To be fair, this was in part because I managed to navigate the department by avoiding professors who told students who started math in college that "the world does not need more mediocre mathematicians."
Message: @sethvneel @ben_golub This isn't entirely true. I started in Math 23 and didn't do grad classes until my junior year and got competitive math PhD offers. Some managed this starting grad classes later than I did. I worry that there's a self-fulfilling prophecy in how we're discussing this issue.
Message: @kanarinka @laurenfklein But notwithstanding all of that, great talk! I've heard snippets of your ideas in different places, and have had the book on queue to read for a while now, and this was a great way to get into the material. I was compelled by the emotional dimension of data presentation.
Message: Will vaccines block #SARSCoV2 carriers? An experimental model of human neutralizing antibodies demonstrates potential of live virus in nasal mucosa (turbinates) https://t.co/hXPfG4x1gg @cellhostmicrobe https://t.co/qVzU1KWKNO
Message: Thank you @cephaloponderer for this insight into their and your amazing work. I definitely experienced the value of this when working with you and other members of the team but only saw a fraction of it. https://t.co/uLoQefmAFS
Message: @roydanroy @ylecun Research is generally easier in the early phases of a field. Many famous contributions are the result of being there first rather than of incredible brilliance.
Message: @brody_lab @MSchottdorf @enieh36 This title is an inspiring juxtaposition of universality and specificity. I can’t wait to discover what it means.
Message: @JaniceWyattRoss @KendraWrites @heymamore @chetanhere Consider sharing "AI, Ain't I A Woman?" with your daughter so she knows she is in similar company of Black women dealing with algorithmic failures: @Oprah @serenawilliams @MichelleObama https://t.co/XxwSW5xmpv
Message: @roydanroy Or recent COLT papers ... Just kidding. Seriously though, trying things out was a lot harder to do in the 1980s and may have to involve different people from the authors.
Message: Why vaccines are so difficult to compare @NatureNews @heidiledford https://t.co/kgGVh12Xca Timely w/new J& — Single dose vs all others 2-doses —Tested in So Africa & —All vaccines marked reduction deaths &
Message: @2rkiva @egrefen @_rockt @riedelcastro FB has voluntarily applied the GDPR rules worldwide. But who is to say that facts should invalidate your assumptions of nefarious intent.
Message: Irony of Ironies: @kanarinka and @laurenfklein are asking "data science for whom" in their talk, and the automatic closed captioning keeps converting it to "data science for him". https://t.co/77eCMzvQK3
Message: Should we prioritize sequencing #SARSCoV2 in children with covid? https://t.co/Rvp4ZG7mFh by @arianaeunjung The N679S variant mentioned in the piece is unproven for having any meaningful differences compared with D614G (ancestral strain) https://t.co/yCeKoD5dcS
Message: @marianeo23 1. I'm French and American. 2. AI is used by FB to *prevent* the platform from being weaponized (eg by filtering hateful and violent speech). 3. Getting your facts right is cheap, but it may cause you to have to revise your preconceptions, which few people are willing to do.
Message: @mioana @arpitrage @annastansbury Agreed. See also this "Digital Capital" paper with @danielrock and @tambe where we find larger market values (Tobin's q) for firms with workers who have more IT and AI skills. https://t.co/aj28lYXxoT
Message: @mioana @KordingLab Yes, firms do seem to have a lot of monopsony power over more educated workers, including those with AI skills. Here's some evidence in our "Digital Capital" paper with @danielrock and @tambe. Another one is on the way more specifically on this question. https://t.co/aj28lYXxoT
Message: @DRedecopp @timnitGebru A good union contract that understands the info silos in tech, helps break them down and ensures trade secrecy claims can't be used to harm workers disclosing in the public interest, certainly would help.
Message: @IndigenousAI ugh. not surprising, though. I went to one in Palo Alto when I moved to the Bay a few years ago and it was already on its last legs
Message: I will not prove anything to you with all due respect and I certainly will not stop continuing to speak up about a corporation which treats people like this with impunity until we have real change & Feel free to move on about your day without coming to my tweets https://t.co/l5TOEQjY05
Message: What a remarkable trans-disciplinary academic team @Yale @YaleMed @YaleSPH @YaleEMD @YaleIBIO w/ @VirusesImmunity @SaadOmer3 @awyllie13 @thehowie @jordan_peccia and all their colleagues, covering all the bases for the pandemic
Message: My favorite @techreview issue each year is their top 10 breakthrough technologies. This year, besides mRNA, several #AI related picks, energy, hyper-GPS https://t.co/jSjLDjhSLshttps://t.co/TWfdbqMPBr
Message: @RonWechsler @DKThomp Given that we are mostly vaccinating older people who have few contacts and medical people who take strong precautions, I wouldn’t expect much impact yet
Message: @math_rachel Well, Sydney and Melbourne are not only much bigger and more culturally diverse in general, but they have more tech – but I was born in Queensland, so I should stick up for the place!
Message: While so many are still busy digging for some malice behind Israel's victory over the Covid-19 vaccine, it is refreshing to read @ShMMor sober account of the factors that led to that victory. https://t.co/s69ocqWaEK
Message: @TracyBethHoeg yes, 100% agree. I might spend a bit of time trying to work these probabilities out today, because I think with the current community infection rates, the odds of catching coronavirus from a vaccinated person are going to be lower than the odds of a traffic accident.
Message: @WillOremus @davegillis @mtaibbi The speed and strength with which things flip from "prescient" to "obvious" in a blink, and then how most everyone forgets and denies the before, is probably my strongest pandemic conclusion and surprise. I knew it happened, of course, but wow... So dunno.
Message: @TracyBethHoeg Shouldn't the odds be quite a bit lower? There is a baseline probability that an unvaccinated person is infected at any given time, and this probability is quite small. Then you need to multiply *this* probability by 0.01?
Message: Best article I’ve read on the pandemic in some time: A thoughtful investigation into global disparities and a meditation on epistemic humility by @DrSidMukherjee https://t.co/Y55iwaRvrL
Message: @kendmil @behrenstimb @BakermansJJW @SaraASolla @worldwideneuro It's high dimensional, depends which dimension you consider altitude. But there must be a manifold answer. https://t.co/GSi2wXu43P
Message: this isn't over until i decide to go to a concert at 5pm on the day of the concert and pay for tickets with some cash i already had in my wallet. nothing less.
Message: i miss spending like $100 going to a dozen concerts in a year, and half of them were awful and i spent twice as much on alcohol, and all of that was awful i miss concerts that were garbage, when going was inconsequential. now we plan our groceries to minimize trips and exposure. https://t.co/iZ9YeMzelx
Message: Each day now we're seeing powerful real world evidence of vaccine protection from deaths and hospitalizations, from Israel, from the UK, and here from the US among nursing hime residents (and more from US coming soon) https://t.co/sE9d0HZxPJ @KHNews https://t.co/Dxre25n1ok
Message: @itsafronomics @femme_economics @drlisadcook I forgot the best part -- I sent him an email after saying that I'd recommend not going around asking about parental occupation at academic events since it may other 1st gen students. He sent me a long email explaining his research in this area (with slides!)
Message: @neuroecology I miss the oxygen station!! Miss sitting with an oxygen mask before my talk, with rapid heart rate after altitude-induced insomnia, as everyone passes by with worried eyes... But also miss talking about science fiction with @janexwang at said oxygen station :)
Message: The mentality of some writers (@glcarlstrom) has reached the point where they can no longer see the moral deformity of their writings. https://t.co/w8Wb6z1LXu
Message: @GovHowardDean Compare the numbers to the placebo group. (Yes confidence intervals are large for low-frequency in both arms events). Also I didn't put in the vaccine efficacy numbers which are statistically stronger and that's why we chose those endpoints. It's good &
Message: Sorry, *seven* medical interventions in placebo group, not 14. Looked at the wrong column. But the vaccinated group number is correct: ZERO. Zero deaths, zero hospitalizations, zero medical interventions after 28 days after one dose of this easy-to-store vaccine.
Message: @GovHowardDean A few people getting mild COVID—essentially a cold—even after vaccination is to be expected—we do not have a single vaccine that doesn't occasionally suffer from breakthrough cases. However, when we can reduce the disease burden globally to another cold, the pandemic is over.
Message: I know the citations in Google Scholar are not always trustworthy, but it's weird to me that there is such a huge difference between citations to Higgs discovery papers by @ATLASexperiment and @CMSExperiment. The citations in @inspirehep are much more similar (and accurate) https://t.co/FHql8WwIqO
Message: 6. More on deaths or hospitalizations (post hoc analysis) All 7 covid deaths in placebo arm, but hospitalizations did occur in vaccine arm before 28 days, thereafter were all prevented https://t.co/AUufx7WvYX
Message: I’m giving a talk from my home today Which is why a construction crew just showed up outside my window to start digging up the street to do utility work Fucking pandemic
Message: La semaine dernière j’ai conseillé a les followers de vendre les bitcoins Je suis bien sûr incapable de prédire l’avenir du bitcoin mais il était évident qu’il y avait une bulle à court terme... https://t.co/rPDRiPc4nf
Message: @aRahasya It's randomized and double-blind. They couldn't have cherry-picked because nobody knew who was getting what! That's the beauty and the magic of these trials.
Message: Yes. I have a few other pieces I'm wrapping this morning, but the efficacy numbers, while not irrelevant, aren't the most important number and are often misunderstood. Will explain longer soon. Look at disease burden, not just practical trial endpoints. https://t.co/zfkKzlFOyi
Message: @DavidHarrisAJC And it will continue to rise, and rise, and rise... until we learn to spell the word Zionophobia. See why: https://t.co/iWqdjNV4T3 See how: https://t.co/7jn0pnLspP
Message: @emilianoucl @tforcworc Oh, I'd say people definitely do complain, and many have noted the increased dependence of university CS programs on these cos. Workers at all of those companies have also organized and drawn attn to these issues
Message: Super-congratulations to @analisereal on his joining the University of Washington, and to causality researchers in the State of Washington on this terrific re-enforcement, which future breakthroughs will affirm. https://t.co/d8oVAsVjio
Message: 3. A look at asymptomatic infections (carrier state) suggests the efficacy for reducing that is similar to overall efficacy https://t.co/z8FynFIPoM
Message: The @JNJNews vaccine FDA briefing documents are out, so we now can see how well this single adenovirus vector vaccine dose works https://t.co/StA97OiJ6O 1. It is efficacious, overall 66%, but not as high as the mRNA vaccines (even in US with D614G strain). https://t.co/o5vGWfSbGI
Message: @SSarabipour @Merz @AHollandLab @iaincheeseman @neuralreckoning @AnnaHscientist @DORAssessment I'm not saying that we shouldnt work on it, I just don't see how OPR solves the problems. But then again I also wasn't very optimistic about virtual conferences until we were forced to have them
Message: Gérard Colomb ancien ministre de l’intérieur explique la décision des écologistes de supprimer la viande dans les cantines lyonnaises La part des familles musulmanes est tellement importante à Lyon que la viande n’est acceptée que si elle est halal... Via @fmomboisse https://t.co/xUOBXMWs2o
Message: Ce qui est INVRAISEMBLABLE c’est que le Maroc vaccine 2 fois plus vite que la France Le Chili 3 fois plus et la Serbie 5 fois plus Sans parler d’Israel... Franchement @EmmanuelMacron devrait avoir HONTE ! https://t.co/i7ER9mfFbm
Message: @avnerarik I am not talking about "a pace workshop", but a clear commitment of the US State Department to eventual peace, entailing warnings to the ayatollahs an the PA that their textbooks -- the greatest obstacle to eventual peace -- ARE being monitored.
Message: @pfau @SaadJbabdi @KordingLab @BakermansJJW @SaraASolla @worldwideneuro Shall we not have this argument about semantics. Linear decompositions are fun and can be revealing, but if you want to use them you need to understand some maths. This code is trying to help people do that.
Message: Pourquoi Israël a gagné la guerre vaccinale Et pourquoi @EmmanuelMacron l’a perdue Israel a vacciné 18 FOIS PLUS que la France et sort du confinement depuis avant-hier A la présidentielle cela va peser lourd ! https://t.co/7SjfBGFieY
Message: @PetarV_93 @ipam_ucla @wouter_kool @thserra It was a great talk, unfortunately I had to leave in the last few min right as you were introducing algorithmic reasoning — so I’m looking forward to watching the end. I’m curious if there are some connections to what we did with graphs and symbolic models https://t.co/FNxy96zpPX
Message: If you want to learn linear manifold methods, or just brush up on linear algebra, @BakermansJJW has made a matlab tutorial to accompany @SaraASolla's @worldwideneuro talk. Talk: https://t.co/zP710jYKiW Code: https://t.co/NkVHUqUffU Featuring Cats and Camels in latent spaces. https://t.co/Oe2OxOKyo0
Message: @teslaownersSV @cleantechnica We’re upgrading all NNs to surround video, using subnets on focal areas (vs equal compute on all uncropped pixels) & This is evolving into solving a big part of physical world AI.
Message: @jeremyphoward @fastdotai Congratulations on the move. I wish you and Rachel the best in your new home though I’m sad that it will be harder for me to see you in person.
Message: I remember getting some of our first investment dollars for Rational Machines, then driving over to Frys to buy parts for the R1000. We celebrated by having lunch at Burger King. Tis the end of an era... https://t.co/ch7P2isRBx
Message: @FemInt @JohnMashey @DrRimmer @QUT @lindasob @InnovationSarah @feraldata @chriseigeland @EliseInTheWoods @terry_carney Thanks, yes it would be great to chat in a few weeks once we are settled!
Message: @EnricoCoiera @fastdotai That would be great :) I'll be in Australia. I'm very keen to work with Aussie medical researchers if I can be useful...
Message: My first interview since I decided to return to Australia. I discuss organizational AI strategy, AI accessibility, the future of deep learning, and much more. Thanks @drjohnflackett of @_AiLab (Adelaide, Australia) for being such a great interviewer! https://t.co/zbyPgqGDeu
Message: I'm really excited to say that I'm sending this tweet from Australia - our new home! I learned lots in SF and had a great experience, but we've decided that long-term the US isn't really the right place for us to raise a family. @fastdotai will keep going, with an Aussie base
Message: I’m sad to leave my dream job as Director of the Center for Applied Data Ethics at USF. I will be joining the USF Data Institute Advisory Board to stay involved. 5/
Message: The last year in the USA has been horrifying. I’m lucky in so many ways: we were able to isolate pretty strictly as a family of 3 (a privilege that many did not have) &
Message: @mcwm a burning man for every occasion - children's birthday parties, coming of age ceremonies, weddings, baby showers... the opportunities are endless!
Message: I do not know myself, because Twitter refuses to "verify" me, and my university says they never dealt with Twitter. It seems I will forever remain "unverified" -- a wandering soul in search of identity. Help anyone? https://t.co/ZTH58aPRoJ
Message: @suzan Origin of IOB tagging: Ramshaw and Marcus (1995). But I doubt that it is a good first thing for the person to read.... https://t.co/7BHGC9qt7q
Message: I am not a candidate for promotion but if asked to fill this form I think I would pass in flying colors citing my record of defending and promoting the inclusion the most excluded and intimidated minority group on campus: Zionist students. The first professor to try it: Hat off! https://t.co/sUtawHt0yy
Message: @itsafronomics @femme_economics @drlisadcook Storytime: I was at an econ workshop as a grad student this one time and this econ grad student was going around asking people if their parents have a PhD. When he got to me (and I said they don't), he asked "but how did you get to Harvard?" He is an inequality researcher :)
Message: @Merz @SSarabipour @AHollandLab @iaincheeseman @neuralreckoning @AnnaHscientist @DORAssessment Which is not to say that there is no better solution, I just haven't seen evidence of good solutions lurking around the corner (unlike online conferences which have obvious advantages to a wide range of problems)
Message: @Merz @SSarabipour @AHollandLab @iaincheeseman @neuralreckoning @AnnaHscientist @DORAssessment I'm more with Alex when it comes to peer review. I don't think that open PR or post-pub-Peer-review are effective solutions to the big problems in our literature. Pre-prints are great, but I don't think we can develop a post-pub review system. The incentives just aren't there
Message: in the aftermath of google firing two *prominent* AI ethics researchers after publishing a paper that implicated google's business, all i can think is how much pressure a student only a year or 2 into their program will feel if their work undermines facebook's business interests https://t.co/vRVREjQ0pa
Message: @geomblog @kate_saenko_ @mmitchell_ai There are a lot of joint appointments and people who are X% at google X% in the university etc who I see are silent. I have seen very few people saying anything in fields like computer vision (eg).
Message: @insanedreamer @erikbryn @ESYudkowsky @chrmanning @FutureJurvetson @tegmark If GPT-3 reads The Matrix script and writes a sequel, I hope it names it properly. The sequel to The Matrix should be called The Tensor.
Message: @geomblog @kate_saenko_ @mmitchell_ai There are a lot of prominent voices very conspicuously silent. For instance The Stanford daily had to run an article noting that members of the AI lab were so quiet before the Stanford AI lab had exactly ONE tweet saying “we support our alumni Timnit Gebru”.
Message: @kate_saenko_ @timnitGebru @mmitchell_ai Not sure what you mean. Many of us (maybe not at top universities but heavily involved with fairness work) have been speaking out. And google funding is miniscule compared to government funding.
Message: @chrislhayes I am not at all against timely panic if warranted! Not at all. But come on, let’s see the preprint before getting this scared. So far, no variant has knocked out all the vaccines.* (We have uncertainty over AZ and one variant—but it’s uncertainty. Tiny sample, CI crossing zero).
Message: @guillermokrh @timnitGebru @Jersey_Hair @kate_saenko_ @mer__edith @mmitchell_ai I fear the company has lost its way and its mission. Fortunately there is still an institution devoted to organizing the world’s knowledge and teaching it to the next generation: Universities! (And their libraries)
Message: @chrislhayes On the one hand, yes yes of course, let’s vaccinate as fast as possible globally. Best protection. OTOH there isn’t an endless fitness landscape here. Look at what happened in SA. We had one that did seem to have immune escape: it went way up and down even without any vaccines.
Message: @chrislhayes The way these things get reported—highlighting drops in neutralizing antibodies which don’t really correspond to similar drops in vaccine effectiveness—has been truly confusing. I do not see a doom scenario from just this article, just one more reason to keep vaccinating fast.
Message: @chrislhayes I found this to be premature. First, where is the paper? Second, let’s accept it’s more transmissible. They’re reporting it’s already dominant. So that would explain what *already* happened in CA. Key question is vaccines. It doesn’t sound like a big threat. (Where’s the paper?)
