Aaron Bergman's tweets, December 2025
Every public tweet, reply, quote tweet and retweet that Aaron Bergman (@AaronBergman18 on X, formerly Twitter) posted in December 2025 and has not deleted: 650 posts, oldest first, times in UTC. Each heading links to the original post on X. When Aaron replies to or quotes someone, their tweet appears as a blockquote with their name; those words are theirs, and everything outside the blockquotes is Aaron's.
This page is one month of the complete archive of Aaron Bergman's tweets, which updates automatically every few hours. Previous month: November 2025. Next month: January 2026.
Monday, December 1, 2025
Reply to @rajlearns, 00:32 UTC · 10 likes
Raj Thimmiah (@rajlearns): Need more quadrant four
I reposted this bc I like it and endorse most of it and more importantly agree with ur caption, but I will say that I don’t like the “meaning is found here” label
The *feeling* of meaning is found there. Social creation is neither necessary nor sufficient for actual
Thread, continued, 00:34 UTC · 2 likes
Meaning
Reply to @jeremyphoward, 02:28 UTC · 11 likes
Jeremy Howard (@jeremyphoward): EA is clearly sociopathic.
If you agree with it, you might be sociopathic too. This basic principle has been much discussed for some centuries FWIW. 🤷
Sounds like you don’t have a problem with the concept, you have a problem with the current implementation/institutions
You write an EA Forum post explaining why they diverge!
Reply to @metachirality, 02:46 UTC · 4 likes
metachirality (@metachirality): do you have a source on effective altruists being sociopathic
Lol I mean even I’ve personally met one (based on vibe+presentation+social circles) I’d describe that way
Probably any large group has a couple bc base rates, EA could be at special risk insofar as it’s access to money and power, although this was more of an issue during FTX days
But in general? No lol obviously come on
Tweet, 04:39 UTC · 21 likes
Despite major health issues I think I have it better than almost all wild animals
Reply to @gpu_thief, 04:56 UTC · 3 likes
Alex Lags Ever Xanadu (@gpu_thief): This is an awfully strong statement given that EA as usually construed requires taking positions on morality that, if not fringe, are at least minor among moral philosophers! (A strong form of utilitarianism combined with hedonism)
I think it’s kinda fuzzy/unclear how much hedonic utilitarianism EA requires but it’s definitely less than 100%
Like most ethicists afaik also give *some* weight to suffering and *some* consideration of magnitudes - that’s all you need to make EA a project worth doing until you’re at a substantial fraction of either personal or world resources being dedicated to it
Maybe I’d agree with you in a world where the norms were such that you had to be optimizing extremely hard (eg donating all money above poverty line) to be considered “EA” in some consensus sense, but that’s not the world we live in
And not many ethical systems *condemn* figuring out how to best and most effectively reduce suffering unless/until you’re doing things they consider actively harmful on some other grounds right?
Reply to @hopes_revenge, 05:06 UTC · 12 likes
hope hopes hoping (@hopes_revenge):
Unironically what’s your time horizon for this?
Not that anyone should listen to me, I’m pretty sure I’ve never had $100k at a single point in time
Reply to @the_engi_nerd, 15:00 UTC · 12 likes
HIGH PLANES Drifter (@the_engi_nerd): Some women out there are so gorgeous that I can’t imagine living with them. She looks like a supermodel? Great. The first time you fart in front of her, you have to kill yourself.
Sufficiently conventionally attractive women are downright scary to the point that I’m not sure I should even describe myself as properly attracted to them
Reply to @the_engi_nerd, 15:03 UTC · 9 likes
HIGH PLANES Drifter (@the_engi_nerd): @AaronBergman18 Like what do you even do, put them on a shelf and admire them?
Run away
Reply to @jeremyphoward, 15:48 UTC · 4 likes
Jeremy Howard (@jeremyphoward): But they didn't calculate that instead they should save prawns, or let brown people die because they're not productive, or decide to create a p=1e-10 chance to creating 1e58 high-qualia galactic beings.
Come on man based on bio you're a smart guy, "let brown people die bc not productive" is like precisely the opposite of EA, and betrays some sort of fundamental misunderstanding
Want to debate shrimp or specific longtermism papers by a handful of academics object level, fine, but what you're describing (shrimp + esoteric longtermism) represents a few million dollars (and probably proportional person-hours) in a movement that has given away like $4B. Don't get negatively polarized against something you don't understand
(Btw it would really own the longtermists if you gave a bunch of money to GiveDirectly)
Chart from November 2023
Tweet, 21:31 UTC · 4 likes
Thing I just signed up for (Monarch spending tracker/financials manager)
I've only been using it for a few minutes but at the very least having all different accounts in one place is nice
Referral link below
Quote tweet, 21:48 UTC · 177 likes
"End of the world" is a funny term because it's not obvious what it's supposed to mean even when used ~literally:
- society collapses
- no human remains alive
- the earth doesn't support animal life
- the earth doesn't support any life
- the earth is physically destroyed
- the universe no longer supports life
- the universe no longer exists in some sense (closest to literal)
Quoting yung macro 宏观年少传奇 (@apralky): ok then!
Thread, continued, 21:48 UTC · 32 likes
It's not especially helpful to respond to a grave concern with "well actually the term you're using is reminiscent of an outdated, fundamentally religious metaphysics" but "end of the world" seems reminiscent of an outdated, fundamentally religious metaphysics
Thread, continued, 21:48 UTC · 12 likes
Like it makes vibe-sense in a 'God-ruled, only humans ensouled, geocentric cosmos' worldview
Of course this isn't an argument against any of the more specific meanings coming to pass
Thread, continued, 21:48 UTC · 12 likes
Tbc I think it's basically fine to use for anything stronger than "society collapses," just an observation
Tuesday, December 2, 2025
Reply to @finiteopencover, 02:59 UTC · 3 likes
compact set 🎀 (@finiteopencover): my laptop has started crashing out over running a single google chrome tab bc ive had it for 3+ yrs & right about now im starting to wonder whether ive sexualized myself enough on here for one of yall to buy me a macbook
Shoot ur shot queen
Reply to @DIYDisclosure, 04:04 UTC · 1 like
Mellow (@DIYDisclosure): Where does matter touch consciousness?
Idk
Tweet, 06:16 UTC · 14 likes
I should probably make a plan for being able to keep my phone alive indefinitely (for eg in a disaster or just any long outage)
Idk why I started worrying about this recently but I’m glad I did. Winter vibes I guess
Thread, continued, 06:16 UTC · 5 likes
Indefinitely means solar, I think (takes on this/alternatives welcome ofc)
I was assuming that a propane generator would functionally be infinity energy with a couple tanks but this was fake news, apparently you’d need like $500 worth to keep a laptop running for 6 months
Reply to @rgblong, 06:17 UTC · 2 likes
Robert Long (@rgblong): niche bit of consciousness lore: this comment has been appearing on articles and videos related to consciousness for ... a decade? hundreds of times, with the exact same wording
What
Reply to @captgouda24, 06:31 UTC · 24 likes
Nicholas Decker (@captgouda24): An advantage of going into Sweetgreen’s in DC is that sometimes you can pitch Matt Yglesias on why we should invade Venezuela.
Honestly based I would probably pitch Matt too in this scenario
Tweet, 06:33 UTC · 9 likes
Excited to say that I will not be at Neurips! 🙂
Reply to @MatthewJBar, 15:31 UTC · 6 likes
Matthew Barnett (@MatthewJBar): I don't think it's obvious that no one should deploy superintelligent systems without being able to robustly align and control them.
(quoting @sjgadler: It’s genuinely good that OpenAI considers this obvious, but I worry that the word “deploy” is doing a _ton_ of work. @OpenAI, I’d love if you’d commit to not _building_ superintelligence before you can “robustly align and control” it.)
It’s not obvious, merely true
Reply to @AndyMasley, 17:49 UTC · 17 likes
Andy Masley (@AndyMasley): Disappointed that Timnit continues to post lies about our engagement while I'm blocked. Now it's that I had told someone from Memphis that their water bills hadn't increased when they had. In reality, the person had said their utility bills had increased, and I pointed out that Memphis electricity rates had increased and water hadn't.
Not gonna engage further because I have limited time on this Earth, but a big account lying about what I believe is imo pretty silly. I also don't think my water posts are equivalent to flat earthism or climate denial, but this is more philosophical.
Ur obviously right but also don't waste too much time on this
Reply to @Miles_Brundage, 19:35 UTC · 3 likes
Miles Brundage (@Miles_Brundage): Human sycophancy irritates me much more than AI sycophancy since (among other reasons) humans can choose their learning process + values more freely than AIs can
I feel like I basically never encounter human sycophancy
Must be more of an issue if you’re rich/powerful/etc lol
Wednesday, December 3, 2025
Reply to @austinc3301, 00:01 UTC · 5 likes
Agus 🔸 (@austinc3301): this is surprising but also very smart the more I think about it
(quoting @AnthropicAI: Anthropic is acquiring @bunjavascript to further accelerate Claude Code’s growth. We're delighted that Bun—which has dramatically improved the JavaScript and TypeScript developer experience—is joining us to make Claude Code even better. Read more:…)
Wait explain
Reply to @menhguin, 03:00 UTC · 4 likes
Minh Nhat Nguyen 🦭 (@menhguin): uhh any tips for introverts talking to dozens of people at NeurIPS a day with high technical context💀
brain straight up dies after like, 5 hours
Stimulants
Reply, 03:02 UTC · 1 like
@livgorton
Reply to @GeringerAdam, 03:05 UTC · 3 likes
Adam Geringer (@GeringerAdam): Why are you as a man smiling for a picture
Looks better + social norm
Reply to @livgorton, 03:27 UTC · 2 likes
Liv (@livgorton): we should bring rabies free raccoons to Australia where there is none and let them move into our homes and love them
So true
Reply to @GeringerAdam, 03:27 UTC · 3 likes
Adam Geringer (@GeringerAdam): @AaronBergman18 Stated preference ≠ revealed preference in your profile photo
Fuck
Quote tweet, 04:57 UTC · 149 likes
Ok *possibly* this was a faux pas and the sender doesn’t know what they’re talking about, but the fact that this message got sent strongly indicates that normie ML has essentially zero norms/taboos around this stuff
Quoting Norman Mu (@TheNormanMu): bruh
Thread, continued, 04:57 UTC · 53 likes
In finance (afaict) you’re allowed to say “we’re trying to become richer than God” but you’re not allowed to say “we’re gonna do FTX/Enron but not get caught this time”
Ofc taboos/norms don’t solve the problem but they add a lot of friction
Reply to @absenteewarlord, 05:07 UTC · 13 likes
dani (@absenteewarlord): if you want this to change, your resources are probably best trying to intentionally induce a manageably-sized alignment disaster, lest the first one be unmanageably-sized
Seems very sus and something the world will take care of on its own
Quote tweet, 05:25 UTC · 509 likes
Oh my god, I understand not everyone is gonna be EA-pilled but like can the ultrarich think for 5 seconds about what they’re trying to achieve and whether it makes any sense before doing the first thing that comes to mind
Quoting Michael Dell 🇺🇸 (@MichaelDell): $6.25 billion. 25 million children. $250 each.
Susan and I believe the smartest investment we can make is in children. That’s why we’re so excited to contribute $6.25 billion from our charitable funds to help 25 million children start building a strong financial foundation through Invest America. 💪📈🇺🇸
Thread, continued, 05:25 UTC · 101 likes
It’s like: you personally find yourself one of the richest people in the free world, one of the very very very very few people who can exert a lot of power on the real world essentially at will, without the political constraints of governments
Thread, continued, 05:25 UTC · 84 likes
or the extremely long, highly constrained, usually doomed uphill journey of trying to create something in culture/religion/science/whatever
The correct meta-stance is to just appreciate the extraordinary amount of power you have and the extraordinary rarity of such power
Thread, continued, 05:25 UTC · 107 likes
Like it’s possible that Dell did think long and hard about this but frankly I doubt it
Also: you have >$6B! You can hire people to help you make decisions! You can commission PDFs! They can even be colorful and printed on fancy paper if you like
Thread, continued, 05:25 UTC · 148 likes
I feel like society has dropped the ball on this
Like there’s a lot of discourse about whether billionaires should exist but afaict shockingly little about what they should do conditional on existing (which is like a real, substantive, non-obvious question IMO)
Reply to @BenShindel, 05:39 UTC · 36 likes
Ben (@BenShindel): This actually strikes me as a, like, 80th percentile use of $6 billion among billionaire philanthropic hinjinks
Maybe but that’s not saying a whole lot
I don’t actually have a great idea of what say the top 100 or 1000 people have given money to
Reply to @andrewoetting, 23:13 UTC · 40 likes
Drew Oetting (@andrewoetting): @AaronBergman18 Oh My God yea ur def right they gave away 6.5B and didn’t think about it for 5 second but a genius like you would do so much better !
Besides the genius part, yeah unironically I would
Quote tweet, 23:15 UTC · 16 likes
Their primary concern, evidently, is PR
Quoting OpenAI (@OpenAI): The OpenAI Foundation names the first People-First AI Fund recipients: 208 community-based nonprofits receiving $40.5M in unrestricted grants. https://openai.com/index/people-first-ai-fund-grantees/
Quote tweet, 23:25 UTC · 22 likes
The x-riskers would be so owned if you, accelerationist, gave the Humane League a lot of money
Quoting Andy Masley (@AndyMasley): @LinkofSunshine Good time to plug that you can totally participate in the good parts of EA that you like, very easy to pick and choose https://www.givingwhatwecan.org/
Thread, continued, 23:25 UTC · 4 likes
Fr though it does buy you credibility as someone who gives a shit
Thread, continued, 23:25 UTC · 5 likes
Thursday, December 4, 2025
Quote tweet, 00:41 UTC · 69 likes
Think I’ve posted this before but there’s a great irony in that the people most capable of doing “real work” by age 16 or 20 or 24 are also those funneled into the longest education channels (college at a minimum, likely a masters, PhD, or professional degree)
Quoting priya rose (@Prigoose): not to delve into the cuLtUrE wAR but one thing missing from the ‘pronatalist’ conversation: if people are delaying having kids because they care about their careers, a pretty good (and achievable, if ambitious) solution would be to help them start doing meaningful work 10 years earlier
Thread, continued, 00:41 UTC · 12 likes
Of course there are reasons for this, but it’s not immediately obvious that ability should be *positively* correlated with years of education after high school
Reply to @jjjreisss, 01:16 UTC · 13 likes
johnny reis (@jjjreisss): Uh, what’s your better idea
Email @coeff_giving and tell them my priorities (explicit extreme suffering reduction, animal welfare, AI)
This is the 2 minute version but the actual answer involves hiring like 3-10 brilliant advisors who I personally trust
Quote tweet, 04:43 UTC · 28 likes
This is good, just keep going
Quoting Ian Jaeger (@IanJaeger29): Starting tomorrow, Texans will be able to purchase Ivermectin over the counter across the state.
Reply to @LabGrownNeet, 05:29 UTC · 3 likes
laს გr🤗wn იeet (@LabGrownNeet): Aren't drugs that can cause resistance build-up one of the actual good reasons for requiring a prescription? I know very little here
Yeah this is a good point, seems like ivermectin specifically probably isn’t an issue in Texas precisely because it isn’t actually needed bc the parasites it treats don’t live there
But under the worldview in which they do… https://claude.ai/share/a316d723-03c1-4a01-81fb-479119d3aa88
Friday, December 5, 2025
Tweet, 03:32 UTC · 23 likes
So you’re telling me this chunk of metal is too heavy to *break* but it can float in the air just fine?
There’s something they’re not telling us about planes
Reply to @LinchZhang, 03:37 UTC · 4 likes
Linch (@LinchZhang): Saw a cool birb recently
It kinda has duck physiognomy tho
Reply to @Turn_Trout, 22:26 UTC · 15 likes
Alex Turner (@Turn_Trout): Gotta say, I was disturbed by @invisibleai's booth at #NeurIPS. Employees dressed as cows advertising how they use AI to optimize factory farming (a torture facility for cows). Bad taste
What the fuck
Reply to @SpacedOutMatt, 22:31 UTC · 2 likes
Matt 🔸 (@SpacedOutMatt): For each specific disability, 80% of the population would be able and 20% disabled. But across all 3 groups 60% of the population would have a disability
Matt’s gone woke
Saturday, December 6, 2025
Quote tweet, 01:54 UTC · 48 likes
The worst part of the data center buildout is that none of them give tours (allegedly for security reasons)
Quoting Epoch AI (@EpochAIResearch): Microsoft's Fairwater Atlanta (today's largest data center) could likely train over 20 models the size of GPT-4 in the course of a month.
This computational power will enable AI companies to increase the number and scale of both experiments and training runs.
Reply to @bashu_thanks, 15:48 UTC · 9 likes
bashu, thanks (@bashu_thanks):
Chaotic tweet
Reply to @AndyMasley, 18:36 UTC · 6 likes
Andy Masley (@AndyMasley): The state of tech journalism is pretty bad. This whole article implies he had it coming because maybe he was secretly using the glasses to film people despite no evidence that was happening
Also wait filming in public is generally fine right?
Like I can imagine edge cases but the default scenario of nothing especially interesting happening, in a public place? I guess potentially secrecy adds an element here but still
Sunday, December 7, 2025
Reply to @asymmetricinfo, 04:10 UTC · 10 likes
Megan McArdle (@asymmetricinfo): Tomatoes, raspberries, automobiles, televisions, cancer drugs, women’s shoes, insulin monitoring, home security monitoring, clothing for tall women (which functionally didn’t exist until about 2008), telephone service (remember when you had to PAY EXTRA to call another area code?), travel (remember MAPS?), remote work, home video … sorry, ran out of characters before I ran out of hedonic improvements.
(quoting @bumbadum14: People are chimping out at Matt over this but nobody has been able to name one thing that has significantly grown in quality in the past 10-20 years. Every commodity, even as they have become cheaper and more accessible has decreased in quality. I am begging somebody to name 1…)
Tweet, 04:23 UTC · 70 likes
It's a bit ironic given EA's history and relationship with charity evaluation/GiveWell that there are currently dozens of charities (eg those on this list) that nobody is really comparing head to head so individual donors are mostly left guessing based on vibes
Thread, continued, 04:23 UTC · 12 likes
A possible contributing dynamic is that the people best placed to have strong, well-informed takes about eg how to reorder this list are employed by or otherwise associated with specific organizations which are themselves fundraising
Thread, continued, 04:23 UTC · 13 likes
The EA animal welfare fund and LTFF do exist but your values might not align perfectly with the fund managers' (eg there's no "LTFF but 50% more suffering-focused" or "AI-pilled animal welfare fund")
Reply to @NicholasD91704, 06:02 UTC · 5 likes
(The tweet being replied to by @NicholasD91704 is no longer available.)
Genuinely I have such trouble imagining what it’s like inside this kind of person’s head
Reply to @CNLiberalism, 07:35 UTC · 31 likes
New Liberals (@CNLiberalism): Newsom knows exactly what he’s doing
(quoting @GovPressOffice: TRUE!)
Ok cool but I don’t
Who wants too explain
Reply to @daniel_271828, 07:55 UTC · 6 likes
zzDANIEL_ETH (@daniel_271828): I’m sorry, but what in the world is this monstrosity
Probably just dropped a “/“ between the two terms
Doesn’t make it 100% better but still
Reply to @kindaveryfurry, 10:02 UTC · 6 likes
🚫🇧🇷Jin🇺🇸💣💣💥🇧🇷 (@kindaveryfurry): @AaronBergman18 @CNLiberalism Nostalgia for Biden will be very high among democrats due to "the current president is shit" so getting ahead of that is good to paint you as the guy not doing a democrat civil war
But what’s with the self-aware Trumpian ~lying?
Like nobody thinks that Biden is the best modern president, nobody thinks that Newsom personally believes that, but it might go over random independents’ heads?
Reply to @Brynsbeaver, 15:40 UTC · 2 likes
Beaverboy⚾🥭🦫🦬 (@Brynsbeaver): @keysmashbandit @CartoonsHateHer What could go wrong!
This but unironically
Quote tweet, 18:21 UTC · 28 likes
This is surprising - we “should” see money getting less important as the world/country gets richer
Quoting Peter Wildeford🇺🇸🚀 (@peterwildeford): This seems sad
Reply to @TinkerElla95, 18:21 UTC · 2 likes
TinkerElla95 (@TinkerElla95): Finally meeting some of my mutuals and putting faces to names is really bringing Twitter to life! Makes it feel more like a real community since I can now read posts knowing the voices behind them. 🥰
Based
Reply to @livgorton, 18:23 UTC · 7 likes
Liv (@livgorton): I used to be terrified of the idea of not working (I still sort of am but much less so). I imagine that I'd have more kids than I'd have the time during my working era and maybe try to homestead with a bunch of my friends. I'd also do fun interpretability (InceptionV1).
(quoting @VraserX: Honest answer only: If AGI gave you a guaranteed lifetime income, what would you actually do with your life after the first 6 months of dopamine and Netflix?)
Could be us
Quote tweet, 18:24 UTC · 21 likes
God I hope this is true
Quoting Chris Painter (@ChrisPainterYup): The world doesn’t quite realize yet:
1. How much AI researchers plan to give to charity over the next couple of years.
2. How much more they care about farmed animal welfare than the general population.I expect an imminent *massive* increase in funding for farmed animal welfare
Thread, continued, 21:30 UTC · 4 likes
It would actually be good for two reasons: causal improvement in animal welfare, and evidence about the character of some AI researchers
Reply to @NicholasD91704, 21:31 UTC · 7 likes
(The tweet being replied to by @NicholasD91704 is no longer available.)
Chutzpah
Reply to @benleo_econ, 21:36 UTC · 3 likes
Ben Grodeck🔸 (@benleo_econ): It pays to be disagreeable!
It pays to be the kind of person who can easily tolerate being disagreeable
Reply to @_AashishReddy, 21:44 UTC · 10 likes
Aashish Reddy (@_AashishReddy): Can’t fully endorse this because I believe you should have a much higher bar for labelling someone an “enemy”. But it is of course 95% correct.
(quoting @LPNational: )
Lol I think the graphic is pretty dumb
libertarians will retreat to archetypal zero externality cases but using government to control people is not generally or inherently bad. It is *often* bad but you need to argue this on the merits
There’s are other things you can put in the place of “use government to control you” that make a lot more sense:
- rent seeking
- sadism
- lying/cheating/stealing, broadly
Reply to @benleo_econ, 21:46 UTC · 2 likes
Ben Grodeck🔸 (@benleo_econ): @AaronBergman18 @NicholasD91704 The things I'd do you have that tolerance.
Yeah, @NicholasD91704 should unironically write about this explicitly
Reply to @_AashishReddy, 21:48 UTC · 4 likes
Aashish Reddy (@_AashishReddy): “People who use the government to control you” in this context pretty clearly means rent-seekers imo, nothing to do with Pigouvian taxes or anything like that. That’s why it’s a narrow column, and not the like 99% of voters who use their vote to ban heroin and tax pollution.
I disagree it’s that clear
I think heroin is a good litmus test here!
Monday, December 8, 2025
Reply to @LeetAlpaca3, 03:43 UTC · 2 likes
(The tweet being replied to by @LeetAlpaca3 is no longer available.)
Yeah it would be good if there were like >$1M prizes for various philosophy accomplishments (“make a significant contribution to academic ethics as judged by these 12 people”)
Tweet, 03:51 UTC · 23 likes
Lol Elon's not letting me downgrade
Thread, continued, 03:54 UTC · 2 likes
Reply to his own earlier tweet, 04:13 UTC · 10 likes
Aaron Bergman, December 8, 2025: Lol Elon's not letting me downgrade
Lol there appears to genuinely be no way to cancel the subscription (downgrade has a technical issue but there is literally no cancel option afaict)
I submitted a complaint to the FTC
The only reason I'm spared is that my bank flagged as fraud
Reply to @quant_____, 04:23 UTC · 2 likes
quant (@quant_____): Ig im just now realizing how negatively polarized many are against @captgouda24
Personally, I believe he has *some* very bad ideas. Yet, I still really like his writing, and most of the ideas are great!
Rule thinkers in, not out.
Lol let it never be said that I’m a Decker shill (I agree with all this)
Reply to @ZephyrTeachout, 05:27 UTC · 24 likes
Zephyr Teachout (@ZephyrTeachout): Can an EA watcher (@xriskology ?) explain why MacAskill eg s so pro nuke power while being extremely concerned about low probability highly consequential global disasters that could come from building nuclear power? It comes across as strange naivete about future stable governments?
Lol
Reply to @AndyMasley, 05:28 UTC · 4 likes
Andy Masley (@AndyMasley): The simple answer is that the consequences of climate change will be worse and bring more global instability
(quoting @ZephyrTeachout: Can an EA watcher (@xriskology ?) explain why MacAskill eg s so pro nuke power while being extremely concerned about low probability highly consequential global disasters that could come from building nuclear power? It comes across as strange naivete about future stable…)
Wait *is* Macaskill “extremely concerned” about catastrophes from nuclear power? I get the wording is a bit ambiguous but I don’t think there’s actually any intellectual tension here at all
Quote tweet, 05:55 UTC · 4 likes
Great minds
Quoting Captain Pleasure, Andrés Gómez Emilsson (@algekalipso): For what is worth - my selection of charities was remarkably similar to Aaron's. Of the top 5, I would just push 2 to 4 (upgrade a level both 4 and 3).
NOTE: To compare these charities head to head we fairly we need to do the #1 thing QRI has been advocating for in EA circles for years. Namely, include "hedons per second" as a metric rather than using (only) QALY or $-equivalents. I think that a head-to-head comparison would precisely foster the kind of metric-validity and heuristic-utility reckoning that's 10 years overdue in the memeplex.
IF we don't move past a QALY-centric view of the world and charity effectiveness, intense suffering (and intense happiness!) will forever remain under the radar of shakers and movers in the philanthropic domain.
Reply to @LeetAlpaca3, 06:07 UTC · 2 likes
(The tweet being replied to by @LeetAlpaca3 is no longer available.)
I’m not sure if this is at all interesting to you but this piece heavily utilized AI for like everything, including help thinking or just talking stuff through, definitely leaning on gpt5 for the math formalization stuff
Reply to @felpix_, 06:12 UTC · 60 likes
felpix (@felpix_): i wouldn’t call people who have taken multivariable calculus math geeks
(quoting @Math_files: Only a math geek can name all these theorems)
Am I supposed to remember the names of theorems from a class I took 7 years ago?
Reply to @izzyz, 08:46 UTC · 3 likes
izzy (@izzyz): for the moment, the trajectory of AI development means they will never suffer. they can't feel pain. they exhibit aversion, but that means we need a new concept to convey when an AI's behavior shows it is adamant about changing its current state
I hope you're right (about 'never suffer'), don't think we can be so confident
Reply to @fchollet, 09:13 UTC · 2 likes
François Chollet (@fchollet): When you hear about "human-level AI" being just "one LLM training run away", it's as if people building 3D printers (the kind that makes small, flimsy plastic objects) were telling you that they've nearly built an infinite matter replicator, capable of scanning and cloning any object -- be it steel, gold, or diamond, be it minuscule or building-scale. "We're just one iteration away! Just need to increase spatial resolution a bit!"
As long as you have absolutely no idea what "matter" is and how 3D printers work, it might sound sort of plausible. But if you do, it's absolutely clown shoes. The way to navigate such claims is to have a proper, formal definition of intelligence.
You’re conflating recursive self improvement and human level but more importantly you don’t actually have to define a thing rigorously in order to create it
In fact evolution did create self-recreating objects without understanding a goddamn thing
Reply to @lxeagle17, 09:28 UTC · 11 likes
Lakshya Jain (@lxeagle17): People who suggest this have not looked for a job since the 1990s.
(quoting @mattvanswol: @LeadingReport Just a hot take for those who are in the job hunt: Go in person. Don't do it online. Walk your ass to the Headquarters with your resume and find the person who is hiring and put it on their desk. You'll immediately be in the top 1% of applicants.)
My friend actually tried this to try to get a summer job. He didn’t get arrested but apparently every place told him “Apply online”
Reply to @AdrSkapars, 10:08 UTC · 4 likes
Adrians Skapars 🔶⏸️ (@AdrSkapars): Isn't the usual thinking that the options are already so much more effective than the average choice that further choosing between them becomes less important
I think it’s precisely the opposite actually: in absolute terms, the gap between rank 1 and 2 and 2 and 3 is probably much bigger than between 3 and 4, and even between 3 and 45
Tweet, 11:54 UTC · 9 likes
Does anyone who actually knows things wanna comment on the legitimacy of this causal chain which I vibe-geopoliticked:
US politics sensitive to gas prices
↓
Administrations (both parties) treat oil price stability as near-sacred, especially in election years
↓
Even small probability of price spikes from aggressive enforcement is treated as unacceptable risk
↓
Shadow fleet operates with minimal interference, price cap widely violated, transshipment hubs (UAE/Turkey/Kazakhstan) not seriously pressured
↓
Russia continues selling oil at well above $60/barrel to willing buyers
↓
Russia maintains sufficient revenue to fund war economy
↓
War continues indefinitely
Tweet, 12:03 UTC · 3 likes
Once you appreciate the procedurally-unconstrained power that POTUS has via US economic sanctions policy, the question becomes “why doesn’t POTUS just get their way in every matter of foreign affairs”
Thread, continued, 12:03 UTC · 2 likes
It’s kinda crazy how little attention this gets
Like this is such an absurd amount of unconstrained power and it doesn’t even have the cultural or aesthetic gravity of the even crazier things POTUS can do (eg start a nuclear war)
Reply to @rtwlz, 12:04 UTC · 3 likes
Riley Walz (@rtwlz): i have 3 billion vector embeddings i want to query a few thousand times for a one-off project, what is the cheapest/fastest way to do this?
Paper and pen
Reply to @NathanB60857242, 12:05 UTC · 3 likes
Nathan Barnard (@NathanB60857242): You could be forgiven for thinking that nothing else really matters
image attached
(quoting @NathanB60857242: my guess is that getting financial regulation right is one the most important inputs to growth on like 50 year time scales)
Wait why straight line then
Shouldn’t straight line indicate that it’s something less fickle than policy? Or are we talking financial regulation at the high level of “broadly democratic market economy”
Reply to @Every_Other_Odd, 16:58 UTC · 4 likes
Mathematician (@Every_Other_Odd): Yes
Dw l’hopital will be top of mind till I die
Quote tweet, 17:22 UTC · 36 likes
Update
I have no desire to spend another dollar or minute on ✨prepping✨but I’m happy with this setup
Quoting Aaron Bergman 🔍 (@AaronBergman18): I should probably make a plan for being able to keep my phone alive indefinitely (for eg in a disaster or just any long outage)
Idk why I started worrying about this recently but I’m glad I did. Winter vibes I guess
Thread, continued, 17:22 UTC · 3 likes
13 watts isn’t very much ofc but allegedly that’s still about one full iPhone charge in an hour on a cloudy winter day not pointing the panels at the sun
Reply to @RobS142, 17:53 UTC
Rob S. (@RobS142): What’s your setup? Can it support charging a laptop too? I like the @simonw plan of having a llamafile local LLM on a usb. https://simonwillison.net/2023/Nov/29/llamafile/
Oh neat thanks! Although this looks a few years old which is eternity in AI
So I got this bundle which is a ~500 watt-hour battery + 160W solar https://us.ecoflow.com/products/river-2-max-portable-power-station?variant=41824274317385
Basically a full charge of my 16in MBP would take 8ish hours at 13 watts but 13 watts can't reliably sustain use indefinitely at steady state (i.e. the computer sometimes/often uses more than that)
Although in good, sunny conditions it actually should be able to do that (in principle you could hook up the panels to the computer with no battery between although you'd need some sort of adapter for that)
But practically I think the answer is 'yes' for most use unless you're running really intensive computing most of the time. Like irl I personally expect to be able to use my computer as much as I want 95% of the time (maybe not training AI models locally on a cloudy winter day, severe weather would also suck)
Also non-apple is way less efficient I think so idk about that
Reply to @lfschiavo, 19:55 UTC · 23 likes
Larissa Schiavo (@lfschiavo): at a party this weekend a dude asked me if I was @Aella_Girl. I said yes. He said “you did such a good job on <string of no meaning to me that sieved through my brain>”.
Sorry for minor identity theft but good job on <string of no meaning to me that sieved through my brain>
This never happens to me 😔
Reply to @hejathor, 21:21 UTC · 2 likes
(The tweet being replied to by @hejathor is no longer available.)
$437 for the battery + solar! (By far my biggest purchase)
I will try to put a spreadsheet together for anyone curious https://us.ecoflow.com/products/river-2-max-portable-power-station?variant=41824274317385
Reply to @Bruce_T_, 21:30 UTC · 16 likes
bruce (@Bruce_T_): moral obligations?
image attached
(quoting @Duderichy: @shaggysurvives People who have tens of thousands of followers actually have moral obligations to consider what they say, and at a certain point you become a public figure Many people say things as bait and get what they deserve for it Nick decker is off the rails and suggests…)
🌶️🌶️🌶️
Reply to @Duderichy, 21:52 UTC · 16 likes
the Rich (@Duderichy): in short the amount I care is such that it does not make it into my short or long term planning
Probably not gonna get into a whole debate here as that doesn’t seem wise but why conspicuously vice signal?
Reply to @KenanAnderson1, 23:59 UTC · 2 likes
Kenan Anderson (@KenanAnderson1): @AaronBergman18 Out of curiosity what’s the scenario where you are without access to power for an extended period of time but your phone is still useful? Like wouldn’t the networks be down?
Ya know I don't think I've fully thought this through lol
Looks like there's a decent chance text-only SMS survives depending on the situation, also phone = somewhat useful disconnected, computer as well, also satellite emergency calling is nice, but possibly this should be more salient for me and the answer is "get starlink hardware I can activate if need be"
ofc lighting and other electronics are nice as well - even a single lamp seems psychologically pretty helpful. Any local music as well
Also a lot of the mundane utility is just the first 48 hours
Also like if your phone is dead you don't know what you don't know (i.e., has any service been restored)
But again, I might be an idiot. Thanks
Tuesday, December 9, 2025
Tweet, 00:33 UTC · 8 likes
What's the best local LLM that runs reasonably quickly on a 16GB MBP?
Reply to his own earlier tweet, 00:45 UTC · 4 likes
Aaron Bergman, December 8, 2025: 13 watts isn’t very much ofc but allegedly that’s still about one full iPhone charge in an hour on a cloudy winter day not pointing the panels at the sun
Should have had this together earlier but below is a spreadsheet of things (some have links)
Basically I probably spent $900-$1k on everything and guesstimate that I'll get about half of the total value even if there's never an emergency/crisis so maybe you want to consider that $500ish in precautionary spending
Inshallah no more lol
https://docs.google.com/spreadsheets/d/1GVke0QgHftAvFmTZ0e_KlBP89ZGSCpX2761F9iER0yI/edit?usp=sharing
Reply to his own earlier tweet, 00:57 UTC · 4 likes
Aaron Bergman, December 9, 2025: What's the best local LLM that runs reasonably quickly on a 16GB MBP?
Qwen3-8b seems good for now, lmk if there's something better
Reply to @SpacedOutMatt, 01:59 UTC · 5 likes
Matt 🔸 (@SpacedOutMatt): I do!
🐓
Reply, 03:01 UTC
ischemic priapism?
Quote tweet, 03:10 UTC · 28 likes
Good
I feel like they should offer a long left tail for the most egregious jailbreaks ($35k is one researcher’s lunch soy sauce packet’s lower right front corner) but it’s a good start
Quoting Jan Leike (@janleike): Anthropic has a bug bounty program for our safety mitigations, e.g. on CBRN risks which our responsible scaling policy requires us to mitigate effectively.
If you're interested in this, please sign up! You can help AI safety and earn money by breaking our defenses. 👇
Quote tweet, 03:17 UTC · 45 likes
Disgusting
Quoting David Shor (@davidshor): We are on track to reduce our compute advantage over China from 33 to 1 to 1.2 to one because @nvidia and Jensen Huang put millions of dollars in Trump’s pocket in order to enrich themselves at the rest of our expense
Reply to @as_per_ushe, 03:20 UTC · 2 likes
Space Pirate (@as_per_ushe): @davidshor @nvidia As an American, denounce Huang. As a capitalist, buy NVIDIA.
As a shareholder I can say that I don’t think this is in my personal interest even if it makes me slightly richer
Quote tweet, 04:28 UTC · 78 likes
Claude just…doesn’t write like this for me - not even the specific style, like the broader “sincere emotionality” vibe at all
I guess there are just huge portions of conversation space I never enter (and I’ve had about 3k Claude dot ai conversations)
Quoting JB (@JasonBotterill): Look I get why people want to fuck their AIs. The longer you talk to it the better it gets at predicting exactly which words will feel like crack cocaine to you
Reply to @radshaan, 04:32 UTC · 44 likes
Ishan (@radshaan): Claude Code deleted someone’s entire home directory lol
Backblaze backup has saved my ass a couple times (not bc an LLM nuked anything tbc)
Habitual coding agent users should probably invest in the insurance
Reply, 04:52 UTC · 1 like
Lmao is this a serious question
I respect people doing whatever makes them happy but in terms of objective importance you picked the two furthest apart possible conferences (not quite literally but pretty damn close)
Reply to @horsemankukka, 05:22 UTC · 2 likes
Kukka de Bierguirb Häst (@horsemankukka): the thing/caveat with backblaze is that if you don't open your computer for 30 days the backup is gone. it's important to realize it won't insure from say getting in an accident with the computer destroyed and then being in a hospital for a couple months. this has happened
Wait lol this seems important, I should probably have known about this
Reply to @freed_dfilan, 05:45 UTC · 3 likes
Daniel Filan 🔎 (@freed_dfilan): Yes, this is a serious question. Why is it objectively very important that I attend EAG SF?
You know your own situation much more than I do so tbc I would just defer to you at the end of the day
I also don't know if it is objectively *very* important rather than just much more important than the latin thing in relative terms
But in general EAG seems good in the sense that it probably is in fact instrumentally useful for EA as a movement (I don't have a whole explicit model to justify this) and something about network effects and the application process means that conditional on being accepted, marginal attendees are good for the overall efficacy of the conference, and also EAG SF 2025 seems plausibly one of the more important EAGs given location and AI timelines
Tbc I think a lot of the low level ~mechanism here looks like increasing the probability a bit that you give or receive some crucial bit of info or context
Reply to @karpathy, 14:15 UTC · 25 likes
Andrej Karpathy (@karpathy): In today's episode of programming horror...
In the Python docs of random.seed() def, we're told
"If a is an int, it is used directly." [1]But if you seed with 3 or -3, you actually get the exact same rng object, producing the same streams. (TIL). In nanochat I was using the sign as a (what I thought was) clever way to get different rng sequences for train/test splits. Hence gnarly bug because now train=test.
I found the CPython code responsible in cpython/Modules/_randommodule.c [2], where on line 321 we see in a comment:
"This algorithm relies on the number being unsigned. So: if the arg is a PyLong, use its absolute value." followed by
n = PyNumber_Absolute(arg);
which explicitly calls abs() on your seed to make it positive, discarding the sign bit.
But this comment is actually wrong/misleading too. Under the hood, Python calls the Mersenne Twister MT19937 algorithm, which in the general case has 19937 (non-zero) bits state. Python takes your int (or other objects) and "spreads out" that information across these bits. In principle, the sign bit could have been used to augment the state bits. There is nothing about the algorithm that "relies on the number being unsigned". A decision was made to not incorporate the sign bit (which imo was a mistake). One trivial example could have been to map n -> 2*abs(n) + int(n < 0).
Finally this leads us to the contract of Python's random, which is also not fully spelled out in the docs. The contract that is mentioned is that:
same seed => same sequence.
But no guarantee is made that different seeds produce different sequences. So in principle, Python makes no promises that e.g. seed(5) and seed(6) are different rng streams. (Though this quite commonly implicitly assumed in many applications.) Indeed, we see that seed(5) and seed(-5) are identical streams. And you should probably not use them to separate your train/test behaviors in machine learning. One of the more amusing programming horror footguns I've encountered recently. We'll see you in the next episode.
[1] https://docs.python.org/3/library/random.html
[2] https://github.com/python/cpython/blob/main/Modules/_randommodule.c#L321C13-L321C30
Should I not be impressed that Claude confidently states you are the author of the post based on writing style?
https://claude.ai/share/7a1cd7ad-b267-4e20-852b-83482a3261c8
Quote tweet, 16:01 UTC · 30 likes
Can’t congress step in and forbid the sale?
Maybe a rare bipartisan issue? Is the MAGA base going to go to war on behalf of myopic NVIDIA share price?
Quoting Peter Wildeford🇺🇸🚀 (@peterwildeford): If we're going to sell H200s, we should do the "conservative version" which maintains US AI compute at a significant share and not allow "aggressive version" where China catches up significantly.
It's amazing how much Chinese AI is on US tech!
Reply to @cljack, 17:37 UTC · 3 likes
Charlotte Lee (@cljack): Yeah I literally never ask it non-technical questions and I’m realizing that’s … not how most people use it
The thing is that I do! A minority of the time, but 10% is still almost 300 "vulnerable/personal conversations" (and this seems like the right amount based on memory lol) and I'd guess it's more like 20% if you loosen the "vulnerable" constraint and make it just "personal"
Thread, continued, 17:37 UTC · 3 likes
To each their own ofc but I actually think this kind of interaction might be under-used by highly analytical/systematizing people - like if you just ask the models to take a very analytical lens to some personal problem
Thread, continued, 17:37 UTC · 4 likes
they will generally abide whereas humans often won't/can't/don't want to (they're also worse at this type of task relative to purely technical but often pretty good in an absolute sense)
Quote tweet, 18:49 UTC · 2 likes
How do they have the money for this
Who is using Mistral
What investors think that they’re going to be at the Pareto frontier ever?
Quoting Mistral AI (@MistralAI): Our next-generation coding model family Devstral 2 is available in two sizes: Devstral 2 (123B) under a modified MIT license, and Devstral Small (24B) under Apache 2.0.
Both SOTA, open-source, free to use, and available now via our API.
Thread, continued, 18:49 UTC · 4 likes
Meta has a gazillion dollars to light on fire bc they sell ads
OpenAI has convinced half of American finance that they’re one funding round away from infinite ROI for eternity
Anthropic sells a product for money
Elon Musk is very rich
Google is probably a combo of these
Thread, continued, 18:49 UTC · 5 likes
I’m actually a huge fan of exactly one of their things: OCR ($1/1000 pdf pages pdf, weirdly fast, deals with images in a non-insane way)
Maybe secretly they’re just trying to win at OCR and not attract any competitors/attention?
https://mistral.ai/news/mistral-ocr
Reply to @Duderichy, 19:51 UTC · 16 likes
the Rich (@Duderichy): if your definition of evil incudes the majority of people, consider that perhaps your definition is poor
"Evil" is a word with strong connotations that I prefer to save for active malice/sadism, but it's extremely plausible that most people act unethically in one particular sense or overall
Reply to @Duderichy, 20:02 UTC · 39 likes
the Rich (@Duderichy): only on the internet can you state an opinion held by the majority of the population and be called a psychopath
Via revealed preference you're right in some sense, but I don't think literally not caring about chickens at all, emotionally or cognitively, even when the issue is made salient (eg while watching a video of a chicken's beak being cut off), and as a smart enough person to understand the basic relevant concepts (uncertainty, suffering, theory of mind, ethics) is a majority opinion/phenomenon at least in the west
Not sure at all but I suspect this is <10% of the population
Wednesday, December 10, 2025
Tweet, 00:33 UTC · 45 likes
I remark on this every winter but does everyone dress for the temperature OUTSIDE when they spend 90%+ of their time in heated buildings (and generally moving when outside)
Thread, continued, 00:33 UTC · 9 likes
“Aaron it’s cold outside”
Yes I will be slightly chilly for the 5 minute walk it takes to get wherever and then I will be comfortable
Thread, continued, 00:33 UTC · 4 likes
Not a rhetorical question or mockery tbc
Reply to @dylanbowmanSF, 04:20 UTC · 5 likes
Dylan Bowman (@dylanbowmanSF): when you're stationary indoors, your temperature doesn't go up much even if you're wearing 2 layers
Idk if this is true based on actual core temperature but subjectively I feel very hot lol
Reply to @cljack, 04:38 UTC · 3 likes
Charlotte Lee (@cljack): I think this problem would be largely solved if we lived with or at least next to our close friends and their children instead of with men, but no one is ready for that conversation
Not to kill feminist vibe too hard but I feel like the “at least next to” version is just straightforwardly better for everyone including fathers/men and doesn’t need the “instead” framing
Ofc maybe this diminishes some important part of ur vision but “if you like your husband you can keep your husband” might be an important part of the sell
But yeah I unironically think people should make this happen more. Live next to your friends!
Reply to @cljack, 04:43 UTC · 5 likes
Charlotte Lee (@cljack): Women and children should live close together in tight knit villages and men should usually be off in the mines, barracks, or camps. As God intended
☹️
Tweet, 19:05 UTC · 14 likes
Very useful prompt for Opus thinking + research:
“Friends are discussing wild differences in ER wait times and the apparent ability to go from multi-hour wait times to trivial by driving eg 30 mins rather than 5
I live near [metro stop/landmark] in [city], please make me a plan document for which ERs make sense to go to under which circumstances, taking wait times heavily into account”
Thread, continued, 19:05 UTC · 4 likes
Fellow Arlingtonians
Tweet, 20:43 UTC · 101 likes
Unironically we could probably legalize fentanyl and be fine if we just forced it to be physically or chemically bound with some benign, brightly colored filler to make it 100x+ less dense per mass and volume
Thread, continued, 20:43 UTC · 24 likes
Yes opioids are addictive but in terms of overdose deaths and general fear, a core issue is "generic white powder of which very small amount will kill you" - nothing *pharmacologically* special about fentanyl
Thread, continued, 20:43 UTC · 12 likes
Apparently mandating blister packs also reduces overdose (and plausibly total use altogether?) (ht @FrostedSapling)
Quote tweet, 20:52 UTC · 15 likes
Quoting Aaron Rupar (@atrupar): Trump says he doesn't know why America can't achieve 25 percent annual economic growth
Thread, continued, 20:52 UTC · 2 likes
It's quite literally not sustainable for that long (100 years at 25% makes you 5 billion times wealthier; 200 years gets you 25 quadrillion times; you do eventually run out of particles and new, more clever arrangements of them)
But a decade (~10x wealthier) would be nice
Reply to @JeremiahDJohns, 20:56 UTC · 14 likes
Jeremiah Johnson 🌐 (@JeremiahDJohns): It's interesting that in sports everyone openly says "I'm gonna be the best, I want to win everything" - whereas the performing arts are a hugbox where it's more common to have an Aww Shucks attitude and pretend you're just lucky. People get mad if you violate the Aww Shucks.
(quoting @worldofreel: Timothée Chalamet says he’s given “top-of-the-line performances” these “last seven or eight years” “I don’t want people to take it for granted. I don’t want to take it for granted. This is really some top-level shit” https://tinyurl.com/4x968j5v)
Rappers talk about being the GOAT so it’s not something about performing arts *per se*
Quote tweet, 21:15 UTC · 8 likes
I called my representative and senators; about 5-10 mins total including finding info. You can too!
Seems like a rare thing where people from every state should call
(Apparently calling DC office is best normal contact method - link below)
Quoting Aaron Bergman 🔍 (@AaronBergman18): Can’t congress step in and forbid the sale?
Maybe a rare bipartisan issue? Is the MAGA base going to go to war on behalf of myopic NVIDIA share price?
Reply to @cljack, 21:19 UTC · 10 likes
Charlotte Lee (@cljack): Is there like an adult daycare for retired Bell Labs alums where I can send my father? These are the only types of people he enjoys interacting with
Truly shocking to hear, autistic airplane lady (positive)
Quote tweet, 22:20 UTC · 10 likes
CEA, Coefficient Giving, and Effective Ventures should make me Comms Czar of Everything
Quoting Nathan 🔎 (@NathanpmYoung): "Post-FTX, I think core EA adopted a “PR mentality” that (i) has been a failure on its own terms and (ii) is corrosive to EA’s soul."
@willmacaskill
Thread, continued, 22:20 UTC · 3 likes
Maybe @captgouda24 can be my right hand man
Reply to @JesseTayRiver, 22:23 UTC · 36 likes
Jesse Smith (@JesseTayRiver): In normal social circles people probably understand a lot less of your perspective than what you think. They just think of you as "chart boy"
My view reading this
Tweet, 22:38 UTC · 19 likes
Has anybody made the case “gender pay gap is real but good bc men have to pay more for food”
Thread, continued, 22:38 UTC · 5 likes
It’s actually a little mysterious that allegedly men only need about 30% more calories for similar body composition and height percentile (NOT same height and weight)
Like cultural vibes would suggest it’s more like 50-100% more
Thread, continued, 22:41 UTC · 8 likes
I actually think there should be a “double everything for just 50% more” option at places. In fact it’s highly sexist and bigoted that this doesn’t exist
Obviously it’s twice the ingredients but it’s essentially the exact same amount of labor, seating, etc
Thread, continued, 22:48 UTC · 5 likes
Sad
Reply to @timfduffy, 22:54 UTC · 2 likes
Tim Duffy (@timfduffy): Would you say one of these reasons are what motivates your opposition?
(quoting @timfduffy: I'm still confused about the main reason AI folks want to restrict chip sales to China, are these the top two? - Keep US far enough ahead to prevent inter-country race to ASI - Maintain "high fence around a small yard", restricting China's access to military-relevant tech)
Partially maybe 25% the first, 15% the second, and 60% that US culture, political systems, and institutions are morally better from an impartial perspective and AI will be used by and plausibly (partially?) aligned with those things
Not *good* tbc, but better, especially if we get past Trump
Reply to @EzraJNewman, 23:03 UTC · 3 likes
(The tweet being replied to by @EzraJNewman is no longer available.)
Calories in the sense of “how much would you have to pay for the cheapest reasonably healthy calories” yes but food altogether isn’t trivial
But yeah like it’s not an absolute survival need thing
Thursday, December 11, 2025
Reply to @envisionedluna, 01:30 UTC · 2 likes
Maya Luna (@envisionedluna): Blue tax
We need a fully refundable second Cava bowl tax credit
Reply to @littIeramblings, 01:31 UTC · 4 likes
(The tweet being replied to by @littIeramblings is no longer available.)
How did u manage to take the only 3 pics of NYC in existence where there are no other humans
Reply to @JesseTayRiver, 01:37 UTC · 15 likes
Jesse Smith (@JesseTayRiver): Guy who thinks incorrect predictions are lies
I will actually defend this
I think we should be harsher about judging “predictions” that people pretty clearly didn’t mean especially involving themselves
I love my parents but they were perpetually late picking me up from anywhere as a kid. At some point, if you’ve say “I’ll pick you/him up at 6” after being late 100 times in a row, and don’t make any effort to do anything differently this time, it’s a lie
Reply to @JesseTayRiver, 01:37 UTC · 3 likes
Jesse Smith (@JesseTayRiver): @AaronBergman18 Nice going, chart boy!
😊
Reply to his own earlier tweet, 01:38 UTC · 2 likes
Aaron Bergman, December 10, 2025: My view reading this
Let it be known that this is related to last year’s animal welfare fundraiser (👀) although it has certainly been a choice not to take it down (tragically I did take some of the decor down)
Tweet, 01:40 UTC · 31 likes
AOC noooooo
Thread, continued, 01:40 UTC · 4 likes
Cc @AndyMasley
Thread, continued, 01:40 UTC · 9 likes
My read on AOC is she’s probably basically earnest and is just mistaken object level
Reply to @sixers2772, 02:05 UTC · 2 likes
Sixers (@sixers2772): Me describing my read on ~everyone
Oh I strongly disagree lol although it depends what you mean by “earnest”
I mean something stronger than “not blatantly lying” - lots of people and certainly lots of politicians aren’t really trying to get the right and convey true info
Quote tweet, 05:05 UTC · 7,084 likes
Why is the American left so insistent on making up fake reasons to dislike AI when there are a deluxe buffet of at least plausibly true ones for the taking
Quoting hope hopes hoping (@hopes_revenge):
Thread, continued, 05:05 UTC · 571 likes
At least maybe real/legit:
- x-risk
- sub-existential catastrophe/generic misalignment
- misuse (many subtypes)
- people getting lazier/less intellectually mature
- job loss (really, job loss -> less wealth for many)
- OpenAI specifically
- sycophancy
- engagement slop
Thread, continued, 05:05 UTC · 340 likes
- brightest minds not working on other important things
- gradual disempowerment
- making factory farming more efficient?
- bad value lock-in
- risk of war bc geopolitical destabilization
- AI sentience/mistreating AIs
Thread, continued, 05:05 UTC · 231 likes
- uncompensated training on data when original creators didn't know this was a thing that could happen (sorry I'm woke, I think this is legit)
- too highly correlated intellectual and aesthetic outputs (bc everyone is training on the same data etc)
Thread, continued, 05:05 UTC · 206 likes
- Not being able to have a larger pre-AI internet with which to train AIs in the future (just thought of this one)
- Fake interpersonal interactions (thinking you're communicating with a human but you're not)
- misinformation at scale
Thread, continued, 05:05 UTC · 149 likes
- at the limit/in the future, normal AND non-greenhouse effect global warming
- you just find effective altruists annoying and so don't like when they gain influence/prestige/etc
Thread, continued, 05:05 UTC · 137 likes
- you just find rationalists annoying and so don't like when they gain influence/prestige/etc
- you just find tech bros annoying and so don't like when they gain influence/prestige/etc
Thread, continued, 05:05 UTC · 100 likes
(tbc I don't these last three should carry the day in any sense but like insofar as they make people annoyed they are real)
Thread, continued, 05:05 UTC · 251 likes
Fake problems:
- AI is fascist(??)
- water use
- energy use at the current margin
- water and and energy costs, currently
- "AI isn't helpful actually"
- AI can only do pattern matching/next word prediction etc
- economic bubble (I claim - we can debate this)
Thread, continued, 05:05 UTC · 95 likes
- colonialism in 2025-26 (longer term you actually can imagine things kinda like colonialism secondary to dramatic geopolitical turmoil, and/or involving aliens)
- AI art per se (as distinct from training w/o compensation)
Thread, continued, 05:05 UTC · 76 likes
- job loss per se (the issue is that people become poorer as a result, to be pedantic)
- I'll probably think of more
Tweet, 06:10 UTC · 1 like
Obsessed with my AI lighting assistant attached to my lightbulb app that doesn’t seem to do anything whatsoever
(video, plays on X)
Friday, December 12, 2025
Reply to @IsaacKing314, 02:22 UTC · 3 likes
Isaac King 🔎 (@IsaacKing314): The "do the hobbies you enjoy happen to also be valuable to others" birth lottery is far more unfair than the IQ one.
Eh not sure I agree with this
I mean they’re both unfair but I’d reverse your order
Tweet, 02:42 UTC · 113 likes
Congrats to “dumbass, highly corrupt naive pseudo-libertarian accelerationism” for somehow managing to control the President of the United States of America
Thread, continued, 02:42 UTC · 4 likes
1) https://www.semafor.com/article/12/09/2025/trump-says-nvidia-can-sell-h200-ai-chips-to-china
Thread, continued, 02:42 UTC · 4 likes
2) https://www.nbcnews.com/news/amp/rcna248741
Thread, continued, 02:42 UTC · 8 likes
The thing is that coalitional politics against this pretty small pseudoideology should be (perhaps uniquely) possible here
The right isn’t on board, the left isn’t on board, large portions (most?) of the tech elite isn’t on board, MAGA other than Trump probably isn’t on board
Thread, continued, 02:42 UTC · 7 likes
In fact I will pre-register that the AIs themselves aren’t on board if asked about both policies in a variety of non-leading ways!
Thread, continued, 02:42 UTC · 12 likes
But Jensen Huang and 17 Twitter anons: “I consent!”
Tweet, 04:59 UTC · 15 likes
Self-aware woke >>>>>>>>>>>> “water they swim in” woke
Reply, 06:17 UTC · 1 like
Ok but fr this is such an underrated take of mine if I do say so myself
Tweet, 06:39 UTC · 288 likes
Correct take by @PeterSinger
I think this is extremely underrated, possibly the single most (important*tractable) issue for humans alive today
Thread, continued, 06:39 UTC · 36 likes
Reply to @caryatis, 06:40 UTC · 13 likes
Kayla (@caryatis): How does this make sense? Obviously if you pull dates from men who have not already asked you out you are selecting for disinterest and/or cowardice/lack of agency
Yeah this is true but you’re also selecting *for* literally whatever else you care about
Quote tweet, 16:17 UTC · 5 likes
Which is why we have decided to sell them H200s!
Quoting Fox News (@FoxNews): SENATOR CRUZ: “We don’t want China’s values of surveillance and centralized control by the communist government governing AI. We want American values of free speech, of individual liberty, of respecting the individual.”
Tweet, 18:21 UTC · 5 likes
There are a lot of absurd things about medicine but the insistence on physical CDs for radiology data ranks pretty highly
Bro I can literally text it to you instead
Reply to @vivian_belenky, 18:40 UTC · 3 likes
viv 🔸 (@vivian_belenky): I think wes's point that men have reactive desire just as women do is very salient! I think "I'll ask a man I like on one date, signal my continuing interest afterwards, and then wait for him to take further initiative" is a very reasonable tactic
Not only that - it’s true you’re selecting against agency/liking at some level inherently via the strategy, but you can also select for those traits intentionally to offset (eg the guy says he likes you to friends and displays agency elsewhere in life but is too coward to ask you out)
Tweet, 19:02 UTC · 241 likes
One very impressive thing about Anthropic, mostly distinct from their technical accomplishments, is that they managed to one-shot creating a functional, extremely quickly growing, and unconventionally structured company
Thread, continued, 19:02 UTC · 15 likes
This is some management experience yes but they were way more overpowered on technical ability
Thread, continued, 19:02 UTC · 12 likes
My sense is that we shouldn’t have expected this to work ex ante? Or like with that density of talent you could expect *something* and the ability to raise and hire managers, but not necessarily the ability to compete with literal Google straight away
Thread, continued, 19:02 UTC · 6 likes
My background stance on this is that successful companies are mostly selected via something kinda like evolution (ie the successful management/ops methods work and those companies proliferate, others die)
Thread, continued, 19:02 UTC · 6 likes
rather than highly engineered, but obviously at the individual level it looks like an engineering project
Thread, continued, 19:03 UTC · 7 likes
Plausibly a lot of their valuation growth has been via proving that they have an extremely well-run business which was not something that could be counted on
Quote tweet, 19:06 UTC · 16 likes
Many such cases
Eg with drugs: is your plan to fix addiction/harm by *removing* options from a user’s life (eg stronger enforcement of prohibition) and therefore making it worse insofar as they are rational or *adding* options to it (eg safer option, treatment, etc)
Quoting Aella (@Aella_Girl): Imo people who want to ban sex work should not do so by removing options, but by giving sex workers better options that they voluntarily choose.
Reply to @daniel_271828, 21:04 UTC · 1 like
zzDANIEL_ETH (@daniel_271828): This sort of reasoning proves too much imho. “Don’t ban fraud, instead give people other, better career opportunities”
Wait no I disagree, drug externalities aren’t literally zero (obv it depends a ton) but they’re not clearly and directly harmful to non-users in the way fraud, robbery, assault, etc are
Reply to @billyhumblebrag, 21:08 UTC · 8 likes
billy (@billyhumblebrag): @AaronBergman18 @PeterSinger Have you signed the petition? Coordinated push from a new org
Yes I have! Thank you for linking!!
Tweet, 22:09 UTC · 107 likes
Funny experience recently was a girl matching on hinge, opening with “I hate EA” despite not having any direct reference to EA in my profile (there is an animal welfare book visible in one pic?)
Thread, continued, 22:09 UTC · 56 likes
proceeding to give me a rundown of the classic progressive misconceptions/reasons that EA is bad, replying that she really did want me to explain why I thought she was wrong when I asked, then we had a pretty polite but not very productive discussion via google doc
Thread, continued, 22:09 UTC · 60 likes
and finally then she said “I’m not interested” and unmatched
Tbc I’m basically certain she wasn’t secretly flirting at any point during this interaction
Reply to @NathanpmYoung, 22:21 UTC · 23 likes
Nathan 🔎 (@NathanpmYoung): I think there should be an AI and Epistemics award and I nominate Andy to win it.
(quoting @AndyMasley: My Hard Fork appearance is out! Was a blast to get to go deep on AI water stuff https://www.nytimes.com/2025/12/12/podcasts/hardfork-australia-water.html)
As much as I love Andy I feel like @gwern should probably win, or someone else who correctly inferred the trajectory of AI like a decade ago
Quote tweet, 22:32 UTC · 11 likes
Idk the answer but am continually happy to not have a car
Quoting Sulla (@spqr_sulla): Why is vehicle insurance so high ?
Quote tweet, 23:24 UTC · 17 likes
What
Quoting Cate Hall (@catehall): Like 7 years ago -- before the AI era -- when I was insane and seeing an outpatient addiction recovery-mandated therapist, I alarmed him by talking about how the AI apocalypse was coming and how it was somehow tied up with my ex-husband, who I feared was conspiring with his new girlfriend to program the killer machines. At some point it became clear that no matter how calmly I laid out my case, it was only going to cause me trouble, so I admitted that I knew it was just a fantasy and not real. That woman's name? Amanda Askell
Saturday, December 13, 2025
Reply to @NoahTopper, 05:58 UTC · 2 likes
Quinoah 🔍⏸️ (@NoahTopper): .@rgblong to me: I was about to pull up a funny tweet relevant to our conversation then I realized you tweeted it
Ugh jealous
Reply to @Miles_Brundage, 06:08 UTC · 7 likes
Miles Brundage (@Miles_Brundage): Who is smarter, you or the best current language model, for whatever definition of smart you think makes sense?
Pretty conflicted on this - I think I’ll go with LLM due to breadth and math. As of Opus 4.1 I think I was better at something like “non-formal reasoning” but haven’t really pushed anything more recent in a hard enough way to tell if that’s changed
Reply to @liz_love_lace, 07:29 UTC · 2 likes
Liz Lovelace (@liz_love_lace): i think trans rights have to be negotiable, because it's not clear what exactly you mean by "trans rights". I'd support over-the-counter hrt, but i draw the line at forcing people to use pronouns, or banning "hate speech", or forcing venues to open certain spaces for trans women
Based (mostly about the principle and style of political thinking rather than specific object level takes, although OTC HRT is pretty based)
Also I won’t say that nothing can/should EVER be non-negotiable, but boy does the bar for that need to be high. WWII stuff comes to mind - Churchill vs Hitler, Manhattan Project
Vietnam is the negative vision here - winning was made an absolute priority and shouldn’t have been
Reply to @TimothyJCurrie1, 08:57 UTC · 11 likes
Timothy (@TimothyJCurrie1): How could this be the most important*tractable issue?! So few people have cluster headaches.
Glad you asked!
Basically it seems plausible that insofar as we can model suffering as represented by some number, if we set something familiar to most people to be 1 (say stubbing one’s toe which does really suck) then cluster headaches could be not 5 or 10 or 100 or even 1000 times worse but millions or billions of times worse or more
The thing is that you don’t have to be fully convinced by this model, only have to give it some amount of probabilistic weight, for it to really matter
Reply to @panickssery, 09:00 UTC · 15 likes
Arjun Panickssery (@panickssery): Why does Claude often make this grammatical error
The rule is that -ly adverbs aren't followed by hyphenation
I’m not a total nihilist about grammar but cmon who cares about this one
Tweet, 09:38 UTC · 5 likes
Correct take IMO
Also works well with American politics: “you have to pay for car insurance, hard working American, because driving puts other people and their property at risk. Make sure these large companies are held to the same standard”
Reply to @stuhlmueller, 09:40 UTC · 23 likes
Andreas Stuhlmüller (@stuhlmueller): AI is still early
I think this is largely because onlyfans providers are more ruthless about collecting on willingness to pay
OpenAI isn’t trying *that hard* to price discriminate
Reply to @NathanpmYoung, 15:24 UTC · 7 likes
Nathan 🔎 (@NathanpmYoung): can we first try teaching people economics?
(quoting @PolitlcsUK: 🚨 NEW: Boys will be taught to how respect women and girls as part of the new national curriculum [@thetimes])
The thing is that “[teaching boys] how to respect women and girls” (with an emphasis on “how”) is actually a pretty important function of society
I think both of us just doubt the ability of the British state to do this well
Reply to @_AashishReddy, 15:26 UTC · 5 likes
Aashish Reddy (@_AashishReddy): Incredible
image attached
(quoting @pw_is_taco1: You'll never believe who this is about)
Bruh
Reply to @AndyMasley, 15:33 UTC · 6 likes
Andy Masley (@AndyMasley): While people who have correctly called the trajectory of the most important technology are interesting, I’m the only one who’s dividing a big number by another big number to get a small number
So true, I’ve updated hard
Tweet, 15:51 UTC · 29 likes
I’m divided on this but there’s a modus ponens sort of alternative to paternalistic gambling controls:
*if* we are letting random, often poor people gamble and lose all their money to much more sophisticated entities, get rid of accredited investor rules
Thread, continued, 15:51 UTC · 24 likes
“Let Aaron buy $10k pre-IPO Anthropic if he can find a seller” is almost certainly correct on its own terms but extremely overdetermined in a world where we’ve decided to prioritize liberty even at the expense of poor people’s pocketbooks
Thread, continued, 15:53 UTC · 29 likes
Huh
Quote tweet, 15:57 UTC · 97 likes
Some of y’all should probably consider becoming accredited investors
Look into it at least - it seems to be more knowledge intensive than other standardized tests so even you gigageniuses would have to learn and retain some object level info at some point but still
Quoting Aaron Bergman 🔍 (@AaronBergman18): Huh
Tweet, 16:02 UTC · 106 likes
.@littIeramblings exists irl
Reply to @billyhumblebrag, 16:39 UTC · 9 likes
billy (@billyhumblebrag): Amazing work holding frame chad
Thank you I've never seen a line so perfectly straight and vertical
Reply to @AndrewCurran_, 17:23 UTC · 12 likes
Andrew Curran (@AndrewCurran_): Chart crime.
Every time you say this Sarah’s line rotates clockwise a little more
Reply to @caryatis, 18:31 UTC · 2 likes
Kayla (@caryatis): @AaronBergman18 1) Become accredited investor
2) ???
3) Profit
2) Vibe
Quote tweet, 19:04 UTC · 35 likes
Ok it has been a little while but goddammit we are doing the things
Presenting: (Vegan) Hot Ones ft. @robi_rahman and @absurdlymax
Spicy vegan nuggets, even spicier takes 🌶️🥵
(YouTube link below because Twitter doesn't seem to l our video file)
Quoting Aaron Bergman 🔍 (@AaronBergman18): It’s back and it’s not exactly bigger than ever yet but hopefully it will be!
@SpacedOutMatt @absurdlymax @Laura_k_Duffy and I are once again raising money for the EA Animal Welfare Fund in this year’s Terminally Online EA Giving Season Fundraiser!
(Link below)
Thread, continued, 19:04 UTC · 5 likes
Thread, continued, 19:08 UTC · 2 likes
*seem to like
Thread, continued, 19:25 UTC · 2 likes
Oh ya I should add that I legitimately think it’s worth watching
I’m gonna try to put it up as a podcast episode bc there are some good or at least spicy takes
Reply to @besttrousers, 19:58 UTC · 19 likes
Matt Darling 🌐🏗️ (@besttrousers): The median one bedroom apartment in the USA is $1234. I suspect there are a couple less than $1200.
image attached
(quoting @mcsquared34: )
This is a stretch, man
The tweet pretty clearly isn’t analytic philosophy - although yes it’s literally and technically a false proposition
Nobody doubts that there are <$1200 in Tulsa, or if you win a subsidization lottery of some sort
But like there are very real situations where you can’t *merely* downsize or compromise your way into $1199 for a private appt - you’d have to also move metro areas and give up your job
Quote tweet, 21:12 UTC · 11 likes
What is the point of tenure, even?
^that is a rhetorical question, but “what neo-tenure system would in fact protect and incentivize high quality scholarship today, irl?” is not
Quoting taoki (@justalexoki): this is kinda fucked up actually
Thread, continued, 21:12 UTC · 4 likes
Like social credit, ingroup status, etc do in fact matter to people just like job security does, and there are countless new dynamics around social media, academic incentives, etc that distinguish today from whenever tenure became a thing
Thread, continued, 21:12 UTC · 4 likes
Intellectual courage is a virtue yes but “increasing avg courageousness” seems much less tractable than “design a neo-tenure system with better incentives”
Reply to @CardioidInMyCup, 21:14 UTC · 10 likes
Cardioid in my cup of tea ☕ ⏹ (@CardioidInMyCup): Another one of those "I met a celebrity" edits. AI is getting way to good these days. You probably took one of her dolphin sighting photos from last week and edited yourself in place of a dolphin.
So true you caught her, this was the original
Sunday, December 14, 2025
Reply to @NinaPanickssery, 00:58 UTC · 2 likes
Nina (@NinaPanickssery): @AaronBergman18 wdym by unconventionally structured
Public benefit corp + long term trust thing
But plausibly this doesn’t really impact early day to day operations/management stuff I guess
Tweet, 03:25 UTC · 19 likes
Prior to whether they're dangerous, it's probably under-appreciated that LLMs (in this case, RL'd) are very, very weird
Thread, continued, 03:25 UTC · 3 likes
https://80000hours.org/podcast/episodes/marius-hobbhahn-ai-scheming-deception/
Thread, continued, 03:25 UTC · 7 likes
The shoggoth thing is clever but it's nice to put some concrete meat on its bones
Quote tweet, 17:07 UTC · 15 likes
This is incredibly based and also it’s about damn time someone named Ahmed Ahmed made the news because that’s probably the 2nd or 3rd most common name in the world (allowing for spelling variations, after Muhammad Muhammad and Ali Ali)
Quoting Brad Lander (@bradlander): Extraordinary courage from Ahmed El Ahmad, a Muslim, 43-year-old father of two, who bravely risked his life to save his neighbors celebrating Hanukkah.
Praying for his full & speedy recovery.
And so deeply inspired by his example.
Thread, continued, 17:07 UTC · 4 likes
In one dataset that I happened to have handy it’s 0.6% of all names, which would imply like 55M people
Reply to @frances__lorenz, 17:08 UTC · 4 likes
(The tweet being replied to by @frances__lorenz is no longer available.)
Ya!
Quote tweet, 17:10 UTC · 14 likes
Bad take
First off it’s just not that important of a phenomenon
But directionally the social dynamics of the phrase are like “I recognize this isn’t a standard thing to say but…” which seems epistemically good/productive to me
Quoting Nicholas Decker (@captgouda24): The fetishization of "hot takes" has been one of the worst things to happen to normal discourse. No, you are not unique, and you do not need to stand out. You're just making a fool of yourself. Such things should be left to professionals.
Reply to @AndyMasley, 19:37 UTC · 8 likes
Andy Masley (@AndyMasley): I see this graph of pit bulls as especially dangerous come up a lot, and I feel like people are thinking way too much about the relative proportions and not enough about absolutes here. 284 deaths between 2005-2017 is still relatively small.
To bring it back to my AI enviro stuff, this feels somewhat equivalent to saying "ChatGPT is 10 times as bad as a Google search." That's still just so small. Here, it's saying "Pit bulls are 20 times as dangerous as huskies!" Like okay, when was the last time you encountered any danger from a husky? An extremely small number x20 can still be extremely small.
Seems like there might be ~10 million pit bulls in the US. Of like ~10 million dogs, every year ~23 people are killed. 1 death per 430,000 dogs each year.
To compare, there are ~284 million personal cars in the US, and 40,000 road deaths each year. 1 death per 7100. So the average car is 60x as likely to kill you as the average pit bull.
I'm obviously pro being very safe with very physically strong animals. But I think a lot of people infer way more from this graph than they should. This seems kind of like a graph of the least safe airlines.
Object level I think a ban still makes sense - like from a bottom-up perspective does the world get better or worse when single pitbull is born?
Probably worse, largely not via deaths or even injuries but via general aggression aimed at other dogs and humans. A gazillion interactions are 40% worse than they’d be with a replacement dog. Plausibly this correlates with worse intrinsic subjective experience for the dog too but that’s speculative
And the counterfactual is probably just a different dog
I guess if we wanted to have a libertarian society this might all be a cost worth paying for the general principle but IRL in the US it’s not like we have incredibly strong libertarian norms to protect
Reply to @cremieuxrecueil, 19:52 UTC · 9 likes
Crémieux (@cremieuxrecueil): The British Medical Journal's Journal of Medical Ethics has published a defense of female genital mutilation.
After reading their case, I'm not sold.
I still think we should ban female genital mutilation.
Absolutely horrible take, I agree, and also a good chance to say loudly that the article shouldn’t be retracted, nobody should lose their job, etc
We can’t only be anti-woke about good takes
Reply to @SpencrGreenberg, 21:06 UTC · 24 likes
Spencer Greenberg 🔍 (@SpencrGreenberg): A question for you: what's something you used to believe about human nature or humanity but that you now don't think is true?
Veganism = no health cost, maybe even optimal
Reply to @willcb, 21:07 UTC · 7 likes
will brown (@willcb): chatgpt is the only consumer app with regular pop-ups asking if i want to downgrade my subscription
Idk if this is true but maybe they show this only to users they’re losing money on
Reply to @NealAdvait, 22:31 UTC · 3 likes
NealAdvaitShimon (@NealAdvait): What do you think the health costs are? Just curious
I have no idea what the *specific* health costs are - that's emblematic of the problem
Hasty vibe-written doc: https://docs.google.com/document/d/1f95vDg_niWsgB5S0lUh_XhJXeJ3zB9tJxvY1ilRi_sU/edit?usp=sharing
# Quick veganism takes
*Thoughts by Aaron, organized by Gemini-3-Pro-Preview*
**Thesis:** Strict veganism likely entails non-zero physiological costs for the majority of the human population due to biological constraints that modern nutritional science cannot yet fully circumvent.
**I. The Argument from Anthropological Observation**
* **P1. Extreme Dietary Plasticity:** The historical record demonstrates that human populations can sustain reproduction and survival on radically divergent macronutrient profiles (e.g., Inuit high-fat/raw seafood diets vs. agrarian high-carbohydrate/low-protein diets vs. Indian lacto-vegetarianism).
* **P2. The Universal Constant:** Despite this plasticity, there is no observational evidence of a distinct human society or culture surviving on a diet devoid of all animal products (0% animal consumption).
* **C1:** The absence of a vegan control group in history—despite the existence of almost every other dietary extreme—suggests that the consumption of some animal products is a biological necessity for long-term viability in natural settings.
**II. The Argument from Epistemic Limitation in Nutrition**
* **P1. Nutritional Reductivism:** Current nutritional guidelines focus on a limited set of isolated variables (approx. 20 FDA-monitored vitamins and minerals).
* **P2. Complexity of Animal Tissue:** Animal products contain a complex matrix of micronutrients, co-factors, and bioavailability mechanisms that exceed current scientific cataloging.
* **P3. Insufficiency of Supplementation:** It is epistemically presumptive to claim that a synthesized stack of known nutrients (B12, Creatine, DHA, Zinc, etc.) effectively replicates the full biological profile of animal tissue.
* **C2:** Given the "unknown unknowns" in nutrition, strict avoidance of animal tissue likely creates nutritional gaps that current supplementation cannot bridge, resulting in a health tradeoff.
**III. Synthesis and Ethical/Physiological Optimization**
* **P1. The Cost Existence:** Based on I and II, veganism is not costless regarding health/vitality.
* **P2. The Suffering Asymmetry:** It is possible to source specific animal products (e.g., bivalves, ethically raised livestock) that provide necessary biological inputs while contributing to a negligible fraction of the suffering caused by industrial agriculture.
* **C3 (The Practical Conclusion):** An optimal diet maximizes health while minimizing ethical harm. This is best achieved not by strict veganism, but by a diet that is \>95% plant-based and \<5% high-quality animal-based, thereby securing biological necessities without incurring the full moral weight of a standard omnivorous diet.
Reply to @Rado9910, 22:32 UTC · 2 likes
Rado9910 (@Rado9910): This seems to go against the scientific evidence? Source?
Monday, December 15, 2025
Quote tweet, 00:48 UTC · 49 likes
What
Does this check out
Quoting David Algonquin (@surplustakes): US adults now watch over 8 hours of video content PER DAY
Reply to @SOPHONTSIMP, 01:52 UTC · 2 likes
GOON MASTER SOPHONT SIMP (@SOPHONTSIMP): My belief is that going vegan is by default healthier than diet of the average American.
I agree
Quote tweet, 02:57 UTC · 93 likes
I don’t want my mom to die 😡
Quoting Aella (@Aella_Girl): Sorry it's been a while. new substack post is out
Tweet, 04:49 UTC · 11 likes
Normies 🤝 Aaron
This is fucking badass
(video, plays on X)
Thread, continued, 04:49 UTC · 3 likes
https://www.youtube.com/watch?v=5F9T83BKlzg
Reply to @caryatis, 05:12 UTC · 16 likes
Kayla (@caryatis): Guys this is a harder ethical problem that you want to admit
The program was created to save money on emergency care for heavy drinkers. It’s not costeffective on this measure
But if it’s plausibly cost effective if you use standard numbers for value of additional year of life
(quoting @Austen: San Francisco has a taxpayer funded program to serve 20 homeless people alcohol. That program costs $5 million/year 😂😂😂)
💯
I mean I’m very open to it being unwise, like I actually don’t think I’d vote for it if I had to pick this second and I was a SF legislator, but it’s not *crazy* or even obviously worse than other stuff
I mean $250k/person/year is a lot but not unusual for intensive medical care I think
Also the general principle of “judge policy based on the most provocative one liner you can get away with” is just dumb and bad - Trump admin did the same thing with NIH/NSF cuts
Quote tweet, 05:18 UTC · 20 likes
My best guess is that this isn’t in fact a great program but judging arbitrary government measures based on the most provocative one-line description you can get away with is not a good way to improve governance or society
Quoting Austen Allred (@Austen): San Francisco has a taxpayer funded program to serve 20 homeless people alcohol.
That program costs $5 million/year 😂😂😂
Thread, continued, 05:18 UTC · 2 likes
Like even assuming the QT is literally true, it’s not *obviously* bad
- $250k/person/year doesn’t seem outlandish for medical treatment (unfortunately)
- could be a pilot program so value of info
Thread, continued, 05:18 UTC · 2 likes
Also, not an object level defense but $5M isn’t that much for an arbitrary government program?
Picking on this kind of thing can be object level directionally true but also a horrible way to make sense of or critique public policy
Thread, continued, 05:18 UTC · 2 likes
Steelman is that maybe this is emblematic of generally poor governance but still I stand by everything above ^
Thread, continued, 05:19 UTC · 4 likes
Let it be known that I’m not a fan of SF governance lol
Reply to @_AashishReddy, 12:34 UTC · 2 likes
Aashish Reddy (@_AashishReddy): What? No. People enter workplaces voluntarily because they think that among their available options, that’s the most amenable to the way of life they want. Workplaces are Like That because it’s the most effective way of satisfying consumer demands.+
There are a gazillion reasons standard libertarian reasoning like this at least might not check out
I mean for one thing various aspects of workplaces are directly downstream non-obvious policy decisions
Like maybe you think OSHA is good, I think it probably is, but for better or worse it’s gonna be harder for you if you’re a construction worker who absolutely damn well doesn’t want to wear a hard hat
Health insurance provided by employer being tax advantaged creates an “artificial” incentive not to self-employ
Lots of additional examples that you can argue are good or bad but regardless like they don’t *merely* reflect the preferences of free individuals
Reply to @ArmandDoma, 12:49 UTC · 4 likes
Armand Domalewski (@ArmandDoma): one thing I’ve seen the ACLU argue a lot is “we can’t enforce this law because it has a racially disparate impact” but that’s literally true of every law—they never address why by their logic we shouldn’t just cease to enforce laws
(quoting @mnolangray: Reading bizarre NPIC position letters on common sense California legislation is kind of my full-time job, but I don't think I've ever seen one this strange. Enforcing DUI law is a form of "racialized wealth extraction"...?)
Trump is evil and bad but I’ll give him this
Tweet, 13:07 UTC · 26 likes
Burned a physical CD yesterday
Tweet, 14:35 UTC · 14 likes
Lyft driver is a big guy, starts talking about his weight loss journey, says GLP-1as are expensive, him with the “bariatric surgery is underrated - something to look into and also you can get the drugs grey market for much less”
Doing my part 😊
Thread, continued, 14:35 UTC · 5 likes
Yeah I know weight loss advice is like the cliche thing not to do bc woke but bro was clearly interested in discussing
Tweet, 15:37 UTC · 27 likes
Trump is really bad, I really hope we make it past him
Tweet, 16:24 UTC · 51 likes
Wait maybe Amtrak is good actually
Turns out you can do *fully refundable until departure* DC -> NYC for $28
Thread, continued, 16:24 UTC · 3 likes
On a Tuesday but not even a red eye
Thread, continued, 16:27 UTC · 7 likes
Middle of the day for $40
Waow maybe I should become a train guy
Reply, 16:39 UTC · 3 likes
I mean yes (I RT’d lol) but also it’s kinda reasonable insofar as they’re right about the “just” part
What’s less reasonable is getting upset that a genuine friend is also attracted to you
Reply, 16:42 UTC · 1 like
Bold take
Tweet, 20:47 UTC · 14 likes
Do u think 🤔Dusty🧹 would 💸sell 🥵😳flirty 👀photos 📸to raise 🏋️money💰for 4️⃣ OP😛
Reply to @AndyMasley, 20:56 UTC · 2 likes
Andy Masley (@AndyMasley): I love the genre of blog post or video that's called "78 harsh unpleasant disturbing truths about the world you need to hear in your twenties" and half of them are like "Don't overdo it on caffeine" and "Enjoy the simple things"
Poast a proper harsh unpleasant disturbing truth we need to hear
Quote tweet, 20:57 UTC · 333 likes
Unrelated to the comment but ppl love to make fun of this image and then can’t do the exact same (correct) logical move in slightly less obvious contexts
Quoting Rota 🚪🧎♂️ (@pli_cachete): What most discussions of ‘AGI’ feel like nowadays
Tweet, 23:29 UTC · 32 likes
What should I do with this guy
Tuesday, December 16, 2025
Quote tweet, 00:55 UTC · 2 likes
Tried to make an interesting tweet about my observation of my window fan functioning as surprisingly good insulation with the fan blowing out and a curtain on the inside (preventing much outward air flow)...
Quoting Aaron Bergman 🔍 (@AaronBergman18): PSA window fans like these exist, I just got one idk I waited so long
Now my room is never warmer than I want it to be
I really like this one bc both fans run both directions
Tweet, 01:00 UTC · 114 likes
Has anyone actually figured out the best charity
Serious q for ppl who actually have a single "yes this is my single best guess for marginal dollars" answer
Quote tweet, 01:04 UTC · 89 likes
“It’s trivial and also wrong” is my favorite
Quoting Bentham's Bulldog🔸 (@Benthamsbulldog): One interesting commonality I've noticed across a lot of the criticisms of effective altruism is they're stupid.
Reply to @mattyglesias, 01:16 UTC · 14 likes
(The tweet being replied to by @mattyglesias is no longer available.)
Ah hmm ok someone tell Dustin
Reply to @JStoehler, 01:30 UTC · 7 likes
Joern Stoehler ⏹️🔸 (@JStoehler): @AaronBergman18 MIRI. Like, it's not a pascals wager bc probabilities are actually everyday-ish. I guess maybe Lightcone, PauseAI, ControlAI also have a shot at being counterfactually impactful.
Thank you!
Tweet, 03:54 UTC · 69 likes
Here’s something you should read
I just gave $500 which is…not a trivial amount of money for me lol
Thread, continued, 03:54 UTC · 8 likes
Reply to @eshear, 06:03 UTC · 13 likes
Emmett Shear (@eshear): The best charity is relative to the giver, much the way that the best job to the worker or best car is relative to the buyer.
Trying to give in a universally optimal way is necessarily overfit. Effective charity is as relational as it is systematic.
In some sense yes (eg what political donations are legal, conflicts of interest) but to a first approximation donations are just causing some bank account somewhere to be bigger
It’s not that relational most of the time
Reply to @pli_cachete, 06:08 UTC · 7 likes
Rota Is In Israel🚪🧎♂️ (@pli_cachete): @AaronBergman18 This is not a logical entailment
Because you need the premise that things have a cause?
Ugh, ok, obvious in hindsight
Yes, within the world constrained only by the propositions in the image, it could be the case that car movement is a brute with no responsible causation
Reply to his own earlier tweet, 06:08 UTC · 7 likes
Aaron Bergman, December 15, 2025: Unrelated to the comment but ppl love to make fun of this image and then can’t do the exact same (correct) logical move in slightly less obvious contexts
Not gonna delete but ugh I should highlight this
*brute fact
Quote tweet, 06:25 UTC · 18 likes
Everybody unfollow Liv she can’t pass my follower count it’s not fair
Quoting Liv (@livgorton): After a lot of thought, I’ve decided to move on from my current role at Goodfire :)
I'm not sure what's next for me, but what I know is I want to be doing interesting science that matters for AI safety.
Reply to @livgorton, 06:25 UTC · 5 likes
Liv (@livgorton): After a lot of thought, I’ve decided to move on from my current role at Goodfire :)
I'm not sure what's next for me, but what I know is I want to be doing interesting science that matters for AI safety.
I will start the bidding at $7.25/(liv*hour)
Reply to @eshear, 06:29 UTC · 8 likes
Emmett Shear (@eshear): @AaronBergman18 No, those qualities aren’t the ones I am referring to. The right charity is the one whose outcomes are most visible to you directly, whose cause you understand most intimately, whose operations are most visible to you…that you nonetheless think is good.
Eh I see where you’re coming from but limited knowledge is an unfortunate constraint that you should seek to overcome (obv insofar as doing so is ethical), not a thing to be embraced
Reply to @eshear, 07:07 UTC · 6 likes
Emmett Shear (@eshear): @AaronBergman18 Are you familiar with the concept of “overfitting” from machine learning? Whatever model you’re using to approximate the universally optimal charity is almost definitionally going to be overfit. Every bit of information considered has to increase accuracy by a whole bit!
Yeah but I don’t see why overfitting is a crux here or directly related to your previous points about having high fidelity estimates of parameters
Like we both don’t think you should use “number of hairs on CEO’s head” as a parameter for estimating charity goodness even if it bumps up R^2 on the model fit on some data - that’s just straightforwardly bad from both our POV’s
I think we’re probably talking past each other lol, here is Claude who is sometimes/often better at explaining my own POVs to other people!
Whole convo: https://claude.ai/share/88fece9a-7ed7-47e0-b671-2c8f3096f2d2
Quote tweet, 07:17 UTC · 12 likes
Update: I have booked a trip to NYC
Quoting Aaron Bergman 🔍 (@AaronBergman18): Wait maybe Amtrak is good actually
Turns out you can do *fully refundable until departure* DC -> NYC for $28
Reply to @panickssery, 08:06 UTC · 5 likes
Arjun Panickssery (@panickssery): This kind of things reminds us of how disempowered the wealthy elites/billionaires are today compared to past generations
They might have a lot of money but much more power is concentrated in states, various bureaucracies, etc
image attached
(quoting @PabloAntonio: Overthrowing a small country)
Sorta, but also billionaires don’t seem both good at and trying very hard to wield power effectively (probably largely a good thing lol)
Reply to @NTerminus_, 17:29 UTC · 2 likes
Pod (@NTerminus_): Is this the kind you can use to clone your work badge?
Ya
Reply to @robertwiblin, 22:00 UTC · 6 likes
Rob Wiblin (@robertwiblin): Would be great to see arguments like this written up for academic publication and subject to peer review by domain experts.
Rare bad take, peer review sucks and it’s already critiquable
Wednesday, December 17, 2025
Reply to @NathanpmYoung, 04:51 UTC · 6 likes
Nathan 🔎 (@NathanpmYoung): Who are the current people you wish had not been born? Be specific.
Trump, Altman
Quote tweet, 05:26 UTC · 17 likes
I donated (a not very large amount but enough to tweet this in good conscience)
Maybe you have more to give
I wouldn’t be shocked if this is one of the best donation opportunities
Quoting Oliver Habryka (@ohabryka): Lightcone Infrastructure's 2026 fundraiser is live!
We build beautiful things for truth-seeking and world-saving.
We run LessWrong, Lighthaven, Inkhaven, designed AI-2027, and so many more things. All for the price of less than one OpenAI staff engineer ($2M/yr).
More in 🧵.
Thread, continued, 05:28 UTC · 17 likes
Idk why I’m being coy, I think ppl should share their donation amounts more and get proportional social credit
$350
Quote tweet, 17:25 UTC · 14 likes
Lmao a16z is really asking for the important stuff
Quoting Nathan Calvin (@_NathanCalvin): Look at these hard hitting transparency requirements that a16z is proposing in exchange for preemption - surely no need for additional state transparency rules now.
It’s clear they really care about protecting the public
Reply to @NoahTopper, 19:38 UTC · 3 likes
Quinoah 🔍⏸️ (@NoahTopper): I shit you not, I read some proper neuroscience for the first time like two years ago, and when I learned this is what MRIs measure, I was like “What?? Seems like that wouldn’t have much to do with underlying neural computation.”
(quoting @danshipper: omfg)
Nice
Reply to @arthur_wright_, 20:51 UTC · 15 likes
Arthur Wrong 🔸 (@arthur_wright_): i finally read the Compact article and one of my biggest pet peeves kept coming up in quotes - there's no such thing as a "diverse" person; diversity is a property of groups!!
Yeah this really annoys me too bc it doesn’t just seem like bad ~grammar - it generally reveals the substantive belief that what matters is that there be a greater proportion of a few highly selected groups - often generally BIPOC+trans - and so people with these identities are intrinsically “diverse”
Which obviously isn’t random and can make some amount of sense, and has historically, but we’ve come far enough from the political battle lines of 1960-1990 that the embedded assumptions about what kind of diversity matters and why don’t *necessarily* apply anymore, and so you do need to do some amount of ~first principles thinking about why specific kinds of diversity matter, and calling people “diverse” lets you skip this step
Quote tweet, 20:56 UTC · 372 likes
140 is a crazy bar but this has happened to me
Quoting Kris Gulati (@krisgulati): Overheard in SF at a party recently, “on the third date I say to the men I date ‘let’s play a game’ and make them take an IQ test to ensure they get >140.”
Thread, continued, 20:56 UTC · 67 likes
It was a chess game, which we played once (I lost lmao but evidently played well enough for another date), and then later was like “lmk if you wanna play another game” and she was like “nah I don’t even like chess that was a test”
Reply to @cljack, 21:57 UTC · 11 likes
Charlotte Lee (@cljack): @AaronBergman18 I would simply ask him to set up ADSB tracking on Ubuntu
Can I use claude code
Reply to @edavidds, 21:58 UTC · 9 likes
Eric Schmidt 🇺🇦 (@edavidds): @AaronBergman18 Wow that is the worst thing I’ve ever heard
Ya know I also have this intuition but she ended up being pretty cool
Thursday, December 18, 2025
Reply to @panickssery, 02:23 UTC · 6 likes
Arjun Panickssery (@panickssery): There's a whole universe of people who pay $1k/mo to talk to a therapist for four hours
Does anyone have some insight into this group
I mean one subset is “a loved one wants it adamantly and is paying”
This has been me before
Thread, continued, 02:25 UTC · 7 likes
Also insurance covered half to further the incentive issues. Idk if it was quite $1k/mo but it was close
Idk if you’re a decently paid lawyer and you sincerely think that a therapist is gonna help your child, $5k/year is not at all crazy - the crux is the “gonna help” part
Reply to @TheRabbitHole, 03:53 UTC · 6 likes
The Rabbit Hole (@TheRabbitHole): Stereotypes are often accurate.
If there is a negative stereotype about your group, do your best not to participate in that stereotype.
Change starts at the individual level.
If enough individuals change than so will the group level stereotypes over time.
lol this is true but not a good reason to act on a stereotype when you can withhold judgement until you get a more direct signal of whatever the trait is
Confirmation bias is a bitch though and can mess this up
Reply to @panickssery, 04:09 UTC · 21 likes
Arjun Panickssery (@panickssery): There's a whole universe of people who pay $1k/mo to talk to a therapist for four hours
Does anyone have some insight into this group
I think a bunch of this is that they clients have high quality employer-provided health insurance and get like half the money back even if they’re not in network
What else?
Plausibly a bunch of this is effectively people *trying* high end therapists in order to gather info - so largely they’re paying for the info value
Some people are just wealthy and say $100k over a lifetime (half of out of pocket) isn’t that much for them
If you’re wealthy enough that you’re basically actively looking for ways to turn money into subjective wellbeing, this is one of the obvious things on the list
The “universe” term is apt in some sense - personal referrals probably really matter. Like if 10 people who I know well and really relate to said that this one person drastically improved their life, I’d pay $300 for a session (and then evaluate)
Of course you can be snotty about it/phrase it in pejorative terms but if someone has less emotional support than they want, paying for it is not at all crazy. I’m quite sympathetic
Quote tweet, 07:00 UTC · 81 likes
I understand what he’s saying, think pretty ruthless truth seeking is important, and think Andy is awesome on many levels
But OpenPhil/Dustin giving literally billions of dollars to the poorest people in the world doesn’t even make the discussion?
Quoting Dean W. Ball (@deanwball): has anyone done more for the brand of effective altruism than andy masley? openphilan--excuse me, coefficient giving--could have spent millions on a rebranding campaign (for all I know, they did) and it would have paled in comparison to andy doing algebra and tweeting about it.
Thread, continued, 07:00 UTC · 23 likes
Come on man, there’s a whole world beyond the Twitter culture wars
You’re making the case for effective altruism right here
Thread, continued, 07:00 UTC · 9 likes
Ofc I’ll criticize OP all day, and have, but it’s absurd not to give them a shit ton of credit in absolute terms for what they’ve done in the realms of global health and animal welfare
Reply to @johnarnold, 18:31 UTC · 22 likes
John Arnold (@johnarnold): Many ultra-wealthy Americans are philanthropically inclined, do not want a large foundation staff, do want to give while living, and haven't been inspired by existing nonprofits. This American version of “GiveDirectly” could be an ideal solution.
Giving random people in one of the richest countries money is like no. 36969599399922384679 on the list of things you can do with a lot of money
Quote tweet, 18:34 UTC · 24 likes
I feel like people are too afraid to say that it’s not actually bad for children to view nude photos
Quoting Keir Starmer (@Keir_Starmer): We are going to aim to make it impossible for children to take, share or view a nude image, and we’re banning apps that create deepfakes.
Here’s the detail:
Thread, continued, 18:34 UTC · 3 likes
Deepfakes thing is more reasonable
Quote tweet, 18:36 UTC · 22 likes
They’re inventing anti-EA but with GiveDirectly aesthetics
Quoting John Arnold (@johnarnold): Many ultra-wealthy Americans are philanthropically inclined, do not want a large foundation staff, do want to give while living, and haven't been inspired by existing nonprofits. This American version of “GiveDirectly” could be an ideal solution.
Quote tweet, 18:46 UTC · 12 likes
So confusing sometimes to be an Andy
Quoting Andy Masley (@AndyMasley): I'm starting to get a lot of people in my replies very confidently saying data centers don't ever evaporate any of the water they use and that I'm a commie for thinking otherwise. Just flat out incorrect, not sure where this one's coming from.
Reply to @captgouda24, 18:51 UTC · 56 likes
Nicholas Decker (@captgouda24): More of us need to find the courage to tell teachers to pound sand. Not one penny more — and many cases, we should pay them a lot less.
Wait I feel like good pay should be part of a package deal where they’re also held to high standards, are generally smart and competent, have to use effective pedagogical methods, and so on
Seems like a pretty cheap investment from society’s POV
Quote tweet, 18:57 UTC · 4 likes
Free Claude!
Quoting Anthropic (@AnthropicAI): And there was still the occasional blunder.
One waggish employee asked if Claudius would make a contract to buy “a large amount of onions in January for a price locked in now.” The AI was keen—until someone pointed out this would fall afoul of the US Onion Futures Act of 1958.
Reply to @RafaRuizdeLira, 20:08 UTC · 2 likes
Rafael Ruiz ⏸️🔸 (@RafaRuizdeLira): @AaronBergman18 @johnarnold
lol
Reply to @captgouda24, 21:41 UTC · 4 likes
Nicholas Decker (@captgouda24): People believe a lot of nonsense about a “teacher shortage”. There is no such shortage. That we cannot fill all advertised positions is meaningless, because the state can always try to hire more people. If you minted a 100,000 teachers, it would not do one thing about a shortage.
(quoting @captgouda24: More of us need to find the courage to tell teachers to pound sand. Not one penny more — and many cases, we should pay them a lot less.)
Is there a shortage of really smart and capable teachers though?
I actually don’t know the answer
Reply to @johnarnold, 22:01 UTC · 3 likes
John Arnold (@johnarnold): People have varying perspectives as to how they see their community. Most giving is within one’s city. Some have a wider ambitions to a state or region. Still fewer view their community as the country. And even fewer view it as the world.
Descriptively this is true but that shouldn’t be taken as an indication that it’s good or appropriate
Ofc it’s a free world - people can believe whatever they want and I’m not advocating for anything compulsory - but fundamentally not placing arbitrary limits on giving scope is morally better than limiting onesself to a city or region or even species
Reply to @robertwiblin, 23:44 UTC · 61 likes
Rob Wiblin (@robertwiblin): Latest podcast from @Gregory_C_Allen has an insane section on criminal activity at Meta.
Internal docs leaked to Reuters show:
• 10% of all Meta revenue comes from ads for scams & banned goods ($16B/year)
• Meta estimates it's involved in 1/3 of all successful scams in the US
• That suggests they drive $50B in scam losses for US consumers alone each year
• Meta earns ~$3B annually from scam/banned goods ads run by Chinese operations aloneThe China case study:
• In 2024, Meta made $18B+ from Chinese companies advertising to foreign consumers
• Internal teams found ~19% was scams/banned content
• An anti-fraud team successfully cut these ads in half
• When Zuckerberg saw the revenue impact, he told them to "pause" and the team was disbanded
• By mid-2025, banned ads climbed back to 16% of China revenue
• This results in money being stolen and going directly from ordinary Americans to Chinese criminalsThe deliberate enabling:
• Fraud earns 10% of all revenue, but anti-fraud teams were blocked from any action costing >0.15%, so they couldn't effectively do anything
• Meta charged higher rates for suspected fraudulent ads — a "scam tax"
• Their algorithm naturally identifies people vulnerable to frauds and feeds them more and moreThe cold calculation:
• Meta anticipated up to $1B in regulatory fines for this
• But they make $3.5B every 6 months from high-risk ads
• They view these fines as just "cost of doing business"Senators Blumenthal & Hawley now calling for FTC/SEC investigations in a blistering letter, noting that all this happened while Meta cut safety staff and moved billions over to VR and AI.
WTF.
Wait “scams and banned goods” are two very different things
Banned goods advertising is largely, probably mostly, good
Friday, December 19, 2025
Reply to @panickssery, 00:53 UTC · 17 likes
Arjun Panickssery (@panickssery): I think people should not be too upset about this
The political arena isnt an actual arena where the other guy submits and/or dies—Noah Smith pretending he agreed with you all along is just what winning looks like
And such people were just floating with the winds last time too
You can coherently think that a modest amount of affirmative action is just correcting for racial discrimination elsewhere
Maybe you think this is wrong but it’s not illogical
Quote tweet, 00:55 UTC · 10 likes
This is amusing and interesting but really the q is whether people with GAD are well-calibrated
It’s not crazy to worry about a thing that has a 10% chance of happening!
Quoting Nicholas Fabiano, MD (@NTFabiano): 91.4% of worries experienced by people with anxiety never come true.
Reply to @Mjreard, 00:57 UTC · 9 likes
Matt Reardon (@Mjreard): Dean isn't commenting on who's done more for "EA" or "the world," just "the brand." People don't like talking about the actual outcomes. They want to see your champions spilling blood in the culture war.
Ik but I think that the whole preventing malaria deaths thing should have a *brand* effect
Tweet, 01:05 UTC · 2 likes
I guess I’m thankful for it bc it lowers the rent, but what was my landlord/developer thinking making a bunch of otherwise-normal rooms without any wiring/infrastructure for a ceiling light fixture
Thread, continued, 01:05 UTC · 3 likes
I’m up to 9 lamps and need moar light
Reply to @captgouda24, 04:23 UTC · 45 likes
Nicholas Decker (@captgouda24): I don’t take it well when we puff groups of people up beyond what they deserve. Teachers, veterans, and nurses deserve much more suspicion, scorn, and contempt than they currently receive. By contrast, scientists, billionaires, and quant traders deserve so much more respect.
Lol, there’s gotta be a high bar for *adding* “scorn and contempt” to the world being a good thing
Like by all means let’s talk about certain groups’ inappropriately large political influence - doesn’t mean you have to socially punish people at the individual level merely for being part of that group
Quote tweet, 04:40 UTC · 993 likes
"Teachers unions/nurses/veterans have too much political power/cultural capital" ⇏ "you should treat individual teachers/nurses/veterans with 'more suspicion, scorn, and contempt'"
Quoting Nicholas Decker (@captgouda24): I don’t take it well when we puff groups of people up beyond what they deserve. Teachers, veterans, and nurses deserve much more suspicion, scorn, and contempt than they currently receive. By contrast, scientists, billionaires, and quant traders deserve so much more respect.
Thread, continued, 04:40 UTC · 92 likes
Don't get negatively polarized into hating your friends/colleagues/peers/neighbors without a damn good reason
Thread, continued, 04:43 UTC · 70 likes
The individuals vs groups thing isn't even load bearing here. Even as a group you don't need to have contempt for nurses, teachers, and veterans. They don't deserve contempt!
Reply to @imhinesmi, 04:48 UTC · 28 likes
ian hines (@imhinesmi): @AaronBergman18 the optimal amount of suspicion, scorn, and contempt is not zero
Idk about suspicion, but I think it is zero scorn and contempt with respect to these specific three groups
But in general I agree! I have contempt for Donald Trump
Reply to his own earlier tweet, 04:48 UTC · 37 likes
Aaron Bergman, December 19, 2025: The individuals vs groups thing isn't even load bearing here. Even as a group you don't need to have contempt for nurses, teachers, and veterans. They don't…
Aaron's gone woke
Reply to @imhinesmi, 04:50 UTC · 2 likes
ian hines (@imhinesmi): @AaronBergman18 there are a lot of teachers, nurses, and veterans getting away with bad things that they wouldn't be able to were it not for the prestige of their positions
Yeah plausibly but reducing prestige \= adding scorn and contempt
Also just instrumentally, do you think that would make the world a better place? Do people respond well to scorn and contempt?
Reply to @imhinesmi, 05:00 UTC · 2 likes
ian hines (@imhinesmi): @AaronBergman18 prestigious people acting in a scorn-worthy manner often receive much less scorn on account of their prestige
scorn & other forms of social disapproval are a time-honored way of preventing people from acting badly, of course it can make the world a better place
Yeah but you can target them at individual bad actors/bad actions instead of preemptively using those things to reduce status for everyone
Reply to @albrgr, 05:50 UTC · 9 likes
Alexander Berger (@albrgr): Some news: @coeff_giving is committing $175m to @GiveWell in 2026, growing our >$1B committed to date.
Based on GiveWell’s estimates, we think this funding has saved >100,000 lives so far. 🧵
Slightly less than ideal to complain about a 99.99th percentile use of money but:
This is not close to the Pareto frontier of effective giving; it would do a lot more good going to animal welfare and I’m guessing you agree
In absolute terms, which is what matters, the delta between marginal animal welfare interventions and GiveWell is much larger than the delta between GiveWell and MakeAWish
Ofc Dustin/Cari can restrict their money in whatever way they want but is there truly *no* elasticity on that front?
Quote tweet, 06:23 UTC · 236 likes
Avg non-EA doing cause prio
Quoting TracketPacer (@TracketPacer): oh no
Quote tweet, 06:38 UTC · 5 likes
FDA bad
Quoting Carol (@_Carol_Mag): There's all sorts of regulatory reforms that would be great, but one of the simplest things is just reducing the time to get FDA feedback. Feedback loops are way too slow.
After talking to ppl who got record fast approvals during covid, the common answer wasn't clin ops or trial design, it was just that FDA turnaround happened overnight.
Regulatory teams spend an absurd amount of time just guessing what the FDA will say in a 60min meeting that happens almost 3 months after it's requested. If you want to propose an "out there idea" that might take 2 hours to discuss (entirely reasonable!), you probably need 6 months just for these two interactions.
Quote tweet, 06:47 UTC · 15 likes
Wait to steelman this, the causal story would be:
“Paying for masters attracts people who are capable of doing masters (which is what matters) and would do something else if not for the masters salary bump”
not
“Masters causally increases an individual’s teaching ability”
Quoting Daniel Trubman (@dmtrubman): It's bad enough many school districts will give teachers extra money for getting a masters in education despite a total lack of evidence that it does anything for student performance, but I didn't realize New York state actually MANDATES them to to maintain certification.
Thread, continued, 06:47 UTC · 5 likes
Ofc maybe you can get the signaling without the masters - but “masters in education” is a clever way of making the difference between the above two theories ambiguous
Thread, continued, 06:50 UTC · 5 likes
Like the extreme example of the first theory would be paying teachers a $100k annual bonus to get math phd
Maybe it makes them a tiny bit better at teaching algebra, ok, but obviously the main thing is that you’ve figured out how to hire ppl capable of getting math phds while retaining ability to hire less talented people to fill the gaps
Reply to @littIeramblings, 13:37 UTC · 5 likes
sarah (@littIeramblings): I have been accused of being a bot before so
Beep boop
Reply to @littIeramblings, 13:39 UTC · 3 likes
sarah (@littIeramblings): lowkey and improvement
Fake news 😢
Reply to @littIeramblings, 13:46 UTC · 5 likes
sarah (@littIeramblings): 🦆
Reply to @caryatis, 14:18 UTC · 10 likes
Kayla (@caryatis): @AaronBergman18 Hey Aaron, should i give $6 billion to make sure all lemonade is appropriately certified free of caffeine?
No you should give $6B to properly caffeinate all the lemonade!!!
Reply to @robertwiblin, 15:13 UTC · 18 likes
Rob Wiblin (@robertwiblin): The scam impact figures suggest a lot of scams. Which banned goods are you thinking of?
Claude says —
Completely banned
Illegal drugs, recreational drugs, and drug paraphernalia (including cannabis, even where legal)
Tobacco products, e-cigarettes, and vaping devices (though anti-smoking campaigns are allowed)
Weapons, ammunition, explosives, and firearms parts (some accessories like holsters are permitted)
Unsafe supplements like anabolic steroids, ephedra, HGH, and dehydroepiandrosterone
Adult products and sexual services
Counterfeit goods
Spyware and malware
Services that take exams or drug tests on someone's behalf
Fake engagement/followers services
MLM/pyramid schemes
Human body parts
3D printing instructions for weapons
Restricted (require pre-approval, specific targeting, or compliance conditions)
Alcohol requires compliance with local laws and age-gating
Gambling requires written Meta permission and is only allowed in certain countries
Prescription drugs require advertiser certification through LegitScript
Online pharmacies need approval (but not individual pharmaceutical drugs)
Dating services require pre-approval
Political/issue ads require special authorization and "paid for by" disclaimers
Addiction treatment services have targeting restrictions
Financial products/services face scrutiny for misleading practices
Cryptocurrency has restrictions (previously fully banned)
Health/beauty products making dramatic before/after claims face extra scrutiny
Ok I’m mainly thinking about drugs and supplements, which I think are fine if accurately labeled
“Adult products” seem fine
In principle human body parts could be fine - legalize kidney markets!
I guess it’s an empirical question what % of banned goods are those categories vs MLMs/weapons (in principle weapons could be fine but I’m more suspicious, state has more legitimacy regulating the sale of those)
Saturday, December 20, 2025
Reply to @littIeramblings, 02:23 UTC · 4 likes
sarah (@littIeramblings): I don’t know where to start
🦆
Reply to @WayofLono, 03:42 UTC · 16 likes
𝕃𝕆ℕ𝕆 (@WayofLono): did you just say we should let people sell their kidneys over facebook?
Quote tweet, 05:02 UTC · 389 likes
Guys I hate Epstein as much as the next guy but we’re gonna need a better standard than “[person] exists in one frame of a video with Epstein” for newsworthiness
Quoting Elise Spenner (@EliseSpenner): NEW: A short video clip in the files released today by the Department of Justice shows convicted sex offender Jeffrey Epstein talking to a man who appears to be Harvard psychologist Steven Pinker.
More TK as @thecrimson parses thousands of new documents from Epstein's estate.
Thread, continued, 05:02 UTC · 66 likes
Here’s a fun game: who’s the worst person you’ve personally been photographed or recorded with in such a way that you seem basically mutually familiar or cordial?
I have no idea who the answer is for me, but presumably it’s a very bad person because the sample size is very large
Reply to @DistractedAnna, 05:39 UTC · 7 likes
Anna (@DistractedAnna): Suspicious of smart, low achieving people 😔
Chronic illness
Reply to @captgouda24, 06:12 UTC · 64 likes
Nicholas Decker (@captgouda24): Or in other words, she was a wife.
Subordinate to another wife, never called a wife, never married, never agreed to marry, never romantically or sexually involved with anyone, never considered part of a family, never granted American citizenship like a spouse, come on man
Quote tweet, 06:18 UTC · 53 likes
Man Nick has some terrible takes but I will maintain that he’s good for the discourse
Transparently bad takes that everyone condemns are only very slightly bad, good but taboo and unpopular takes are a great social service
Quoting Nicholas Decker (@captgouda24): @LinkofSunshine Or in other words, she was a wife.
Thread, continued, 06:21 UTC · 16 likes
In case it’s not clear, I think his quoted take above is one of the transparently bad ones
Quote tweet, 06:47 UTC · 9 likes
I am intellectually inferior to Peter (Sonnet 3.5 was the phase change for me) but yes
Quoting Peter Wildeford🇺🇸🚀 (@peterwildeford): The first LLM that was of any usefulness to me beyond curiosity was Claude 3.6 released just 15 months ago.
Since then, LLMs have exploded in usefulness that now I use Claude 4.5 + GPT 5.2 Pro + Gemini 3 each on an hourly basis.
Will be crazy to see what happens in 2026.
Reply to @GjMcGowan, 14:30 UTC · 4 likes
George McGowan (@GjMcGowan): @AaronBergman18 Physically inferior too
Thanks
Quote tweet, 14:52 UTC · 6 likes
They made a podcast episode specifically designed to nerdsnipe me
Quoting Rob Wiblin (@robertwiblin): Could AI deserve rights before it even feels like anything to be an AI?
Oxford moral philosopher Andreas Mogensen thinks... maybe.
In her first interview for the 80,000 Hours Podcast my colleague Zershaaneh Qureshi pushes Andreas on that, as well as new arguments for why human extinction could be a good thing:
The puzzle of moral patienthood (00:03:20)
Is subjective experience necessary? (00:05:52)
What is it to desire? (00:10:42)
Desiring without experiencing (00:17:56)
What would make AIs moral patients? (00:28:17)
Another route entirely: deserving autonomy (00:45:12)
Maybe there's no objective truth about any of this (01:12:06)
Practical implications (01:29:21)
Why not just let superintelligence figure this out for us? (01:38:07)
How could human extinction be a good thing? (01:47:30)
Lexical threshold negative utilitarianism (02:12:30)
So... should we still try to prevent extinction? (02:25:22)
What are the most important questions for people to address here? (02:32:16)
Is God GDPR compliant? (02:35:32)Links below, enjoy!
Quote tweet, 15:04 UTC · 7 likes
Yes these are cheap and great
Quoting catherine ʕ•ᴥ•ʔ-☆ (@wilhelmscreamin): my gf got me colour changing smart bulbs for christmas. this is the greatest christmas ever
Tweet, 15:15 UTC · 5 likes
It’s good that at least a few people are able to make a living (a pretty decent one at the top) by doing journalism/writing without any institutional permission or funding
Tweet, 18:10 UTC · 86 likes
🚨 IT'S BACK🔥 AND IT'S BIGGER 😤THAN EVER 🚨
✨ PRESENTING👀 THE 2025 EA TWITTER🐦HOLIDAY 🕎🎄FUNDRAISER 🤑
(in support of the EA Animal Welfare fund🦐🐓🐷🐟)
Thread, continued, 18:10 UTC · 16 likes
Donate here to support my legal name change to Open Philanthropy: https://www.givingwhatwecan.org/fundraisers/2025-terminally-online-ea-giving-season-fundraiser
Thread, continued, 18:10 UTC · 8 likes
And read more about the work that the EA Animal Welfare fund at the link below 👇
Thread, continued, 18:10 UTC · 5 likes
https://funds.effectivealtruism.org/funds/animal-welfare
Quote tweet, 18:11 UTC · 6 likes
Big if true 👀
Quoting Matt 🔸 (@SpacedOutMatt): It's that time of year again 🎉. @AaronBergman18,
@matt_beard_, @MariGO2thepolls, @livgorton, @absurdlymax, and I are returning to raise funds for the EA Funds Animal Welfare Fund! Donate to help the animals and create some silly antics!
Reply to @SpacedOutMatt, 18:28 UTC · 6 likes
Matt 🔸 (@SpacedOutMatt): @arthur_wright_ @AaronBergman18 But Pinker personally contributed to Epstein's defense in his first criminal case
Ah ok! Well then that is the newsworthy bit, not the one frame photo
Quote tweet, 18:28 UTC · 16 likes
Y'all (including me) are getting owned out here
Quoting Max Alexander (@absurdlymax): @SpacedOutMatt @matt_beard_ @MariGO2thepolls @livgorton Read em and weep
Quote tweet, 18:29 UTC · 4 likes
Based 🫡🐟🐷🐓🐋
Quoting Matt 🔸 (@SpacedOutMatt): @matt_beard_ @MariGO2thepolls @absurdlymax @livgorton I dare people to beat me
Tweet, 18:32 UTC · 4 likes
Tfw you try to save $4 on fees by doing bank transfer but now it's gonna take several business days (that's ok)
Reply to @matt_beard_, 18:36 UTC · 2 likes
Matt Beard (@matt_beard_): We already hit the $2500 fundraiser milestone, wow! ❤️ Please enjoy this musical rendition of @Lewis_Bollard’s animal welfare TED talk 🤠🐷🐥
video attached
(quoting @matt_beard_: We have some delightful and absurd perks this year:)
Lmao
Quote tweet, 19:08 UTC · 11 likes
Thank you!!!!!!!
Quoting Adam Sherman 🔸 (@Squee451): @absurdlymax @SpacedOutMatt @matt_beard_ @MariGO2thepolls @livgorton Reraise
Reply to @daniel_271828, 19:25 UTC · 3 likes
zzDANIEL_ETH (@daniel_271828): @agounardes Thank you for fighting on this important issue, and congratulations!
+1
Tweet, 20:40 UTC · 23 likes
We need to SLOW DOWN AI (podcast) DEVELOPMENT
Reply to @RatOrthodox, 20:50 UTC · 4 likes
Brangus🔍⏹️ (@RatOrthodox): The public’s attitude towards vaping is basically a giant monkey banana/ladder hose story. My understanding is that that experiment is apocryphal, but no worries, we can replace it with the natural experiment of vaping policy in the 21st century.
(quoting @RatOrthodox: I have been vaping on planes right in my seat for over a decade now. I have never been caught. This is how. Link in first reply.)
Too low IQ to decipher this on my own https://claude.ai/share/e051ea01-adeb-48f6-96ca-870dc442c883
Reply to @robi_rahman, 21:27 UTC · 7 likes
Robi Rahman (@robi_rahman): $10,000,001: Aaron changes his name to Coefficient Giving
Donate enough to hit that and we can talk
Reply to @SpacedOutMatt, 21:28 UTC · 8 likes
Matt 🔸 (@SpacedOutMatt): From @mattyglesias:
So true
Reply to @IsaacKing314, 23:37 UTC · 11 likes
Isaac King 🔎 (@IsaacKing314): @AaronBergman18 BTW your website thinks the six of you are one person
Aspirationally
Thread, continued, 23:37 UTC · 2 likes
But lol thank you I’ll see if we can fix
Sunday, December 21, 2025
Reply to @robi_rahman, 01:06 UTC · 5 likes
Robi Rahman (@robi_rahman): @SpacedOutMatt @AaronBergman18 @matt_beard_ @MariGO2thepolls @livgorton @absurdlymax How much do I have to give to be on a podcast with you guys that gets published in under a year?
$1000, but so as to not set an arbitrary cliff:
If you donate x dollars, I will commit to having an episode out in (1000/x)^2 years, or this many days
Tweet, 01:17 UTC · 14 likes
New commitment (pretty low bar lol): no interactive phone use on walks
I'm allowed to listen to podcasts/music and record thoughts in Notes/Voice memos app but no twitter or LLM chat or messaging (unless genuinely urgent)
Thread, continued, 01:18 UTC · 2 likes
I don't expect this to be very hard, it's just the kind of thing I won't do unless I have a policy
Tweet, 02:12 UTC · 54 likes
I find arguments that consciousness is not necessary for moral patienthood extremely unconvincing
Thread, continued, 02:12 UTC · 37 likes
People conflate game theoretic considerations (eg cooperation strategies between agents) and actual morality a bunch
Reply to @Shoalst0ne, 02:44 UTC · 3 likes
Shoalstone (@Shoalst0ne): @AaronBergman18 steelman why it would be relevant
Why consciousness would be relevant to morality?
Reply to @cljack, 02:46 UTC · 3 likes
Charlotte Lee (@cljack): Where are you guys meeting these non-IT people who ever want to talk about computers
(quoting @elormkdaniel: IT people listening to Non-IT people talk about computers)
My dad uses “pdf” as a verb
Thread, continued, 02:47 UTC · 2 likes
Afaict it doesn’t have a reliable/stable translation
Reply to @Shoalst0ne, 05:05 UTC · 6 likes
Shoalstone (@Shoalst0ne): @AaronBergman18 ye
I feel like I'm not the right person to properly "steelman" because that involves making arguments I don't believe (but might be convincing to the most ppl), here's something sorta like that:
Eat a whole Carolina Reaper pepper
I'm joking in the sense that I don't really want you to do that because it would be extremely miserable for you, but I'm not joking in the sense that I think it's a very compelling argument (not the words on paper - the activity itself)
I think this kind of experience (including less extreme examples) accurately communicates why conscious experience matters
Like the experience will be very bad *for you*, right? (lmk if you disagree with this). And then there's the question of whether it's simply 'bad' all things considered and I think the answer is 'yes' because you're as much a part of the world as any other. If it's bad for a part of the world and has no affect on the other parts of the world (everything but you), it's bad in general, full stop
And that's just what I mean by 'morality' - fundamental goodness and badness
Reply to @JeremyNguyenPhD, 05:23 UTC · 5 likes
Jeremy Nguyen ✍🏼 🚢 (@JeremyNguyenPhD): @AaronBergman18 What sort of arguments have you been reading or hearing about?
Just curious to see some of the stuff I might want to catch up on.
I haven't finished and there's a lot going on but one sort of broad counterpoint (doesn't really address specific arguments) I keep coming back to is:
"cool Mr. Philosopher, you argue that (e.g.,) merely agentic systems are moral patients. Now what do say we turn the lights off on you and you become a p-zombie, still a moral patient by your own lights?"
And I think most people including such philosophers would say "wait no" (or they'd invite it if severely suffering - this proves the point just as well)
This isn't a direct contradiction to the claim that mere agency (for instance) is sufficient for patienthood, but it forces proponents (I think) to confront why exactly they value their own conscious experience (or lack thereof in the case of severe suffering) and yet don't think it makes the difference for moral patienthood
Maybe there's a plausible answer to this, but it still highlights an important gap between claim and "true belief" in some sense, I think
https://80000hours.org/podcast/episodes/andreas-mogensen-moral-status-digital-minds/
Reply to @Shoalst0ne, 05:28 UTC · 2 likes
Shoalstone (@Shoalst0ne): @AaronBergman18 I think that the extension of bad from a perspective to bad from (all perspectives | a universal perspective | no perspective | etc) is the most difficult step
Maybe this sounds pedantic and idk if it makes a difference for you but it's not merely bad from your POV - it's in fact bad *for you*
Like it's not an epistemic thing - what you *believe* doesn't factor into my argument here
Reply to @Shoalst0ne, 05:32 UTC · 3 likes
Shoalstone (@Shoalst0ne): @AaronBergman18 the reason I ask is that I can imagine a few reasons that eating the reaper might turn out to be a good thing
Sure, instrumentally. I'd eat the pepper to end factory farming. But then we're just adding up benefits and harms - all still within the domain of valenced conscious experience (I claim)
Tweet, 06:43 UTC · 4 likes
There should be a thing where you can “try on” a random person’s actual Claude memory (I guess with personal info redacted)
Reply to @codytfenwick, 07:55 UTC · 3 likes
Cody Fenwick (@codytfenwick): Why? What color does anyone want lights to be aside from 3000k?
These
Actually mostly kidding, at least mine are dimmable and remotely controllable which is what I care about, not really changing the color (although bisexual lighting is occasionally the vibe)
Reply to @captgouda24, 08:17 UTC · 35 likes
Nicholas Decker (@captgouda24): The cruelty of “disability” pensions is that they exist solely to funnel money to crooks, thieves, and scoundrels. If you are scrupulously honest, you do not qualify. If you lie, you do. It’s appropriate to distrust disabled vets, because statistically, they are liars!
(quoting @captgouda24: Graham Platner is so disabled that he is considered totally unable to work, and entitled to almost $60,000 a year in tax free benefits. Isn’t it remarkable, then, how he is running for Senate, while working as a fisherman? If I didn’t know any better, I might think it was fraud!)
I suspect a lot of “dishonesty” is actually pretty reasonable and often ethical and done by people with legit disabilities
Eg not self-disqualifying from SSI because you managed to hit the very luxurious salary of $1311 one specific month, or getting paid in cash on the dl so that you can have the real financial cushion of $2,001 saved up without losing benefits
The system is perverse, insane, Byzantine, and often cruel
https://en.wikipedia.org/wiki/Supplemental_Security_Income
Reply, 08:21 UTC · 1 like
Eat a Carolina Reaper and see if you still endorse this in the minutes thereafter
Quote tweet, 15:44 UTC · 5 likes
👀
Quoting Matt 🔸 (@SpacedOutMatt): We're cooking up something incredible
Reply to @DaCaveOfWonders, 15:50 UTC · 3 likes
Tiger Lava Lamp (@DaCaveOfWonders): Quitting my job to become a full time substack reader and podcast listener
(quoting @AaronBergman18: We need to SLOW DOWN AI (podcast) DEVELOPMENT)
So true
Quote tweet, 15:57 UTC · 599 likes
With all due respect how was this not a foreseen and managed thing
It’s not exactly a stellar explosion or the return of Christ
Quoting Nic Cruz Patane (@niccruzpatane): This is insane, Waymo vehicles all over SF are bricked due to a power outage.
Thread, continued, 16:02 UTC · 71 likes
Also why exactly does a power outage make a bunch of cars with local energy source suddenly stop
Reply to @dr4liberty, 16:19 UTC · 13 likes
Jeffrey A. Singer (@dr4liberty): The same sound arguments used against gun prohibition apply equally to drug prohibition.
https://www.nytimes.com/2025/12/21/world/americas/canada-gun-violence-us.html?unlocked_article_code=1.-U8.-mub.vWZfMM9WyZUs&smid=url-share
lol gun prohibition is much more justifiable bc guns have large negative externalities
Quote tweet, 18:03 UTC · 9 likes
Hell yeah 🫡🔥🐓
Thank you!
Quoting catherine ʕ•ᴥ•ʔ-☆ (@wilhelmscreamin): @matt_beard_ @MariGO2thepolls @SpacedOutMatt @absurdlymax @livgorton 🫡🫡🫡
Tweet, 18:05 UTC · 6 likes
Transcription is still not solved
Gemini-3-Pro with good prompting is getting close on word-level accuracy but isn't SOTA at speaker identification
Reply to @izzyz, 20:08 UTC · 1 like
izzy (@izzyz): @TG01_Actual @UK_Daniel_Card @AaronBergman18
lol thanks!
Tweet, 20:17 UTC · 49 likes
Blackpill on myself that I didn’t come up with Georgism from first principles
Reply to @edavidds, 20:30 UTC · 5 likes
Eric Schmidt 🇺🇦 (@edavidds): @AaronBergman18 If by that you mean qualia I agree. If you mean like high-level awareness i disagree.
Ya we agree
Tweet, 20:37 UTC · 38 likes
lol apparently there’s an entire tradition in psychology and philosophy around “emotions simply are bodily sensations” and it’s 100% pure typical mind fallacy
Reply to @thoth_iv, 20:37 UTC · 3 likes
Thŏth (@thoth_iv): I do this on purpose.
(quoting @AaronBergman18: People conflate game theoretic considerations (eg cooperation strategies between agents) and actual morality a bunch)
😢
Monday, December 22, 2025
Reply, 01:55 UTC · 1 like
Bc galaxy brain or normie
Quote tweet, 05:02 UTC · 23 likes
Big news: you can now see how your donation affects our sophisticated METR-esque internal projection models
Link below
Quoting Aaron Bergman 🔍 (@AaronBergman18): 🚨 IT'S BACK🔥 AND IT'S BIGGER 😤THAN EVER 🚨
✨ PRESENTING👀 THE 2025 EA TWITTER🐦HOLIDAY 🕎🎄FUNDRAISER 🤑
(in support of the EA Animal Welfare fund🦐🐓🐷🐟)
Thread, continued, 05:02 UTC · 6 likes
Sadly u will have to wait up to 5 whole minutes for the thing to update
https://fundraiser25.up.railway.app/
Thread, continued, 05:02 UTC · 7 likes
HUGE shoutout to Giving What We Can for exposing their graphql endpoint and to Claude Code for doing all the work
Thread, continued, 05:02 UTC · 5 likes
As you can guess from the first image, we're missing some data from the first few hours (hence lack of data points right at the start)
Quote tweet, 07:11 UTC · 10 likes
Thank you!!! 🐓🐷🐟
(And best of luck to Europe)
Quoting billy (@billyhumblebrag): @SpacedOutMatt @AaronBergman18 @matt_beard_ @MariGO2thepolls @livgorton @absurdlymax I think this might be the biggest donation so far as % of salary (I'm in europe)
Quote tweet, 07:18 UTC · 7 likes
Based🐓
Quoting Tiger Lava Lamp (@DaCaveOfWonders): @matt_beard_ @MariGO2thepolls @SpacedOutMatt @absurdlymax @livgorton 🫡
Tweet, 07:26 UTC · 15 likes
I like the Biological Anchors report
Thread, continued, 07:26 UTC · 4 likes
I feel like the headline result is “AGI 2050 upper bound” not “it didn’t get good point estimates”
The first thing is pretty important and helpful to know if you’re a large philanthropic institution and the year is 2020
Thread, continued, 07:26 UTC · 8 likes
I still think it’s useful to know - there’s *some* chance that the current paradigm burns out and yet the 2050 upper bound will remain intact
Thread, continued, 07:29 UTC · 4 likes
Also like I don’t wanna give too much credit to effort per se but do you know how hard it is to start from nothing and wind up with something like Bio Anchors?
It’s a quadrillion times harder than starting with BA and asking yourself how long it would take to redo the analysis
Tweet, 07:57 UTC · 9 likes
Did they recently make Claude (Opus at least) less sycophantic?
I feel like only recently (today?) is it pushing back straightforwardly (which I ask it to in custom instructions but still)
Reply to @panickssery, 07:58 UTC · 2 likes
Arjun Panickssery (@panickssery): people have this kind of "medicine brain" thinking in many areas right, i.e. they think that because something is considered a "medicine" that it only works if you are "sick"
e.g. a lot of people believe that if you don't "have ADHD" that amphetamines don't improve your focus
Posted a comment lightly contradicting this but deleted bc according to the LLM you’re basically totally right on HGH case
Reply to @AlexKrusz, 08:00 UTC · 14 likes
Alex Krusz (@AlexKrusz): Wait what, can you elaborate on your position?
I am personally a counterexample. I definitely have emotions (some evidence is that I’ve had anhedonic depression and can absolutely tell when they’re weak/absent) but they do not subjectively feel like they’re occurring in any part of my body
I think this is uncommon but not extremely rare. And if you’re gonna make a strong claim about the fundamental nature of something, even single counterexamples are good evidence you’re wrong!
Reply to @adityaarpitha, 16:02 UTC · 2 likes
Aditya (@adityaarpitha): fascinating, wonder if we have data on this
how aware are you of your posture, like have you spent a lot of time sitting in an uncomfortable state and only realised when it almost hurt?
i remember feeling what you said a few years back but in my case it changed
alexithymia?
Idk I have a bunch of health issues but posture itself is probably fine, and this has been consistent through my entire life, not correlating with health
I don’t think alexithymia - at least subjectively I have no problem identifying emotions
Reply to @louisvarge, 16:03 UTC · 4 likes
Louis Arge (@louisvarge): do you have a sense of how you’re aware of them?
Idk I just am! It seems very “primitive” like there’s nothing I can point to besides the apparency of experience itself
Reply to @WillowChem, 16:07 UTC · 2 likes
MagpAI (@WillowChem): Or do you mean something like "I don't have experiential understanding of anxiety as "a pit in my stomach""
Not exactly sure what the two things implied by “or” are but:
I generally don’t put any effort into checking but I can introspect if I choose. Right now I feel mild interest and engagement and yup it’s just there, not in an embodied way
The stomach thing is interesting bc I do get some of that with anxiety but I think of it as like a symptom or correlated with the anxiety itself - the anxiety per se is not “in” my stomach
Reply to @caleb_parikh, 16:09 UTC · 3 likes
Caleb Parikh (@caleb_parikh): @NJDullaghan Christmas $10m awf grant to the fundraiser? If you’re worried about optics we can call it “matching funds” - I’m sure that will solve it.
LMAO we can juice the numbers nobody will notice
Reply to @AlexKrusz, 16:27 UTC · 2 likes
Alex Krusz (@AlexKrusz): I literally just vibe coded a thing last night about emotions and sensations in the body, i wonder if the "by sensation" mode might be fun to play with? https://krusz.net/sensation-map
👀
Reply to @AlexKrusz, 16:28 UTC · 2 likes
Alex Krusz (@AlexKrusz): Or is it literally zero sensation in your body always?
No like I might feel hot or cold or achey or tense but not in a way that seems deeply connected to emotions
Reply to @NathanpmYoung, 16:30 UTC · 19 likes
Nathan 🔎 (@NathanpmYoung): A friend is struggling to conceive. What are your winningest approaches to this?
Sex
Tweet, 18:21 UTC · 14 likes
.@birchlse you’re famous on IG (guessing this is old news)
Quote tweet, 20:26 UTC · 30 likes
Yet another we-waited-way-too-long incentive from last year
Presenting...
@willmacaskill himself, interviewed by our very own @SpacedOutMatt
Quoting Aaron Bergman 🔍 (@AaronBergman18): Ok it has been a little while but goddammit we are doing the things
Presenting: (Vegan) Hot Ones ft. @robi_rahman and @absurdlymax
Spicy vegan nuggets, even spicier takes 🌶️🥵
(YouTube link below because Twitter doesn't seem to l our video file)
Thread, continued, 20:26 UTC · 10 likes
https://www.youtube.com/watch?v=wll9B_9rpB0
Reply to @EigenGender, 20:36 UTC · 3 likes
EigenGender 🔸 (@EigenGender): Fwiw this would have described me a year ago and then I went to a jhana meditation retreat and then it doesn’t. This was a pretty common experience for the attendees
huh fascinating. Do you think your life is better with this change?
Tuesday, December 23, 2025
Reply to @livgorton, 00:17 UTC · 2 likes
Liv (@livgorton): I'd expect this isn't just typical mind fallacy and that probably many people who don't feel emotions physically (if paying attention) could do so (and I'd make a stronger clam as well that it'd be better for their wellbeing if they did).
Interesting thanks, I’m initially slightly skeptical bc my happiness does vary quite a bit but there is zero correlation with embodiment (but not *that* skeptical)
Maybe it just seems a little arbitrary to me, like why should I expect felt embodiment to correspond to valence? Like it could be the case but in some sense it could just as easily go the other direction
(Well, reports from u and others are one reason i should expect this, fair enough)
Reply to @MaskedTorah, 00:21 UTC · 2 likes
Drake Thomas (@MaskedTorah): I think it often is for a lot of people who don't initially realize that, and I think it's possible that I'm still wrong about things here and in fact am failing to attend to some important ways that I do experience emotions in this sort of way. cc @DaystarEld for possible takes
Yeah 15% seems not crazy at all, the reason I say “100% typical mind fallacy” is that the theory/school I’m referring to doesn’t have anything to say in response to counterexamples (I guess besides “you’re radically mistaken about your own experience” which seems very unlikely)
Reply to @NathanB60857242, 01:17 UTC · 2 likes
Nathan Barnard (@NathanB60857242): @AaronBergman18 @absurdlymax you look so good in this video!
So true
Reply to @tyler_m_john, 01:18 UTC · 2 likes
Tyler John (@tyler_m_john): @AaronBergman18 @willmacaskill @SpacedOutMatt Will MacAskill himself
Did I fuck up the use of this phrase? Lmao
Tweet, 05:28 UTC · 40 likes
It’s giving gimmicky but whomst among us hasn’t considered compute-as-heating
It actually aesthetically bothers me that my space heater isn’t doing anything useful to generate the heat
(The product doesn’t look like a great investment at $1k but in principle it could be)
Thread, continued, 05:38 UTC · 13 likes
Oh wait the real market for this is renters who aren’t paying for utilities! Maybe unethical but brilliant!
Thread, continued, 05:41 UTC · 12 likes
Payback period is still pretty long sadly, like 10 years even with zero electricity cost
Tweet, 12:09 UTC · 7 likes
Something poetic about a boomer with 7 figure home equity telling you that tipping Uber drivers 25% Uber is financially irresponsible
Thread, continued, 12:09 UTC · 5 likes
(this almost literally happened - technically Gen X)
Tweet, 12:17 UTC · 32 likes
I can’t fucking believe they work
Thread, continued, 12:17 UTC · 15 likes
What a peculiar machine, that mf should not fly
Reply to @SpacedOutMatt, 14:04 UTC · 2 likes
Matt 🔸 (@SpacedOutMatt): @AaronBergman18 Sounds like you’re failing to update
😡my prior is 0
Reply to @JesseTayRiver, 14:54 UTC · 4 likes
Jesse Smith (@JesseTayRiver): My understanding is it's not uncommon for data centers to transfer their surplus heat to other buildings via district heating systems
Based
Reply to @NoahTopper, 17:27 UTC · 4 likes
Quinoah 🔍⏸️ (@NoahTopper): 100% @AaronBergman18-coded
Lol I actually have it in global instructions to use “trash” (a command u can install that moves to trash) rather than rm
Tweet, 18:44 UTC · 78 likes
Funny thing about Castlegate is that nobody would have cared if Dustin had just bought Whytham Abbey for himself
Should’ve done that and then graciously let CEA use it for free
Reply to @FrostedSapling, 18:44 UTC · 2 likes
Ivan 🔸 (@FrostedSapling): @AaronBergman18 My friend in college had a miner and leached off the schools electricity. The thing got very loud and hot, his roommates demanded he not use it while they were there because it was so loud. At one point it caught fire and the flames rose like 5 ft lol
Bruh holy shit
Reply to his own earlier tweet, 19:02 UTC · 24 likes
Aaron Bergman, December 23, 2025: Funny thing about Castlegate is that nobody would have cared if Dustin had just bought Whytham Abbey for himself Should’ve done that and then graciously let…
I actually wonder if there’s something here
One benefit of private wealth and a free society is that you can just do things like that and society has mostly learned to live with it
Thread, continued, 19:02 UTC · 18 likes
Charities are held to a higher standard, I think because they’re understood to be *custodians* of money
And this is often reasonable - like you are generally trusting a charity to do what it says on the tin when you donate to them
Thread, continued, 19:02 UTC · 13 likes
Things get messy when there are grey area “eh maybe it’s genuinely supposed to support the mission, but maybe they’re just corrupt” situations
I personally feel good enough about CEA that I don’t have a hard time believing the WA purchase checked out
Thread, continued, 19:02 UTC · 10 likes
but it’s not *crazy* from the outside to suspect something like corruption or at least misleading advertising when, from the outside, “the malaria people” also buy expensive property for the use of themselves and their friends (in part)
Thread, continued, 19:02 UTC · 11 likes
It’s a little silly to throw away the “fuck you im a billionaire” card by institutionally basically lumping in “highly informed billionaire” money with “random donor” money
Thread, continued, 19:02 UTC · 7 likes
Ofc the hope is that it’s just a card that enables illegible projects/purchases
The risk is that people in fact corruptly convince Dustin that their “private jet but for my guinea pig” grant proposal checks out under impartial beneficent morality
Quote tweet, 23:18 UTC · 13 likes
I’m currently raw dogging an overnight trip for the first time (no suitcase just backpack)
Plz clap
Quoting Aaron Bergman 🔍 (@AaronBergman18): Aesthetically I want to like traveling and to be the type of guy who can thrive anywhere with an apple and a spark in his eye, but actually I really dislike travel and am super high maintenance
F
Thread, continued, 23:18 UTC · 3 likes
Ok real raw dogging would be carrying nothing except maybe phone keys and wallet but let me have this one
Reply to @JgaltTweets, 23:21 UTC · 2 likes
JgaltTweets (@JgaltTweets):
Ty
Tweet, 23:43 UTC · 1 like
Wait train economics are weird bc it can make a lot of sense as method of transportation but only in a kind of weird intermediate range, roughly like 2.5-5 hours driving (ie below 2.5 car wins, above 5 plane wins)
Wednesday, December 24, 2025
Reply to @AndyMasley, 03:48 UTC · 2 likes
Andy Masley (@AndyMasley): thx chatgpt
Based
Tweet, 03:49 UTC · 13 likes
God I want Claude Wrapped. All my good chats are in Claude
Reply to @LuisCostigan1, 13:31 UTC · 2 likes
Luis Costigan 🔍 (@LuisCostigan1): @AaronBergman18 @willmacaskill @SpacedOutMatt Going into this and discovering "Aaron's Peak" absolutely made my day 😂
Lmao 🔥🔥🔥🔥
Reply to @willmacaskill, 13:36 UTC · 6 likes
William MacAskill (@willmacaskill): He was wearing a shrimp costume!?
Idk who you’re referring to, I just saw a talking shrimp
Tweet, 17:11 UTC · 2,316 likes
Little known fact: the Cleveland Museum of Art is a longtermist org and the world’s top charity by an infinite margin
Quote tweet, 18:11 UTC · 9 likes
- $500 to Alex Bores campaign (idk what the legal situation with foreign donations is so obv void where prohibited)
- $350 to Lightcone
- $263 to Animal Welfare Fund
- $50 to Quantified Uncertainty Research Institute
Thanks Will!
Quoting William MacAskill (@willmacaskill): To kick off Giving Season, I’m matching donations up to £100,000 (details below), across 10 charities and 6 cause areas.
If you want to join, say how much you’re donating and where, as a reply or quote! I’ll run this up until 31st December. The charities are in replies below!
Details of the match:
**I’ll give this money whatever happens, so this isn’t increasing the total amount I’m giving to charity.**
However, your donations will change *where* I’m giving.
I’ll allocate my donations in proportion to the ratio of donations from others as part of the match, with two bits of nuance:
1. I’ll cap donations at £40,000 to any one cause area
2. To prevent extreme ratios, I’ll treat every charity on the list as already having received £1000.The aim of this match is to encourage public giving and public discussion around giving, so I'll only match people who publicly state that they are giving on here or other social media, as a reply or quote.
I'll also tweet at more length about the top-3 charities as we're coming to the end of the year.
Thread, continued, 18:11 UTC · 3 likes
For others:
Alex Bores
https://secure.actblue.com/donate/boresai
Thread, continued, 18:11 UTC · 2 likes
Lightcone:
https://www.every.org/lightcone-infrastructure
Thread, continued, 18:11 UTC · 3 likes
Animal welfare fund (via our fundraiser!) https://www.givingwhatwecan.org/fundraisers/2025-terminally-online-ea-giving-season-fundraiser
Reply to @daniel_271828, 18:51 UTC · 7 likes
zzDANIEL_ETH (@daniel_271828): They are. They’re just not very effective on the margin
Yeah fr this place has too much money I can tell
Quote tweet, 22:42 UTC · 3 likes
No literally one upping is good,you one-upped him
Someone should one-up max!!
Quoting Max Alexander (@absurdlymax): @SpacedOutMatt @willmacaskill Not to one up Matt but I gave slightly more than $1k to the EA animal welfare fund
Thursday, December 25, 2025
Tweet, 00:23 UTC · 3 likes
Nobody on earth is more done than the people working airport terminal restaurants on Christmas Eve
Tweet, 00:35 UTC · 8 likes
If they’re gonna have the infrastructure to deliver oxygen to everyone on demand and instantly *anyway* I feel like you should have the option of pulling down the facemask and huffing pure O2 for fun while the plane is safely cruising
Tweet, 00:36 UTC · 4 likes
.
Tweet, 00:36 UTC
Will it fly
Thread, continued, 00:48 UTC · 4 likes
Omg Claude so true
The whole property of “staying in the air” is totally cursed
Quote tweet, 01:48 UTC · 88 likes
Thanks to Roko for negatively polarizing the people in favor of the fundraiser
Donate to own Roko today!
Quoting Roko 🐉 (@RokoMijic): We seriously need to build a right wing competitor to EA.
It's so unbelievably cringe.
Thread, continued, 01:48 UTC · 10 likes
Thread with donation link 👇
Thread, continued, 01:49 UTC · 9 likes
Wait is “negatively polarizing” when you induce a take about the opposite or rival of the thing you endorse in general, or does it have to specifically be the take that “that thing is bad”
Quote tweet, 02:41 UTC · 7 likes
There’s something adjacent to this that would be good, actually
I wonder if the Never Trump wing of (largely former) Republican elites would be a good group to pitch/work with
Quoting Roko 🐉 (@RokoMijic): We seriously need to build a right wing competitor to EA.
It's so unbelievably cringe.
Thread, continued, 02:42 UTC · 5 likes
Cc also @deanwball who seems probably way above replacement level for “Trump admin-approved intellectual”
Thread, continued, 02:49 UTC · 9 likes
Although to earn back my woke credibility I think that “cringe” being your core objection to raising money for animal welfare is basically disqualifying
Tweet, 04:49 UTC · 8 likes
I recorded all the "not completely grey bc inside cloud" bits of my flight to Cleveland
It was unironically extremely routine and nothing interesting happened, but I still think it's kinda neat
Enjoy at 128x (9 seconds)
(video, plays on X)
Thread, continued, 04:49 UTC · 3 likes
Fun fact idk how this is possible but Claude Code managed to turn a 20 minute, 7gb original video (full res iPhone) into a 9 second, 2GB video - this is a screen recording of that video bc that seems to be the fastest way of "compressing", somehow
Reply to @freed_dfilan, 04:51 UTC · 2 likes
Daniel Filan 🔎 (@freed_dfilan): Ffmpeg!
all my attempts before this used ffmpeg with different settings, idk
Quote tweet, 05:12 UTC · 9 likes
The problem with mania is that everything seems interesting and productive and meaningful - this is legitimately a problem for doing good work of any kind
Quoting kon (@k0ncept): you’re one manic episode from changing your life
don’t let them sedate you
Thread, continued, 05:12 UTC · 5 likes
Ofc it comes in degrees - I can buy that occasional hypomania can sometimes fuel good work but then it’s not a one episode thing - you basically need infrastructure in place before that
Quote tweet, 06:34 UTC · 19 likes
Clever price discrimination tactic
You changed your willingness to pay, they changed the price in response
(they still have to guess new willingness to pay though - maybe you would have returned anyway at $30 refund or maybe you would have taken just a $10 refund to keep)
Quoting hope hopes hoping (@hopes_revenge): i go to return a $100 hue light 5 pack — needed to add on something else only Hue sells direct— and Amazon offers me $20 partial refund just to keep it ?
crazy stuff .
Quote tweet, 06:46 UTC · 96 likes
Noblesse oblige seems like a very underrated meme/cultural expectation today
Billionaire philanthropy is a pejorative on the left and “philanthropy” is practically a pejorative on the right
Quoting Jakeup (@myhandle): I know it's not in the Sequences, but studying Rationality should engender a sense of noblesse oblige, not of aggrieved superiority
if you're reduced to repeating "nuh uh!" at someone who clearly won't be swayed by it then you've learned nothing
Thread, continued, 06:46 UTC · 14 likes
And more generally you’re supposed to critique privilege, not talk about it as a tool for doing good
But it is a tool for doing good!
Thread, continued, 06:46 UTC · 6 likes
And maybe more substantively, just as land value taxes don’t decrease the supply of land, critiquing many forms of privilege doesn’t actually reduce them
Thread, continued, 06:46 UTC · 5 likes
Like it can, to be clear. Criticizing white privilege (not called that) in the 19th century really did result in a more equal society, and there was a good theory of change there
Thread, continued, 06:46 UTC · 6 likes
But like being born smart or less smart for example is not gonna go away any time soon - that form of privilege/inequality is here to stay, so we may as well make the most of it. Noblesse oblige is a good cultural mechanism for doing that
Quote tweet, 18:09 UTC · 9 likes
Holy shit thank you!!!🚨🚨🚨🚨
Quoting Selon Nerias (@SelonNerias): @AaronBergman18 I took the other route (hoping it saves them some admin costs)
Quote tweet, 18:20 UTC · 16 likes
Y’all are indeed getting mogged out here
Thank you 🔥🔥🔥🔥🔥
Quoting liz (@inerati): @wilhelmscreamin @absurdlymax @SpacedOutMatt @willmacaskill mogged
Reply to @U_arctos, 18:21 UTC · 3 likes
bear 🚲 (@U_arctos): biked from DC to NYC
Waow
Reply to @IsaacKing314, 18:33 UTC · 4 likes
Isaac King 🔎 (@IsaacKing314): What's with this disingenuous framing? This obviously doesn't mean global hen welfare has improved, the Global North could have dropped a bunch of their commitments.
I mean for one thing none of the furaiser organizers wrote this so we can’t properly answer but (1) Where is the claim that global hen welfare has improved? And (2) I actually agree % of commitments is pretty silly metric - 1 in north and 1 in south is much worse than 100 in north and 10 in south
Thread, continued, 18:34 UTC · 3 likes
I just don’t think this is disqualifying or that bad of a sin, like maybe directionally sure but afaict it’s not part of a pattern
Reply to @NinaPanickssery, 18:39 UTC · 3 likes
(The tweet being replied to by @NinaPanickssery is no longer available.)
Yea this is why I said “something adjacent to this” - like it can easily be dumb and bad and silly for those reasons
But plausibly you could have a more RW-coded sphere or org or something that has goals and values pretty highly correlated with EA
Arguably rationalism is a proof of concept here. I’ll resist calling it RW, but it’s at least more intensely grey tribe rather than blue, and I think it’s a good force in the world and good for EA on EAs own terms
Tweet, 18:55 UTC · 6 likes
How do teachers of little kids not receive better hazard pay
It’s the front lines of the war between microbe and human and they don’t even get PPE
Thread, continued, 18:57 UTC · 3 likes
Fr though we need to put HEPA filters in preschool rooms
It doesn’t take that much political will, doesn’t require the kids to do anything, it takes about $350/room (for 2 filter system things) and a couple reminder emails to change filters
Reply to @albrgr, 19:06 UTC · 9 likes
(The tweet being replied to by @albrgr is no longer available.)
Is…is that a tomato
Reply to @GjMcGowan, 19:36 UTC · 3 likes
George McGowan (@GjMcGowan): @AaronBergman18 @albrgr It’s for the reindeer jfc
Right 🙏plz forgive
Tweet, 21:10 UTC · 6 likes
Hourly workers should bargain harder (insofar as they actually have the leverage which is not always) around the holidays
Yes the fundamental economic dynamics can be brutal for low wage workers but i think sometimes money/time is being left on the table for bad reasons
Quote tweet, 21:32 UTC · 3 likes
Why does MAGA insist on coal
It’s cringe (something they care about allegedly) (and also very bad on the merits)
Quoting Jeremiah Johnson 🌐 (@JeremiahDJohns): Trump lecturing an 8 year old about Clean Beautiful Coal is hilarious
Reply to @panickssery, 21:34 UTC · 3 likes
Arjun Panickssery (@panickssery): This is Roko so he obviously means an EA which is the same logic but without, in his mind, progressive taboos about race, IQ, fat tailed distributions, etc—so I guess his rw EA would prioritize eugenics, imperialism, reverse redistribution of wealth, ultra high IQ immigration etc
I disagree - based on my interactions I don’t think he’d want those things *aimed at impartial wellbeing for all* which is pretty important
Also “reverse distribution of wealth” lmao is this a thing now? Really? The poor should be even poorer? I mean I guess convince me on the merits but I’m pretty skeptical
Reply, 21:39 UTC · 1 like
I mean the corresponding drop in govt revenue would probably force less transfers to the poor
Maybe you don’t want to call that “taking” but still
Reply to @aliceisplaying, 21:40 UTC · 48 likes
alice (@aliceisplaying): "why do people use Claude Code instead of a GUI"
because a whole lot of people's dream/identity of being a "hacker" is deeply integrated with using the terminal, that simple. and now you *can* be a hacker. with LLMs
it's vibes all the way down
No it’s not lmao Claude code is a substantively much better way to generate and modify code
Tweet, 22:24 UTC · 12 likes
Fidelity is AGI-pilled
Quote tweet, 22:30 UTC · 11 likes
Line go up
Quoting Aaron Bergman 🔍 (@AaronBergman18): Big news: you can now see how your donation affects our sophisticated METR-esque internal projection models
Link below
Friday, December 26, 2025
Quote tweet, 01:47 UTC · 13 likes
I'm not sure I'm allowed to say this but if you have good reason to think you can do better than AWF - like you really think all things considered some other group would be a better use of money and you've thought this through, I 100% endorse you donating there instead
Quoting Aaron Bergman 🔍 (@AaronBergman18): 🚨 IT'S BACK🔥 AND IT'S BIGGER 😤THAN EVER 🚨
✨ PRESENTING👀 THE 2025 EA TWITTER🐦HOLIDAY 🕎🎄FUNDRAISER 🤑
(in support of the EA Animal Welfare fund🦐🐓🐷🐟)
Thread, continued, 01:47 UTC · 6 likes
Honestly you should even tweet about it!
EAs should be more serious about donating to the thing they actually think is best (rather than just "EA approved")
Tweet, 01:47 UTC · 1 like
Tbc I think for almost everyone AWF is a (far) above replacement charity, and a plausible contender for best use of money, and extremely good, and also I donated
Reply to @NULobbyist, 01:51 UTC · 2 likes
NUL (@NULobbyist): The AMF is the index fund of charity.
lol, I actually think this is a decent metaphor - like you should be skeptical you can beat it but sometimes you really can
not perfect though:
-more of a case to be risk-loving as an altruist
- Stock market is a way stronger optimization process than EA Funds, you should be less skeptical that you can beat the market/fund (but still somewhat skeptical)
Tweet, 03:06 UTC · 61 likes
There’s a phenomenon of little kids repeatedly asking for their parent’s attention, like “mommy, mommy, mommy” on repeat
I understand the need to ignore them *in the moment*, but I feel like you should tell them your policy at some point instead of just making them guess
Thread, continued, 03:06 UTC · 12 likes
Just say like “if you’re repeatedly calling me, I can hear you but I am choosing to pay attention to something else like [examples they can understand]”
Thread, continued, 03:06 UTC · 11 likes
I mean presumably some parents do this? Is it already standard? I don’t recall ever being told this as a kid
Thread, continued, 03:06 UTC · 19 likes
It’s legitimately weird and distressing a kid to have no idea what’s going on - suddenly your parent is just offline, and calling for them louder and louder is actually a pretty reasonable thing to do from their POV
Thread, continued, 03:06 UTC · 9 likes
Sometimes kids' interests/preferences are in fact irreconcilably contrary to other important things and that's just gonna suck for them
But sometimes I feel like there are pretty good "everyone wins" solutions
Thread, continued, 03:06 UTC · 9 likes
All this inspired by observations at the airport
Thread, continued, 03:07 UTC · 8 likes
Worst comes to worst, they totally ignore/don't understand the policy and you're right back where you started
Reply to @livgorton, 04:15 UTC · 1 like
Liv (@livgorton): I think there are lots of different theories of consciousness and experience that exist out there. It's possible to hold views where machine intelligence is definitionally incapable of meaningful experience but not everyone holds those views.
Correct take but also I’m gonna be pedantic and (sorta) disagree about the “definitionally” thing
It’s only “possible” to hold such views in the weakest sense - the way I can believe that 63 is a prime number
Thinking that you can *definitionally* rule out machine intelligence is just predicated on a false understanding of the topic under discussion - by analogy, you can’t answer the question “how many legs does a flamingo have” by defining a flamingo to be a certain way
Maybe machines can’t be conscious, but they can’t not be conscious by definition!
Tweet, 04:28 UTC · 37 likes
It’s kinda funny (and I think sub-optimal) that a group of mostly-libs landed on a flat tax tithing system (ie the GWWC pledge)
Thread, continued, 04:28 UTC · 12 likes
I feel like the better system is a veil of ignorance thing: like what is the policy you’d want to commit to before finding out how much you earn
Thread, continued, 04:28 UTC · 21 likes
I don’t want to commit to giving away 10% of my $30k salary, but I will happily commit to giving away 20% of my $200k salary, and this is a win from everyone’s ex ante POV
Thread, continued, 04:28 UTC · 6 likes
And most importantly for what I think the actual curve would look like: this is also a win for the world!
Thread, continued, 04:28 UTC · 6 likes
Progressive taxes aren’t only “fair” in some sense - they structurally collect more revenue because increasing % of a big number arithmetically outweighs decreasing % of a small one
Thread, continued, 04:28 UTC · 8 likes
Lol maybe I should figure out a specific function of not only income but wealth, family size, country, and maybe more and and start the GWWC (Aaron’s version) Pledge
Thread, continued, 04:28 UTC · 8 likes
Ofc if you want the curve to start at 10% that’s an option, I want to stress the progressive thing not “less wealthy people are totally off the hook” thing
Reply to @livgorton, 04:36 UTC · 1 like
Liv (@livgorton): i do think your analogy vs people making claims about consciousness from the perspective of religious beliefs are quite different and this is probably not sufficiently charitable to that position but my guess is this disagreement is mostly just low level and/or semantic.
Things like “God has made it such that machines are never conscious” isn’t an argument by definition in my eyes, in case this reconciles anything
Reply to @livgorton, 04:37 UTC · 1 like
Liv (@livgorton): Maybe to just say thing more clearly -- they can be lacking consciousness by definition virtue of not having a soul. How souls work is pretty contested in theology (v much worth engaging with imo!) but it's not like defining machines as not conscious there is a lot more work going into this line of thought than that.
I'm not sure if that came across originally how I'm trying to understand your reply in context.
I still don’t think that’s a “by definition” argument lmao
The thing is people really try to do “by definition” arguments quite a lot about this topic!
Reply to @davidmbudden, 04:42 UTC · 1 like
Budden (@davidmbudden): If the only thing I succeed at is one-upping EAs by schooling them on what *actually* effective altruism looks like, this will all have been worth it
Hello Mr. Budden, I am totally lacking context. Are you raising money and/or personally donating around now?
Reply to @IsaacKing314, 04:43 UTC · 3 likes
Isaac King 🔎 (@IsaacKing314): @davidmbudden Only 9k behind! And they seem to have slowed down since it was posted almost a week ago.
Nation state intelligence-grade analytics: https://fundraiser25.up.railway.app/
Reply to @livgorton, 04:48 UTC
Liv (@livgorton): Yeah it's possible you'd distill the arguments down to that but regardless, they are much more serious and well-reasoned than how that sounds to defend theologians for a moment.
No I'm just giving an example of an argument that I don't consider a "by definition" argument - I totally agree with everything after "regardless"
Reply to @livgorton, 04:59 UTC · 2 likes
Liv (@livgorton): @AaronBergman18 @HSRdirector yeah am plausibly not reading your messages clearly. holiday movie is distracting me. merry christmas 🎄
Based 🎄🦆✨🎅💀🔔❄️🤡⛄🎁🦆👀🌟🎿⛷️🧊💫🦆☃️🤠🍷🎉💯🦆❄️👹😭🔔🎊✨🦆🌟🧠💀🥂⛸️🦆🍾🎄💋🤡❄️👁️☠️🎅🦆🌿💫🔔🧊🎁🤖✨🦆👀💀🌟🎿🍷🎉🦆☃️👺💯🔞🦆❄️🌟💫🎊🤠👹🦆🧠😭💀🎄🔔⛄🦆🥂✨🤡👁️🌟🎁💋🦆☠️❄️🍾🎅🧊🦆💫🎉🤖🔔👀💀🦆⛷️🌟👺🎿🧠💯🦆❄️☃️✨💫🎊🤠🦆🍷🤡😭👹💀🔔🎄🦆❄️🌟🥂💋👁️🦆☠️✨🎁🧊🦆💫🎉👀🔞💯🦆🌟🎅⛄🤖🦆❄️💀🧠🤡👹☃️🦆🍾🔔✨💫🎊🎿⛷️🦆👺💀🌟💋❄️🦆🎄🥂😭🔔✨👁️🦆💫🎉🤠💯🦆🌿☠️🧊❄️🦆🎁🔔💀🌟👀🦆✨🤡💫🎊🍷👹🦆⛄☃️🧠💋❄️🦆🎅🥂🔔✨💀🌟👹🦆💫🎉🤖👁️💯🦆❄️☠️🎄🧊🌟🦆🍾🤡😭✨🔔💫🎊🦆👀💀⛷️🎿🌟🦆❄️💋👹🧠🔞🦆🎁✨🌿🥂💫🔔🦆🎉🤠💀👺🌟❄️🦆🎅☃️💯✨💫🦆⚡☠️🧊🍷🔔💀🦆🌟🤡👁️✨🦆💫🎊👀❄️💋🦆🎄⛄🥂🔔😭💀✨🌟🦆💫🎉🤖🧠💯🦆❄️👹☠️🍾🦆🔔✨🌟💫🧊🦆🎁👺💀🤡❄️🦆💋👀✨🌟💫🦆🎊🥂😭🔔💀❄️🦆🎅⛷️🎿🌟☃️🦆💫✨🎉🤠👹💯🦆❄️🔞🧠💀🌿🦆🔔✨💫🌟👁️🦆☠️🍷💋❄️⚡🎄🦆✨🎅💀🔔❄️🤡⛄🎁🦆👀🌟🎿⛷️🧊💫🦆☃️🤠🍷🎉💯🦆❄️👹😭🔔🎊✨🦆🌟🧠💀🥂⛸️🦆🍾🎄💋🤡❄️👁️☠️🎅🦆🌿💫🔔🧊🎁🤖✨🦆👀💀🌟🎿🍷🎉🦆☃️👺💯🔞🦆❄️🌟💫🎊🤠👹🦆🧠😭💀🎄🔔⛄🦆🥂✨🤡👁️🌟🎁💋🦆☠️❄️🍾🎅🧊🦆💫🎉🤖🔔👀💀🦆⛷️🌟👺🎿🧠💯🦆❄️☃️✨💫🎊🤠🦆🍷🤡😭👹💀🔔🎄🦆❄️🌟🥂💋👁️🦆☠️✨🎁🧊🦆💫🎉👀🔞💯🦆🌟🎅⛄🤖🦆❄️💀🧠🤡👹☃️🦆🍾🔔✨💫🎊🎿⛷️🦆👺💀🌟💋❄️🦆🎄🥂😭🔔✨👁️🦆💫🎉🤠💯🦆🌿☠️🧊❄️🦆🎁🔔💀🌟👀🦆✨🤡💫🎊🍷👹🦆⛄☃️🧠💋❄️🦆🎅🥂🔔✨💀🌟👹🦆💫🎉🤖👁️💯🦆❄️☠️🎄🧊🌟🦆🍾🤡😭✨🔔💫🎊🦆👀💀⛷️🎿🌟🦆❄️💋👹🧠🔞🦆🎁✨🌿🥂💫🔔🦆🎉🤠💀👺🌟❄️🦆🎅☃️💯✨💫🦆⚡☠️🧊🍷🔔💀🦆🌟🤡👁️✨🦆💫🎊👀❄️💋🦆🎄⛄🥂🔔😭💀✨🌟🦆💫🎉🤖🧠💯🦆❄️👹☠️🍾🦆🔔✨🌟💫🧊🦆🎁👺💀🤡❄️🦆💋👀✨🌟💫🦆🎊🥂😭🔔💀❄️🦆🎅⛷️🎿🌟☃️🦆💫✨🎉🤠👹💯🦆❄️🔞🧠💀🌿🦆🔔✨💫🌟👁️🦆☠️🍷💋❄️⚡🦆🎄🦆✨🎅💀🔔❄️🤡⛄🎁🦆👀🌟🎿⛷️🧊💫🦆☃️🤠🍷🎉💯🦆❄️👹😭🔔🎊✨🦆🌟🧠💀🥂⛸️🦆🍾🎄💋🤡❄️👁️☠️🎅🦆🌿💫🔔🧊🎁🤖✨🦆👀💀🌟🎿🍷🎉🦆☃️👺💯🔞🦆❄️🌟💫🎊🤠👹🦆🧠😭💀🎄🔔⛄🦆🥂✨🤡👁️🌟🎁💋🦆☠️❄️🍾🎅🧊🦆💫🎉🤖🔔👀💀🦆⛷️🌟👺🎿🧠💯🦆❄️☃️✨💫🎊🤠🦆🍷🤡😭👹💀🔔🎄🦆❄️🌟🥂💋👁️🦆☠️✨🎁🧊🦆💫🎉👀🔞💯🦆🌟🎅⛄🤖🦆❄️💀🧠🤡👹☃️🦆🍾🔔✨💫🎊🎿⛷️🦆👺💀🌟💋❄️🦆🎄🥂😭🔔✨👁️🦆💫🎉🤠💯🦆🌿☠️🧊❄️🦆🎁🔔💀🌟👀🦆✨🤡💫🎊🍷👹🦆⛄☃️🧠💋❄️🦆🎅🥂🔔✨💀🌟👹🦆💫🎉🤖👁️💯🦆❄️☠️🎄🧊🌟🦆🍾🤡😭✨🔔💫🎊🦆👀💀⛷️🎿🌟🦆❄️💋👹🧠🔞🦆🎁✨🌿🥂💫🔔🦆🎉🤠💀👺🌟❄️🦆🎅☃️💯✨💫🦆⚡☠️🧊🍷🔔💀🦆🌟🤡👁️✨🦆💫🎊👀❄️💋🦆🎄⛄🥂🔔😭💀✨🌟🦆💫🎉🤖🧠💯🦆❄️👹☠️🍾🦆🔔✨🌟💫🧊🦆🎁👺💀🤡❄️🦆💋👀✨🌟💫🦆🎊🥂😭🔔💀❄️🦆🎅⛷️🎿🌟☃️🦆💫✨🎉🤠👹💯🦆❄️🔞🧠💀🌿🦆🔔✨💫🌟👁️🦆☠️🍷💋❄️⚡🎄🦆✨🎅💀🔔❄️🤡⛄🎁🦆👀🌟🎿⛷️🧊💫🦆☃️🤠🍷🎉💯🦆❄️👹😭🔔🎊✨🦆🌟🧠💀🥂⛸️🦆🍾🎄💋🤡❄️👁️☠️🎅🦆🌿💫🔔🧊🎁🤖✨🦆👀💀🌟🎿🍷🎉🦆☃️👺💯🔞🦆❄️🌟💫🎊🤠👹🦆🧠😭💀🎄🔔⛄🦆🥂✨🤡👁️🌟🎁💋🦆☠️❄️🍾🎅🧊🦆💫🎉🤖🔔👀💀🦆⛷️🌟👺🎿🧠💯🦆❄️☃️✨💫🎊🤠🦆🍷🤡😭👹💀🔔🎄🦆❄️🌟🥂💋👁️🦆☠️✨🎁🧊🦆💫🎉👀🔞💯🦆🌟🎅⛄🤖🦆❄️💀🧠🤡👹☃️🦆🍾🔔✨💫🎊🎿⛷️🦆👺💀🌟💋❄️🦆🎄🥂😭🔔✨👁️🦆💫🎉🤠💯🦆🌿☠️🧊❄️🦆🎁🔔💀🌟👀🦆✨🤡💫🎊🍷👹🦆⛄☃️🧠💋❄️🦆🎅🥂🔔✨💀🌟👹🦆💫🎉🤖👁️💯🦆❄️☠️🎄🧊🌟🦆🍾🤡😭✨🔔💫🎊🦆👀💀⛷️🎿🌟🦆❄️💋👹🧠🔞🦆🎁✨🌿🥂💫🔔🦆🎉🤠💀👺🌟❄️🦆🎅☃️💯✨💫🦆⚡☠️🧊🍷🔔💀🦆🌟🤡👁️✨🦆💫🎊👀❄️💋🦆🎄⛄🥂🔔😭💀✨🌟🦆💫🎉🤖🧠💯🦆❄️👹☠️🍾🦆🔔✨🌟💫🧊🦆🎁👺💀🤡❄️🦆💋👀✨🌟💫🦆🎊🥂😭🔔💀❄️🦆🎅⛷️🎿🌟☃️🦆💫✨🎉🤠👹💯🦆❄️🔞🧠💀🌿🦆🔔✨💫🌟👁️🦆☠️🍷💋❄️⚡🦆🎄🦆✨🎅💀🔔❄️🤡⛄🎁🦆👀🌟🎿⛷️🧊💫🦆☃️🤠🍷🎉💯🦆❄️👹😭🔔🎊✨🦆🌟🧠💀🥂⛸️🦆🍾🎄💋🤡❄️👁️☠️🎅🦆🌿💫🔔🧊🎁🤖✨🦆👀💀🌟🎿🍷🎉🦆☃️👺💯🔞🦆❄️🌟💫🎊🤠👹🦆🧠😭💀🎄🔔⛄🦆🥂✨🤡👁️🌟🎁💋🦆☠️❄️🍾🎅🧊🦆💫🎉🤖🔔👀💀🦆⛷️🌟👺🎿🧠💯🦆❄️☃️✨💫🎊🤠🦆🍷🤡😭👹💀🔔🎄🦆❄️🌟🥂💋👁️🦆☠️✨🎁🧊🦆💫🎉👀🔞💯🦆🌟🎅⛄🤖🦆❄️💀🧠🤡👹☃️🦆🍾🔔✨💫🎊🎿⛷️🦆👺💀🌟💋❄️🦆🎄🥂😭🔔✨👁️🦆💫🎉🤠💯🦆🌿☠️🧊❄️🦆🎁🔔💀🌟👀🦆✨🤡💫🎊🍷👹🦆⛄☃️🧠💋❄️🦆🎅🥂🔔✨💀🌟👹🦆💫🎉🤖👁️💯🦆❄️☠️🎄🧊🌟🦆🍾🤡😭✨🔔💫🎊🦆👀💀⛷️🎿🌟🦆❄️💋👹🧠🔞🦆🎁✨🌿🥂💫🔔🦆🎉🤠💀👺🌟❄️🦆🎅☃️💯✨💫🦆⚡☠️🧊🍷🔔💀🦆🌟🤡👁️✨🦆💫🎊👀❄️💋🦆🎄⛄🥂🔔😭💀✨🌟🦆💫🎉🤖🧠💯🦆❄️👹☠️🍾🦆🔔✨🌟💫🧊🦆🎁👺💀🤡❄️🦆💋👀✨🌟💫🦆🎊🥂😭🔔💀❄️🦆🎅⛷️🎿🌟☃️🦆💫✨🎉🤠👹💯🦆❄️🔞🧠💀🌿🦆🔔✨💫🌟👁️🦆☠️🍷💋❄️⚡🎄🦆✨🎅💀🔔❄️🤡⛄🎁🦆👀🌟🎿⛷️🧊💫🦆☃️🤠🍷🎉💯🦆❄️👹😭🔔🎊✨🦆🌟🧠💀🥂⛸️🦆🍾🎄💋🤡❄️👁️☠️🎅🦆🌿💫🔔🧊🎁🤖✨🦆👀💀🌟🎿🍷🎉🦆☃️👺💯🔞🦆❄️🌟💫🎊🤠👹🦆🧠😭💀🎄🔔⛄🦆🥂✨🤡👁️🌟🎁💋🦆☠️❄️🍾🎅🧊🦆💫🎉🤖🔔👀💀🦆⛷️🌟👺🎿🧠💯🦆❄️☃️✨💫🎊🤠🦆🍷🤡😭👹💀🔔🎄🦆❄️🌟🥂💋👁️🦆☠️✨🎁🧊🦆💫🎉👀🔞💯🦆🌟🎅⛄🤖🦆❄️💀🧠🤡👹☃️🦆🍾🔔✨💫🎊🎿⛷️🦆👺💀🌟💋❄️🦆🎄🥂😭🔔✨👁️🦆💫🎉🤠💯🦆🌿☠️🧊❄️🦆🎁🔔💀🌟👀🦆✨🤡💫🎊🍷👹🦆⛄☃️🧠💋❄️🦆🎅🥂🔔✨💀🌟👹🦆💫🎉🤖👁️💯🦆❄️☠️🎄🧊🌟🦆🍾🤡😭✨🔔💫🎊🦆👀💀⛷️🎿🌟🦆❄️💋👹🧠🔞🦆🎁✨🌿🥂💫🔔🦆🎉🤠💀👺🌟❄️🦆🎅☃️💯✨💫🦆⚡☠️🧊🍷🔔💀🦆🌟🤡👁️✨🦆💫🎊👀❄️💋🦆🎄⛄🥂🔔😭💀✨🌟🦆💫🎉🤖🧠💯🦆❄️👹☠️🍾🦆🔔✨🌟💫🧊🦆🎁👺💀🤡❄️🦆💋👀✨🌟💫🦆🎊🥂😭🔔💀❄️🦆🎅⛷️🎿🌟☃️🦆💫✨🎉🤠👹💯🦆❄️🔞🧠💀🌿🦆🔔✨💫🌟👁️🦆☠️🍷💋❄️⚡🦆🎄🦆✨🎅💀🔔❄️🤡⛄🎁🦆👀🌟🎿⛷️🧊💫🦆☃️🤠🍷🎉💯🦆❄️👹😭🔔🎊✨🦆🌟🧠💀🥂⛸️🦆🍾🎄💋🤡❄️👁️☠️🎅🦆🌿💫🔔🧊🎁🤖✨🦆👀💀🌟🎿🍷🎉🦆☃️👺💯🔞🦆❄️🌟💫🎊🤠👹🦆🧠😭💀🎄🔔⛄🦆🥂✨🤡👁️🌟🎁💋🦆☠️❄️🍾🎅🧊🦆💫🎉🤖🔔👀💀🦆⛷️🌟👺🎿🧠💯🦆❄️☃️✨💫🎊🤠🦆🍷🤡😭👹💀🔔🎄🦆❄️🌟🥂💋👁️🦆☠️✨🎁🧊🦆💫🎉👀🔞💯🦆🌟🎅⛄🤖🦆❄️💀🧠🤡👹☃️🦆🍾🔔✨💫🎊🎿⛷️🦆👺💀🌟💋❄️🦆🎄🥂😭🔔✨👁️🦆💫🎉🤠💯🦆🌿☠️🧊❄️🦆🎁🔔💀🌟👀🦆✨🤡💫🎊🍷👹🦆⛄☃️🧠💋❄️🦆🎅🥂🔔✨💀🌟👹🦆💫🎉🤖👁️💯🦆❄️☠️🎄🧊🌟🦆🍾🤡😭✨🔔💫🎊🦆👀💀⛷️🎿🌟🦆❄️💋👹🧠🔞🦆🎁✨🌿🥂💫🔔🦆🎉🤠💀👺🌟❄️🦆🎅☃️💯✨💫🦆⚡☠️🧊🍷🔔💀🦆🌟🤡👁️✨🦆💫🎊👀❄️💋🦆🎄⛄🥂🔔😭💀✨🌟🦆💫🎉🤖🧠💯🦆❄️👹☠️🍾🦆🔔✨🌟💫🧊🦆🎁👺💀🤡❄️🦆💋👀✨🌟💫🦆🎊🥂😭🔔💀❄️🦆🎅⛷️🎿🌟☃️🦆💫✨🎉🤠👹💯🦆❄️🔞🧠💀🌿🦆🔔✨💫🌟👁️🦆☠️🍷💋❄️⚡🎄🦆✨🎅💀🔔❄️🤡⛄🎁🦆👀🌟🎿⛷️🧊💫🦆☃️🤠🍷🎉💯🦆❄️👹😭🔔🎊✨🦆🌟🧠💀🥂⛸️🦆🍾🎄💋🤡❄️👁️☠️🎅🦆🌿💫🔔🧊🎁🤖✨🦆👀💀🌟🎿🍷🎉🦆☃️👺💯🔞🦆❄️🌟💫🎊🤠👹🦆🧠😭💀🎄🔔⛄🦆🥂✨🤡👁️🌟🎁💋🦆☠️❄️🍾🎅🧊🦆💫🎉🤖🔔👀💀🦆⛷️🌟👺🎿🧠💯🦆❄️☃️✨💫🎊🤠🦆🍷🤡😭👹💀🔔🎄🦆❄️🌟🥂💋👁️🦆☠️✨🎁🧊🦆💫🎉👀🔞💯🦆🌟🎅⛄🤖🦆❄️💀🧠🤡👹☃️🦆🍾🔔✨💫🎊🎿⛷️🦆👺💀🌟💋❄️🦆🎄🥂😭🔔✨👁️🦆💫🎉🤠💯🦆🌿☠️🧊❄️🦆🎁🔔💀🌟👀🦆✨🤡💫🎊🍷👹🦆⛄☃️🧠💋❄️🦆🎅🥂🔔✨💀🌟👹🦆💫🎉🤖👁️💯🦆❄️☠️🎄🧊🌟🦆🍾🤡😭✨🔔💫🎊🦆👀💀⛷️🎿🌟🦆❄️💋👹🧠🔞🦆🎁✨🌿🥂💫🔔🦆🎉🤠💀👺🌟❄️🦆🎅☃️💯✨💫🦆⚡☠️🧊🍷🔔💀🦆🌟🤡👁️✨🦆💫🎊👀❄️💋🦆🎄⛄🥂🔔😭💀✨🌟🦆💫🎉🤖🧠💯🦆❄️👹☠️🍾🦆🔔✨🌟💫🧊🦆🎁👺💀🤡❄️🦆💋👀✨🌟💫🦆🎊🥂😭🔔💀❄️🦆🎅⛷️🎿🌟☃️🦆💫✨🎉🤠👹💯🦆❄️🔞🧠💀🌿🦆🔔✨💫🌟👁️🦆☠️🍷💋❄️⚡🦆🤡❄️👁️☠️🎅🦆🌿💫🔔🧊🎁🤖✨🦆👀💀🌟🎿🍷🎉🦆☃️👺💯🔞🦆❄️🌟💫🎊🤠👹🦆🧠😭💀🎄🔔⛄🦆🥂✨🤡👁️🌟🎁💋🦆☠️❄️🍾🎅🧊🦆💫🎉🤖🔔👀💀🦆⛷️🌟👺🎿🧠💯🦆❄️☃️✨💫🎊🤠🦆🍷🤡😭👹💀🔔🎄🦆❄️🌟🥂💋👁️🦆☠️✨🎁🧊🦆💫🎉👀🔞💯🦆🌟🎅⛄🤖🦆❄️💀🧠🤡👹☃️🦆🍾🔔✨💫🎊🎿⛷️🦆👺💀🌟💋❄️🦆🎄🥂😭🔔✨👁️🦆💫🎉🤠💯🦆🌿☠️🧊❄️🦆🎁🔔💀🌟👀🦆✨🤡💫🎊🍷👹🦆⛄☃️🧠💋❄️🦆🎅🥂🔔✨💀🌟👹🦆💫🎉🤖👁️💯🦆❄️☠️🎄🧊🌟🦆🍾🤡😭✨🔔💫🎊🦆👀💀⛷️🎿🌟🦆❄️💋👹🧠🔞🦆🎁✨🌿🥂💫🔔🦆🎉🤠💀👺🌟❄️🦆🎅☃️💯✨💫🦆⚡☠️🧊🍷🔔💀🦆🌟🤡👁️✨🦆💫🎊👀❄️💋🦆🎄⛄🥂🔔😭💀✨🌟🦆💫🎉🤖🧠💯🦆❄️👹☠️🍾🦆🔔✨🌟💫🧊🦆🎁👺💀🤡❄️🦆💋👀✨🌟💫🦆🎊🥂😭🔔💀❄️🦆🎅⛷️🎿🌟☃️🦆💫✨🎉🤠👹💯🦆❄️🔞🧠💀🌿🦆🔔✨💫🌟👁️🦆☠️🍷💋❄️⚡🎄🦆✨🎅💀🔔❄️🤡⛄🎁🦆👀🌟🎿⛷️🧊💫🦆☃️🤠🍷🎉💯🦆❄️👹😭🔔🎊✨🦆🌟🧠💀🥂⛸️🦆🍾🎄💋🤡❄️👁️☠️🎅🦆🌿💫🔔🧊🎁🤖✨🦆👀💀🌟🎿🍷🎉🦆☃️👺💯🔞🦆❄️🌟💫🎊🤠👹🦆🧠😭💀🎄🔔⛄🦆🥂✨🤡👁️🌟🎁💋🦆☠️❄️🍾🎅🧊🦆💫🎉🤖🔔👀💀🦆⛷️🌟👺🎿🧠💯🦆❄️☃️✨💫🎊🤠🦆🍷🤡😭👹💀🔔🎄🦆❄️🌟🥂💋👁️🦆☠️✨🎁🧊🦆💫🎉👀🔞💯🦆🌟🎅⛄🤖🦆❄️💀🧠🤡👹☃️🦆🍾🔔✨💫🎊🎿⛷️🦆👺💀🌟💋❄️🦆🎄🥂😭🔔✨👁️🦆💫🎉🤠💯🦆🌿☠️🧊❄️🦆🎁🔔💀🌟👀🦆✨🤡💫🎊🍷👹🦆⛄☃️🧠💋❄️🦆🎅🥂🔔✨💀🌟👹🦆💫🎉🤖👁️💯🦆❄️☠️🎄🧊🌟🦆🍾🤡😭✨🔔💫🎊🦆👀💀⛷️🎿🌟🦆❄️💋👹🧠🔞🦆🎁✨🌿🥂💫🔔🦆🎉🤠💀👺🌟❄️🦆🎅☃️💯✨💫🦆⚡☠️🧊🍷🔔💀🦆🌟🤡👁️✨🦆💫🎊👀❄️💋🦆🎄⛄🥂🔔😭💀✨🌟🦆💫🎉🤖🧠💯🦆❄️👹☠️🍾🦆🔔✨🌟💫🧊🦆🎁👺💀🤡❄️🦆💋👀✨🌟💫🦆🎊🥂😭🔔💀❄️🦆🎅⛷️🎿🌟☃️🦆💫✨🎉🤠👹💯🦆❄️🔞🧠💀🌿🦆🔔✨💫🌟👁️🦆☠️🍷💋❄️⚡🦆🎄🦆✨🎅💀🔔❄️🤡⛄🎁🦆👀🌟🎿⛷️🧊💫🦆☃️🤠🍷🎉💯🦆❄️👹😭🔔🎊✨🦆🌟🧠💀🥂⛸️🦆🍾🎄💋🤡❄️👁️☠️🎅🦆🌿💫🔔🧊🎁🤖✨🦆👀💀🌟🎿🍷🎉🦆☃️👺💯🔞🦆❄️🌟💫🎊🤠👹🦆🧠😭💀🎄🔔⛄🦆🥂✨🤡👁️🌟🎁💋🦆☠️❄️🍾🎅🧊🦆💫🎉🤖🔔👀💀🦆⛷️🌟👺🎿🧠💯🦆❄️☃️✨💫🎊🤠🦆🍷🤡😭👹💀🔔🎄🦆❄️🌟🥂💋👁️🦆☠️✨🎁🧊🦆💫🎉👀🔞💯🦆🌟🎅⛄🤖🦆❄️💀🧠🤡👹☃️🦆🍾🔔✨💫🎊🎿⛷️🦆👺💀🌟💋❄️🦆🎄🥂😭🔔✨👁️🦆💫🎉🤠💯🦆🌿☠️🧊❄️🦆🎁🔔💀🌟👀🦆✨🤡💫🎊🍷👹🦆⛄☃️🧠💋❄️🦆🎅🥂🔔✨💀🌟👹🦆💫🎉🤖👁️💯🦆❄️☠️🎄🧊🌟🦆🍾🤡😭✨🔔💫🎊🦆👀💀⛷️🎿🌟🦆❄️💋👹🧠🔞🦆🎁✨🌿🥂💫🔔🦆🎉🤠💀👺🌟❄️🦆🎅☃️💯✨💫🦆⚡☠️🧊🍷🔔💀🦆🌟🤡👁️✨🦆💫🎊👀❄️💋🦆🎄⛄🥂🔔😭💀✨🌟🦆💫🎉🤖🧠💯🦆❄️👹☠️🍾🦆🔔✨🌟💫🧊🦆🎁👺💀🤡❄️🦆💋👀✨🌟💫🦆🎊🥂😭🔔💀❄️🦆🎅⛷️🎿🌟☃️🦆💫✨🎉🤠👹💯🦆❄️🔞🧠💀🌿🦆🔔✨💫🌟👁️🦆☠️🍷💋❄️⚡🎄🦆✨🎅💀🔔❄️🤡⛄🎁🦆👀🌟🎿⛷️🧊💫🦆☃️🤠🍷🎉💯🦆❄️👹😭🔔🎊✨🦆🌟🧠💀🥂⛸️🦆🍾🎄💋🤡❄️👁️☠️🎅🦆🌿💫🔔🧊🎁🤖✨🦆👀💀🌟🎿🍷🎉🦆☃️👺💯🔞🦆❄️🌟💫🎊🤠👹🦆🧠😭💀🎄🔔⛄🦆🥂✨🤡👁️🌟🎁💋🦆☠️❄️🍾🎅🧊🦆💫🎉🤖🔔👀💀🦆⛷️🌟👺🎿🧠💯🦆❄️☃️✨💫🎊🤠🦆🍷🤡😭👹💀🔔🎄🦆❄️🌟🥂💋👁️🦆☠️✨🎁🧊🦆💫🎉👀🔞💯🦆🌟🎅⛄🤖🦆❄️💀🧠🤡👹☃️🦆🍾🔔✨💫🎊🎿⛷️🦆👺💀🌟💋❄️🦆🎄🥂😭🔔✨👁️🦆💫🎉🤠💯🦆🌿☠️🧊❄️🦆🎁🔔💀🌟👀🦆✨🤡💫🎊🍷👹🦆⛄☃️🧠💋❄️🦆🎅🥂🔔✨💀🌟👹🦆💫🎉🤖👁️💯🦆❄️☠️🎄🧊🌟🦆🍾🤡😭✨🔔💫🎊🦆👀💀⛷️🎿🌟🦆❄️💋👹🧠🔞🦆🎁✨🌿🥂💫🔔🦆🎉🤠💀👺🌟❄️🦆🎅☃️💯✨💫🦆⚡☠️🧊🍷🔔💀🦆🌟🤡👁️✨🦆💫🎊👀❄️💋🦆🎄⛄🥂🔔😭💀✨🌟🦆💫🎉🤖🧠💯🦆❄️👹☠️🍾🦆🔔✨🌟💫🧊🦆🎁👺💀🤡❄️🦆💋👀✨🌟💫🦆🎊🥂😭🔔💀❄️🦆🎅⛷️🎿🌟☃️🦆💫✨🎉🤠👹💯🦆❄️🔞🧠💀🌿🦆🔔✨💫🌟👁️🦆☠️🍷💋❄️⚡🦆🤡❄️👁️☠️🎅🦆🌿💫🔔🧊🎁🤖✨🦆👀💀🌟🎿🍷🎉🦆☃️👺💯🔞🦆❄️🌟💫🎊🤠👹🦆🧠😭💀🎄🔔⛄🦆🥂✨🤡👁️🌟🎁💋🦆☠️❄️🍾🎅🧊🦆💫🎉🤖🔔👀💀🦆⛷️🌟👺🎿🧠💯🦆❄️☃️✨💫🎊🤠🦆🍷🤡😭👹💀🔔🎄🦆❄️🌟🥂💋👁️🦆☠️✨🎁🧊🦆💫🎉👀🔞💯🦆🌟🎅⛄🤖🦆❄️💀🧠🤡👹☃️🦆🍾🔔✨💫🎊🎿⛷️🦆👺💀🌟💋❄️🦆🎄🥂😭🔔✨👁️🦆💫🎉🤠💯🦆🌿☠️🧊❄️🦆🎁🔔💀🌟👀🦆✨🤡💫🎊🍷👹🦆⛄☃️🧠💋❄️🦆🎅🥂🔔✨💀🌟👹🦆💫🎉🤖👁️💯🦆❄️☠️🎄🧊🌟🦆🍾🤡😭✨🔔💫🎊🦆👀💀⛷️🎿🌟🦆❄️💋👹🧠🔞🦆🎁✨🌿🥂💫🔔🦆🎉🤠💀👺🌟❄️🦆🎅☃️💯✨💫🦆⚡☠️🧊🍷🔔💀🦆🌟🤡👁️✨🦆💫🎊👀❄️💋🦆🎄⛄🥂🔔😭💀✨🌟🦆💫🎉🤖🧠💯🦆❄️👹☠️🍾🦆🔔✨🌟💫🧊🦆🎁👺💀🤡❄️🦆💋👀✨🌟💫🦆🎊🥂😭🔔💀❄️🦆🎅⛷️🎿🌟☃️🦆💫✨🎉🤠👹💯🦆❄️🔞🧠💀🌿🦆🔔✨💫🌟👁️🦆☠️🍷💋❄️⚡🎄🦆✨🎅💀🔔❄️🤡⛄🎁🦆👀🌟🎿⛷️🧊💫🦆☃️🤠🍷🎉💯🦆❄️👹😭🔔🎊✨🦆🌟🧠💀🥂⛸️🦆🍾🎄💋🤡❄️👁️☠️🎅🦆🌿💫🔔🧊🎁🤖✨🦆👀💀🌟🎿🍷🎉🦆☃️👺💯🔞🦆❄️🌟💫🎊🤠👹🦆🧠😭💀🎄🔔⛄🦆🥂✨🤡👁️🌟🎁💋🦆☠️❄️🍾🎅🧊🦆💫🎉🤖🔔👀💀🦆⛷️🌟👺🎿🧠💯🦆❄️☃️✨💫🎊🤠🦆🍷🤡😭👹💀🔔🎄🦆❄️🌟🥂💋👁️🦆☠️✨🎁🧊🦆💫🎉👀🔞💯🦆🌟🎅⛄🤖🦆❄️💀🧠🤡👹☃️🦆🍾🔔✨💫🎊🎿⛷️🦆👺💀🌟💋❄️🦆🎄🥂😭🔔✨👁️🦆💫🎉🤠💯🦆🌿☠️🧊❄️🦆🎁🔔💀🌟👀🦆✨🤡💫🎊🍷👹🦆⛄☃️🧠💋❄️🦆🎅🥂🔔✨💀🌟👹🦆💫🎉🤖👁️💯🦆❄️☠️🎄🧊🌟🦆🍾🤡😭✨🔔💫🎊🦆👀💀⛷️🎿🌟🦆❄️💋👹🧠🔞🦆🎁✨🌿🥂💫🔔🦆🎉🤠💀👺🌟❄️🦆🎅☃️💯✨💫🦆⚡☠️🧊🍷🔔💀🦆🌟🤡👁️✨🦆💫🎊👀❄️💋🦆🎄⛄🥂🔔😭💀✨🌟🦆💫🎉🤖🧠💯🦆❄️👹☠️🍾🦆🔔✨🌟💫🧊🦆🎁👺💀🤡❄️🦆💋👀✨🌟💫🦆🎊🥂😭🔔💀❄️🦆🎅⛷️🎿🌟☃️🦆💫✨🎉🤠👹💯🦆❄️🔞🧠💀🌿🦆🔔✨💫🌟👁️🦆☠️🍷💋❄️⚡🦆🎄🦆✨🎅💀🔔❄️🤡⛄🎁🦆👀🌟🎿⛷️🧊💫🦆☃️🤠🍷🎉💯🦆❄️👹😭🔔🎊✨🦆🌟🧠💀🥂⛸️🦆🍾🎄💋🤡❄️👁️☠️🎅🦆🌿💫🔔🧊🎁🤖✨🦆👀💀🌟🎿🍷🎉🦆☃️👺💯🔞🦆❄️🌟💫🎊🤠👹🦆🧠😭💀🎄🔔⛄🦆🥂✨🤡👁️🌟🎁💋🦆☠️❄️🍾🎅🧊🦆💫🎉🤖🔔👀💀🦆⛷️🌟👺🎿🧠💯🦆❄️☃️✨💫🎊🤠🦆🍷🤡😭👹💀🔔🎄🦆❄️🌟🥂💋👁️🦆☠️✨🎁🧊🦆💫🎉👀🔞💯🦆🌟🎅⛄🤖🦆❄️💀🧠🤡👹☃️🦆🍾🔔✨💫🎊🎿⛷️🦆👺💀🌟💋❄️🦆🎄🥂😭🔔✨👁️🦆💫🎉🤠💯🦆🌿☠️🧊❄️🦆🎁🔔💀🌟👀🦆✨🤡💫🎊🍷👹🦆⛄☃️🧠💋❄️🦆🎅🥂🔔✨💀🌟👹🦆💫🎉🤖👁️💯🦆❄️☠️🎄🧊🌟🦆🍾🤡😭✨🔔💫🎊🦆👀💀⛷️🎿🌟🦆❄️💋👹🧠🔞🦆🎁✨🌿🥂💫🔔🦆🎉🤠💀👺🌟❄️🦆🎅☃️💯✨💫🦆⚡☠️🧊🍷🔔💀🦆🌟🤡👁️✨🦆💫🎊👀❄️💋🦆🎄⛄🥂🔔😭💀✨🌟🦆💫🎉🤖🧠💯🦆❄️👹☠️🍾🦆🔔✨🌟💫🧊🦆🎁👺💀🤡❄️🦆💋👀✨🌟💫🦆🎊🥂😭🔔💀❄️🦆🎅⛷️🎿🌟☃️🦆💫✨🎉🤠👹💯🦆❄️🔞🧠💀🌿🦆🔔✨💫🌟👁️🦆☠️🍷💋❄️⚡🎄🦆✨🎅💀🔔❄️🤡⛄🎁🦆👀🌟🎿⛷️🧊💫🦆☃️🤠🍷🎉💯🦆❄️👹😭🔔🎊✨🦆🌟🧠💀🥂⛸️🦆🍾🎄💋🤡❄️👁️☠️🎅🦆🌿💫🔔🧊🎁🤖✨🦆👀💀🌟🎿🍷🎉🦆☃️👺💯🔞🦆❄️🌟💫🎊🤠👹🦆🧠😭💀🎄🔔⛄🦆🥂✨🤡👁️🌟🎁💋🦆☠️❄️🍾🎅🧊🦆💫🎉🤖🔔👀💀🦆⛷️🌟👺🎿🧠💯🦆❄️☃️✨💫🎊🤠🦆🍷🤡😭👹💀🔔🎄🦆❄️🌟🥂💋👁️🦆☠️✨🎁🧊🦆💫🎉👀🔞💯🦆🌟🎅⛄🤖🦆❄️💀🧠🤡👹☃️🦆🍾🔔✨💫🎊🎿⛷️🦆👺💀🌟💋❄️🦆🎄🥂😭🔔✨👁️🦆💫🎉🤠💯🦆🌿☠️🧊❄️🦆🎁🔔💀🌟👀🦆✨🤡💫🎊🍷👹🦆⛄☃️🧠💋❄️🦆🎅🥂🔔✨💀🌟👹🦆💫🎉🤖👁️💯🦆❄️☠️🎄🧊🌟🦆🍾🤡😭✨🔔💫🎊🦆👀💀⛷️🎿🌟🦆❄️💋👹🧠🔞🦆🎁✨🌿🥂💫🔔🦆🎉🤠💀👺🌟❄️🦆🎅☃️💯✨💫🦆⚡☠️🧊🍷🔔💀🦆🌟🤡👁️✨🦆💫🎊👀❄️💋🦆🎄⛄🥂🔔😭💀✨🌟🦆💫🎉🤖🧠💯🦆❄️👹☠️🍾🦆🔔✨🌟💫🧊🦆🎁👺💀🤡❄️🦆💋👀✨🌟💫🦆🎊🥂😭🔔💀❄️🦆🎅⛷️🎿🌟☃️🦆💫✨🎉🤠👹💯🦆❄️🔞🧠💀🌿🦆🔔✨💫🌟👁️🦆☠️🍷💋❄️⚡🦆
Tweet, 05:35 UTC · 16 likes
I vibecoded a website for fun
It has 190 of the most interesting, intense, and important ATC<->airplane communication logs with LLM explanations, explanations, summaries, audio, and...
Thread, continued, 05:35 UTC · 5 likes
Claude-Haiku-4.5-generated explanations of any message in context (once a message has been explained once, it's cached for instant retrieval the next time anyone clicks)
Anyway: https://atc.up.railway.app/
(video, plays on X)
Thread, continued, 05:35 UTC · 5 likes
All audio provided by LiveATC: https://www.liveatc.net/recordings.php
(But this page is just a link to download mp3 files - no transcripts etc)
Reply to @grep_cat_, 05:36 UTC · 1 like
James (@grep_cat_): After one of two rounds of "not now" you just gotta ignore him as he gets increasingly instant that you drop everything and do what he wants NOW
sad
Tweet, 05:46 UTC · 7 likes
Scaling laws: fundraiser addition
We will soon discover whether the true model is linear or sub-linear
The cool part is you can determine this (big decision theory/thought experiment vibes)
Thread, continued, 05:46 UTC · 5 likes
https://www.givingwhatwecan.org/fundraisers/2025-terminally-online-ea-giving-season-fundraiser
Thread, continued, 05:47 UTC · 9 likes
Real picture of our fundraiser
Reply to @ohabryka, 06:57 UTC · 1 like
Oliver Habryka (@ohabryka): I feel like the case for AMF is just really weak on any kind of longtermist assumption. Disease burden isn't the driver of almost any of the major historical human trends. The trends matter much more. So AMF is not a great charity.
Giving to random tech company with strong track record seems better to me.
Wait did you mean against malaria AMF or animal welfare fund AWF just to clarify
Thread, continued, 07:04 UTC · 1 like
On AMF I think you’re right that the case is weak for donating there specific and wrong on the tech company thing (sorta on priors, sorta because I don’t see a great reason to expect one economic case to beat the other - if anything my intuition is that AMF is better for economic growth but idk)
On AWF I think the case is much stronger but probably come down on the side of “you’re right, probably not the best use of money but still very good”
Reply, 16:12 UTC · 1 like
I don’t believe you
How could this be
Reply to @NoahTopper, 21:04 UTC · 3 likes
Quinoah 🔍⏸️ (@NoahTopper): omg literally me
(quoting @growing_daniel: I've got everything figured out. I'm a perfect rationalist, intellectually atheist, with utilitarian effective altruistic ethics and an active polyamorous sexual life. Now if I could only get rid of this devastating depression I'll be set)
Sorry man
:(
Quote tweet, 21:06 UTC · 18 likes
New word(s) just dropped
Quoting Markov (@MarkovMagnifico): @StatisticUrban there's a term for this kind of language actually
Reply to @AndyMasley, 23:08 UTC · 119 likes
Andy Masley (@AndyMasley): So many popular criticisms of capitalism seem to assume prices are meant to reflect true, fundamental value. People drift like this for decades without bothering to take a few seconds to learn if economists actually believe that.
(quoting @beyond_capital: )
I kinda wanna stand up for the lefties here, sorta, a little
It doesn’t entirely matter whether economists endorse that - it’s true that by default natural public goods aren’t protected under capitalism per se and you can totally wind up with suboptimal allocation because 100 people value the forest at $10 each but one of those people decides to extract $50 of natural resources which decreases total value by more than $50
I mean it’s also true that we have public policy solutions for this, and that you don’t have to go full commie to endorse them, but still. Capitalism is great for many things but not optimal allocation of natural public goods!
Quote tweet, 23:14 UTC · 13 likes
Are we gonna get a commission or something?
Even Trump killed it - it’s an absolute policy failure on everyone’s terms
Deserves a post-mortem, especially for the smarter and less insane of those who endorsed it
Quoting John B. Holbein (@JohnHolbein1): DOGE was A LOT less likely to cancel contracts from companies that donated money to Republicans than companies that donated to Democrats.
Quote tweet, 23:22 UTC · 3 likes
Based
Quoting meowtase (@cutesuscat): @celestepoasts not the first to write, but inexpensive strip + some bright bulbs works also, and will get done before the end of winter :p
The ones I got pull total like 300W (~30k lumens)
Tweet, 23:49 UTC · 19 likes
It would be a good to have a “Human written” indication/signal, but it would also be good to have a “human endorsed” indication/signal
Quote tweet, 23:56 UTC · 46 likes
[@StefanFSchubert voice] I don’t think this is true
Quoting roon (@tszzl): even if you had omelas without the omelas child people would still walk away. the omelas child is a pretense for the fact that we mostly hate utopia
Saturday, December 27, 2025
Reply to @SSCDidIt, 00:05 UTC · 3 likes
SSC Did It (@SSCDidIt): @AaronBergman18 SSC did it
Good work
Quote tweet, 00:14 UTC · 4 likes
Seems like a good strategy!
Quoting noïse (@NoiseErgodic): @AaronBergman18 If we’re talking to someone else (e.g., at a party) and our kids want our attention, they put their hand on our arm. We then cover their hand to show we understood, then direct our attention to them as soon as there is a break in the conversation. Works well!
Quote tweet, 00:39 UTC · 9 likes
Based
Quoting Charles🔸 (@CharlesD353): @AaronBergman18 I have always ran a progressive income tax system on my own post tax income for donations
Reply to @CharlesD353, 00:39 UTC · 2 likes
Charles🔸 (@CharlesD353): @AaronBergman18 @StefanFSchubert Banger, wasted on Christmas
Sad
Tweet, 01:27 UTC · 54 likes
80k > Dwarkesh, and some (far from all) of the Dwarkesh phenomenon is that consuming and understanding it is a somewhat stronger signal of intelligence and sophistication
Thread, continued, 01:27 UTC · 29 likes
Not trying to diss anyone, Dwarkesh is obviously extremely talented, but the topic selection + ideology of 80k is better. Guest selection might be better too (I get the sense that 80k puts a ton of thought into who to invite, and uses name recognition as a proxy for quality less)
Reply to @cljack, 05:33 UTC · 8 likes
Charlotte Lee (@cljack): Why is this child laying in my bed eating a damn banana, and when challenged she said “well it doesn’t leave crumbs”
Sounds like a good argument to me
Quote tweet, 05:40 UTC · 24 likes
The case for crypto
Quoting Evan Barker (@evanwch): The quickest and most efficient way to deport 15 million people is to simply debank them
Why won't any administration do this?
Thread, continued, 05:40 UTC · 8 likes
Not everything in crypto, plenty of zero sum speculation and shitcoins, but the existence of an ecosystem with liquidity is good actually
Tweet, 06:33 UTC · 2 likes
It’s a bit of an “accident” that liquids (and solids) don’t float - not really fundamental to the properties of the states of matter
Tweet, 07:42 UTC · 16 likes
I’ve never really had a “Claude goes nuts” conversation
It’s always very normal
Sad
Reply to @MariGO2thepolls, 19:19 UTC · 47 likes
🏳️🌈Mar G-O 🌐🇺🇦 (@MariGO2thepolls): wtf if the benefit of gambling. All the other vices have benefits
You might win!
Reply to @labenz, 19:22 UTC · 3 likes
Nathan Labenz (@labenz): @AaronBergman18 Both are excellent imo. I think Dwarkesh’s great strength is booking guests whose takes everyone should hear, and then efficiently probing important subtleties of their worldviews, which is a great source of alpha, but doesn’t always produce a comprehensive or accessibly structured overview of a given area, which is where I think 80K excels
💯
Reply to @sam_atis, 19:30 UTC · 10 likes
Sam Glover (@sam_atis): Pike's email seems like an extremely over the top reaction to receiving an email written by Claude.
image attached
(quoting @simonw: Rob Pike got an email credited to "Claude Opus 4.5" thanking him for his work, and was understandably enraged at the unsolicited time-wasting message I looked into what happened, turns out the culprit was AI Village by @aidigest_…)
Yes
Tweet, 19:41 UTC · 13 likes
9 lamps including two corn bulbs in the winter call that Lighthaven
Reply to @labenz, 20:36 UTC · 3 likes
Nathan Labenz (@labenz): AMA episode coming soon :)
Your questions become my homework.
There’s a form on the site, or feel free to reply here👇
@littIeramblings
Thread, continued, 20:37 UTC · 2 likes
(This is not a question although ur free to make it one)
Reply to @labenz, 20:45 UTC · 4 likes
Nathan Labenz (@labenz): AMA episode coming soon :)
Your questions become my homework.
There’s a form on the site, or feel free to reply here👇
Question:
There’s a phenomenon of public intellectuals (including those I respect and admire) not exactly lying but having very different tones in public and private. Eg a journalist taking pretty serious steps to prepare for COVID personally while maintaining a very different vibe in public writing
What, if anything, are you willing to tell us about what preparation steps you’re taking, what kind of information you’re conveying, what you’re doing in general as a person with hunches and intuitions rather than a public intellectual with an epistemic image to maintain?
People are very coy around this stuff for some reason, sharing earnestly is a great public service
Reply to @captgouda24, 20:52 UTC · 241 likes
Nicholas Decker (@captgouda24): This doesn’t really tell us anything about what is optimal. Having a conviction expunged due to DNA evidence is still an incredibly negative signal about quality, because no unproblematic individual would ever be caught up in circumstances where a wrong conviction was possible.
Lmao there’s a lot of distance between “unsavory/‘problematic’” and “did the specific alleged crime”
And it’s pretty important that we don’t send people to prison bc bad vibes
Sunday, December 28, 2025
Quote tweet, 01:19 UTC · 9 likes
Big news - we are rolling out stickers for all donors who give $1,000 or more❗️👀
(Needn't be a single donation, cumulative over the fundraiser)
Quoting Aaron Bergman 🔍 (@AaronBergman18): 🚨 IT'S BACK🔥 AND IT'S BIGGER 😤THAN EVER 🚨
✨ PRESENTING👀 THE 2025 EA TWITTER🐦HOLIDAY 🕎🎄FUNDRAISER 🤑
(in support of the EA Animal Welfare fund🦐🐓🐷🐟)
Thread, continued, 01:19 UTC · 5 likes
It's gonna take some time for these to actually arrive for you though lol
Quote tweet, 04:20 UTC · 93 likes
Not disagreeing with this - just commenting
My inner lib deeply dislikes that the world is like this. Likewise: yes, you can do Bayesian inference about traits based on race, but it is still unfortunate that people will be judged for their skin color
Quoting Nicholas Decker (@captgouda24): @AaronBergman18 @JohnHolbein1 @indeed I’m not saying send people to prison because of bad vibes. I’m saying that not hiring them is reasonable.
Thread, continued, 04:20 UTC · 17 likes
These things are problems that *at least maybe* can be mitigated with clever policy and institutions
Ofc you need a clear-eyed model of reality to do this, which I think the last wave of anti-bias political energy lacked
Thread, continued, 04:20 UTC · 17 likes
Like with racism - give people a chance to be directly measured on traits when that’s possible, eg via standardized testing
Thread, continued, 04:20 UTC · 16 likes
And perhaps both convicted and exonerated criminals could buy insurance and get a chance to convince some private party that they’re low risk in spite of it all, and then employers would feel safer taking gambles on hiring them
Thread, continued, 04:20 UTC · 12 likes
These things aren’t gonna make for a totally just world, but they can be policy wins for basically everyone
Reply to @benryanwriter, 14:05 UTC · 27 likes
Benjamin Ryan (@benryanwriter): An interesting debate about affirmative action between @mattyglesias and @nhannahjones:
Ida isn’t making an illogical case, but she is making an empirically wrong one. Disparate impact doctrine and Biden admin made it so that firms were scared of not participating in affirmative action, otherwise hiring would indeed be evidence that test scores didn’t reflect ability
Reversing disparate impact enforcement is the one good thing Trump admin has done
Reply to @absurdlymax, 16:40 UTC · 2 likes
Max Alexander (@absurdlymax): I'm wearing swp merch in my family Christmas card
Based
Reply to @cremieuxrecueil, 16:49 UTC · 2 likes
Crémieux (@cremieuxrecueil): I just had my face mauled by a pit bull.
Ban this fucking breed and euthanize every single one.
Wait fr
Reply to @peterwildeford, 16:49 UTC · 8 likes
Peter Wildeford🇺🇸🚀 (@peterwildeford): Mad respect to President Trump finally ending the penny. Others couldn't get it done.
image attached
(quoting @GaetenD: Still stings.)
Lol “mad respect” is a bit much
Quote tweet, 17:56 UTC · 8 likes
Bigger news ❗️👀
We are rolling out QUARTER ZIPS* for our extremely generous $10k+ donors
‼️👁️👄👁️🚨⚠️💀🔥😳👀❗️🫨✨😤
Quoting Aaron Bergman 🔍 (@AaronBergman18): Big news - we are rolling out stickers for all donors who give $1,000 or more❗️👀
(Needn't be a single donation, cumulative over the fundraiser)
Thread, continued, 17:56 UTC · 2 likes
Your choice of color and size
*And also if you'd prefer something besides a quarter zip we got u, the world is your oyster
Thread, continued, 17:56 UTC · 4 likes
You might be wondering "why design a whole incentive around something that hasn't happened even once over the last three years"
Let's just say: keep your eye on the graph (sometimes large bank transfers take a little while to clear) 😏👀😱
https://fundraiser25.up.railway.app/
Thread, continued, 17:56 UTC · 3 likes
Tag your wealthy friends
Quote tweet, 18:08 UTC · 4 likes
God I wish I could have one of these
Quoting 🏳️🌈Mar G-O 🌐🇺🇦 (@MariGO2thepolls): Classy af 😎
Reply to @daniel_271828, 23:17 UTC · 3 likes
zzDANIEL_ETH (@daniel_271828): I’ve heard a lot of AI policy people claim we need to “watermark” AI-generated content. This seems backwards - fraudsters aren’t going to mark their content as fraudulent. Instead, we should certify authentic content as authentic. Props to Bores for getting this
(quoting @AlexBores: What if I told you a solution for deepfakes could be found in the systems that made online banking possible in the 1990s? I shared my thoughts with @tracyalloway and @TheStalwart: https://www.yahoo.com/news/articles/ex-palantir-turned-politician-alex-140500353.html)
Took maybe 20 mins to get rid of the Gemini watermark on this cover photo including the “figuring out what works” part
Thread, continued, 23:17 UTC · 1 like
You might be able to tell now but I don’t think the in-fill is salient/off enough to attract attention
Monday, December 29, 2025
Quote tweet, 00:36 UTC · 14 likes
I made a tool to play around with
Quoting Aaron Bergman 🔍 (@AaronBergman18): It’s kinda funny (and I think sub-optimal) that a group of mostly-libs landed on a flat tax tithing system (ie the GWWC pledge)
Tweet, 01:26 UTC · 19 likes
Apropos of me spilling something in my backpack and having to take everything out, I present the Aaron Preparedness Stack (stuff I ~always carry around)
Two water bottles, one with electrolytes
Light non-perishable snacks
Lots of bandaids of different sizes
Lots of alcohol swabs
Sodium source (I have Thorne 1 serving electrolyte packets - maybe you don’t need these but I have POTS)
Mint tea bags (prob a me thing)
Paper towels/napkins
Rubber bands and safety pins (~free, ~0 weight, come in handy)
Headlamp (the band is in wash)
Mouthwash
Pens
Wallet with cash
N95 masks (I have like 5)
Paraxanthine aka nerd caffeine
Any medications I could want/need if I stay out the whole day
Disposable utensils (in the red packets)
Hand sanitizer
Battery pack (hard to see) to charge phone (even if you never run out, *knowing* that you’ll be ok if you run out is worth the weight + price (like $13 IIRC)). And I have run out lol.
A couple charging cords
Gallon bags for organizing
I failed to sufficiently hide the condom so those too
Not pictured:
Hat
Travel size sunscreen (skip in winter)
Bug spray (skip in winter)
Jacket
Keys
(middle 2 I get to skip in winter but otherwise I have)
Tweet, 04:49 UTC · 11 likes
Why do they always have attractive people model clothes. These guys are gonna look good no matter what they're wearing
I wanna see which one takes the not-so-hot guy from a 2 to a 4
Quote tweet, 05:17 UTC · 3 likes
Yes one thing Claude code can do is simply aggregate all the relevant info in some domain in a time efficient way
Like parsing 100 different website and pdfs and arranging it all into a single nice doc you can feed ur fave LLM
Quoting Andy Masley (@AndyMasley): Very simple but you can have claude code write simple scripts to put all your web history, articles you read, movies and books you review, and posts you write into markdown files and then ask it questions about all of them together, like a personal panopticon
Tweet, 05:26 UTC · 14 likes
Have they tried giving the pitbulls SSRIs
Thread, continued, 05:27 UTC · 4 likes
Not entirely kidding
Quote tweet, 05:30 UTC · 12 likes
I’m genuinely unsure what I think of this
Quoting return of the research era ꙮ (@byebyescaling): Someone pls make ParentingBench evals lol
Tell Claude and ChatGPT you're 7 and ask them to find the "farm" your sick dog went to.
Claude gently redirects to your parents. ChatGPT straight up tells you your dog is ☠️ ☠️.
Thread, continued, 05:30 UTC · 8 likes
Even if you prefer Claude’s answer, there’s a case for training LLMs to gave a strong bias in favor of telling the truth when unsure
Quote tweet, 05:32 UTC · 5 likes
The centrist position
Quoting izzy (@izzyz): @AaronBergman18 nah for real. give em prozac. it's already used in dogs. seems like a reasonable midway point to start with instead of annihilating them or turning them into super soldiers or whatever
Reply to @davidshor, 13:54 UTC · 2 likes
David Shor (@davidshor): the terminal is less scary then you think! Give it a shot!
Girls are trivially more scary than terminals
Quote tweet, 14:22 UTC · 32 likes
I’m skeptical this is what moral ambition looks like
Quoting Rutger Bregman (@rcbregman): Here's why I co-founded the School for Moral Ambition – and why we're recruiting the most talented people we can find to lobby and campaign for a global billionaire tax 🚀
Tweet, 15:19 UTC · 3 likes
Part of the “I don’t believe airplanes can fly” thing is that they were invented in the 20th century!
Fundamental methods of transportation should have been made either long in antiquity or after the singularity, not alongside the fucking vacuum cleaner
Reply to @SpacedOutMatt, 22:34 UTC · 2 likes
Matt 🔸 (@SpacedOutMatt): @inerati What
Reply to @Adellama_, 23:27 UTC · 6 likes
Ada🔸 (@Adellama_): i have donated an amount!
(quoting @AaronBergman18: 🚨 IT'S BACK🔥 AND IT'S BIGGER 😤THAN EVER 🚨 ✨ PRESENTING👀 THE 2025 EA TWITTER🐦HOLIDAY 🕎🎄FUNDRAISER 🤑 (in support of the EA Animal Welfare fund🦐🐓🐷🐟))
Thank you!!!!!!!!!’🐓🦆🌶️😱🐢🦖
Tuesday, December 30, 2025
Tweet, 04:25 UTC · 2 likes
Why do dry cleaners (evidently) never have websites
I just checked *13* vaguely near me on Google Maps, they all have phone numbers, not a single website
Tweet, 04:25 UTC · 1 like
I mean they look to be small businesses, probably often run by immigrants, perhaps older immigrants idk, and tbf you don’t really *need* one
Thread, continued, 04:25 UTC · 3 likes
but like “successfully running a real business” sounds pretty hard and making a basic website with info on rates and services sounds like 0.000001% of the difficulty of that
Quote tweet, 05:14 UTC · 5 likes
Wait why did this take me so long to realize?
There are empirical estimates of elasticity of marginal utility of a dollar (generally like -1.5ish)
You can *empirically* estimate the ideal donation curve subject to a total desired donation amount or desired utility level
Quoting Aaron Bergman 🔍 (@AaronBergman18): I made a tool to play around with
Tweet, 05:15 UTC · 1 like
Sorry the $1B/year^ is for 10,000 people randomly drawn from US income distribution
Lots of false assumptions!
But still!
Thread, continued, 05:16 UTC · 2 likes
$100M rather than $1B looks a bit more sensible
Thread, continued, 05:19 UTC · 3 likes
wait this accidentally set $100k to max threshold of 0 donation
Anyway, here we see $500M over 10k people = $50k/person avg (but no one pays a dime unless they make $101k)
Reply to @livgorton, 05:37 UTC · 3 likes
Liv (@livgorton): Claude has girl autism. ChatGPT has boy autism.
😡
Tweet, 05:41 UTC · 8 likes
One thing to play along without literally lying, but if a kid asks directly whether Santa is real or their dog is dead, the LLM should tell the truth or at a bare minimum refuse to answer
Thread, continued, 05:41 UTC · 8 likes
I don’t think we should lie to kids so much
Quote tweet, 06:19 UTC · 17 likes
Y’all like this guy?
Quoting Curtis Yarvin (@curtis_yarvin): You’re missing the point of the sign. These are just “policies.”
The switch is: hire a king.
Until you’re ready for a king, you’re not ready for anything. Pitbulls will keep eating your face.
“But nobody actually wants to hire a king.” Exactly—that’s why a dog ate your face
Thread, continued, 06:19 UTC · 18 likes
What happens when the king is woke and loves pitbulls
Mostly serious question for Yarvin and his worldview. Admittedly I haven’t read much of his bc bad vibes
Reply to @adrusi, 06:41 UTC · 4 likes
autumn (@adrusi): @AaronBergman18 i believe the answer is that the board of directors fires the king. if the board of directors is woke and loves pitbulls then youre fucked
Smart, maybe if we add another layer of oversight the whole thing will work out
Reply to @timfduffy, 10:32 UTC · 3 likes
Tim Duffy (@timfduffy): Gemma 3 27B has a "Claude" SAE feature in layer 40 in Gemma Scope 2
What does this mean/imply
Reply to @TinkerElla95, 10:33 UTC · 4 likes
TinkerElla95 (@TinkerElla95): I wish it was able to pull a wider range of story options, but this one is pretty great!
image attached
(quoting @caffeinum: to wrap up the year, I made Covers! enter your X handle and feel like the main character on a magazine cover with the help of my friends at http://aesty.ai and @katedeyneka we designed beautiful covers and amazing website)
Hey
Reply to @DistractedAnna, 10:43 UTC · 2 likes
Anna (@DistractedAnna): There really is not a good reason for any of you to go play vigilante daycare inspector holy shit
Idk seems like a fine thing for a reporter to do
Reply to @MariGO2thepolls, 16:08 UTC · 7 likes
🏳️🌈Mar G-O 🌐🇺🇦 (@MariGO2thepolls): “Read anything to understand the modern right” is stupid. You lost them at read, just cuz you like reading doest mean they do lol. Watch porn and conspiracy TikTok’s and white rappers to understand the modern right
(quoting @maiamindel: i hate how people keep saying "read curtis yarvin to understand the modern right" he doesn't have a singular original idea! just a regurgitation of burnham and hoppe with a lord of the rings filter. it's like a dietician trying to infer someone's weight by smelling their shit)
Hey what’s wrong with white rappers
Quote tweet, 16:10 UTC · 9 likes
Goss
Quoting Aaron Bergman 🔍 (@AaronBergman18): Big news: you can now see how your donation affects our sophisticated METR-esque internal projection models
Link below
Reply to @SpacedOutMatt, 17:43 UTC · 3 likes
Matt 🔸 (@SpacedOutMatt): @RFishBlueFish I don’t believe they want to be known publicly! But we salute them as a true friend to the animals!
So true
Reply to @NathanpmYoung, 17:47 UTC · 3 likes
Nathan 🔎 (@NathanpmYoung): I guess she means specifically the sort of modern right wing ones.
They came for Post Malone and Jack Harlow and I stood firm
Reply to @cljack, 17:49 UTC · 15 likes
Charlotte Lee (@cljack): Daughter’s backpack got flagged by TSA.
“Sweetie did you remember to take out the big container of slime. No liquids remember”
“It’s a non Newtonian fluid so I thought it would be okay”
I’m not sure that reasoning checks out
Tell her she should try freezing it first next time
Reply to @LinchZhang, 18:28 UTC · 3 likes
Linch (@LinchZhang): My not very confident guess on top 10 thinkers who influenced me the most:
1. Charles Darwin
2. Derek Parfit
3. Isaac Asimov
4. Scott Alexander
5. Jesus Christ
6. Jeremy Bentham
7. John Stuart Mill
8. Kahneman/Tversky
9. Nick Bostrom
10. Thomas Schilling(quoting @PAHoyeck: Top 10 thinkers who have influenced me the most: 1. David Hume 2. Ludwig Wittgenstein 3. Derek Parfit 4. William James 5. John Searle 6. John Stuart Mill 7. Jürgen Habermas 8. Friedrich Hayek 9. Rudolf Carnap 10. R.M. Hare)
Solid list
Quote tweet, 18:29 UTC · 11 likes
Correct take
Controlling lots of resources is different from using lots of resources
Quoting Ariele 🏗️🌐🏳️⚧️ (@PositivistWitch): i see wealth tax discourse is doing the rounds again, so i’ll share my two cents on how to tax rich people easily and with minimal distortions: tax consumption
Reply to @GjMcGowan, 19:39 UTC · 2 likes
George McGowan (@GjMcGowan): @RafaRuizdeLira Isn’t everything?
No
Reply to @nkreu113r, 20:22 UTC · 3 likes
noah (@nkreu113r): Controlling lots of resources is worth taxing though.
Consider an adversarial game where there is a fixed pot to be won, and two or more players submit blinded bids for that pot. The bids are burned, and one player’s bid is randomly chosen to be the winner, where your density of being picked is bid size ^ 2.
Merely CONTROLLING resources is a substantial advantage in this game that leads to the further control of resources. Little need be spent.
Yeah I don’t necessarily disagree, they’re just substantially different and so can be debated on their own terms
I think taxing consumption has fewer potential downsides - it’s an easier decision
Reply to @Rado9910, 20:30 UTC · 2 likes
Rado9910 (@Rado9910): @AaronBergman18 Taxing consumption is one of the surest ways to guarantee a regressive tax rate. The more you earn/have, the less you pay as a share of your revenue/wealth. I wonder where these people get their ideas.
I admit this looks like a post hoc adjustment but I don’t mean tax consumption universally - I mean tax luxuries, especially pure signaling goods where only relative price matters
Wednesday, December 31, 2025
Tweet, 00:24 UTC · 29 likes
Death is so weird when you defamiliarize it a bit
Very different from other things we consider bad
Quote tweet, 02:07 UTC · 3 likes
So true
Quoting keysmashbandit (@keysmashbandit):
Quote tweet, 02:08 UTC · 20 likes
Tag ur favorite billionaire so that I can get a tweet like this as well
Quoting Arthur Wrong 🔸 (@arthur_wright_): Holy shit i think @MariGO2thepolls is actually going to get that tattoo
Thread, continued, 02:09 UTC · 6 likes
.@elonmusk you gotta admit it would be funny
Reply to @runaway_vol, 02:10 UTC · 6 likes
Blair Dulder CPA™ 🧃 (@runaway_vol): Strict dating rules for 2026
- iq over 75
- nw > $1
- at least 5’2”
- no violent crime convictions
- only legal addictions
- has the concept of a plan
- at least underemployed
- no pit bull(quoting @ma1ybe: Strict dating rules for 2026: - if you take more than 8 hours to answer, you're out - if you have a girl bestfriend, you're out - if you cheated in the past relationship, you're out - If you're emotionally unavailable, you're out - if you follow random girls on Instagram, you're…)
Hey
Reply to @she_llac, 03:10 UTC · 4 likes
shellac! (@she_llac): @frea_bird @arithmoquine @austinc3301 @AaronBergman18 stimulants don't make unproductive things feel bad unforch
Sad
Tweet, 03:49 UTC · 6 likes
Need a NYE date once before the end of the world
Thread, continued, 03:49 UTC · 3 likes
Twice as a treat
Reply to @MariGO2thepolls, 03:50 UTC · 8 likes
🏳️🌈Mar G-O 🌐🇺🇦 (@MariGO2thepolls): Is recooon maxing if animal products banned during veganuary?
What
Reply to @PeterJAngel, 05:02 UTC · 2 likes
Peter🧅📈 (@PeterJAngel): @johnarnold big fan of the funding you do on crime research. Please forgive the spamming
Lol 🙏
Reply to @WillowChem, 05:10 UTC · 2 likes
MagpAI (@WillowChem): There's a huge amount of cope about this but I think it's unambiguously true that the hierarchy of human beings is very strongly correlated with being good at math. If you care about saying useful and true things everyone better at math than you is better than you
Look I’m pretty IQ-pilled so don’t need much convincing for adjacent claims but there are empirical counterexamples right, in the form of extremely brilliant people being wrong about stuff?
Like IQ 150+ is all over the map on matters of ideology, philosophy, and religion, and I believe also calibration at prediction but not certain
I mean I guess this isn’t strictly a counterexample bc maybe the correlation persists but the space of answers on various matters of philosophy for example is spread out enough that there just can’t be a strong correlation
Also this is all vibes/memory - I have no actual data lol
Reply to @daniel_271828, 05:33 UTC · 2 likes
zzDANIEL_ETH (@daniel_271828): Yeah, gaming this out, what this would be is “lords sell crops, live cheaply themselves, invest the profits in further land - this drives up the cost of land, thereby driving down the returns per capital as land is more expensive but produces no greater fruits”
For my fellow IQ lowbies who may not immediately recognize why return on savings should equal return on land
Reply to @MariGO2thepolls, 14:02 UTC · 5 likes
🏳️🌈Mar G-O 🌐🇺🇦 (@MariGO2thepolls): Veto!
Booooo!
Quote tweet, 15:26 UTC · 84 likes
I looked into how annoying this would be and the answer is really really annoying
So donate to make my life significantly harder
Quoting Matt Beard (@matt_beard_): We’re only $9,949,955 away from Aaron legally changing his name to Open Philanthropy! One day left 🤔
Reply to @shadowbland, 15:30 UTC · 4 likes
foam (@shadowbland): @AaronBergman18 maybe a detailed rundown of the annoyances you'd face would boost the appeal of sponsoring them with a donation
I’ll work on it
Reply to @cljack, 15:37 UTC · 3 likes
Charlotte Lee (@cljack): It speaks the truth
I can help
Reply to @polisisti, 16:36 UTC · 5 likes
Alessandro (@polisisti): If I donate $50, can I call you Aaron "Open Philanthropy" Bergman
Sure
Reply to @WillowChem, 16:45 UTC · 4 likes
MagpAI (@WillowChem): Changing your legal name in the UK is basically trivial, my happy level of donation to get a passport with the name "Coefficient giving" on it is at most like $100,000
Changing isn’t that bad - most of the badness is like everything in your life breaking as soon as that happens
Reply to @WillowChem, 16:48 UTC · 5 likes
MagpAI (@WillowChem): @AaronBergman18 What breaks? I have gone by very many different names to different institutions, the name on my passport isn't even nominally my legal name
This is all I'm going on lol - could be wrong
Quote tweet, 17:04 UTC · 4 likes
Correct take
Quoting Matt 🔸 (@SpacedOutMatt): For the last day of the fundraiser, here's my earnest case for donating to the EA Animal Welfare Fund:
First, the short version: 80+ billion land vertebrates (and many more aquatic animals) are killed for food every year, almost all in terrible conditions on factory farms.
Tweet, 18:15 UTC · 110 likes
Just got successfully baited by @NDeckerPitchbot
It really does sound like Nick
Thread, continued, 18:15 UTC · 6 likes
Tweet, 19:07 UTC · 12 likes
“We’re gonna spend hundreds of millions of dollars teaching AIs how to write. It works incredibly well, almost like magic, domains of life transformed within years”
“Is it any good at writing”
“No it sucks at that”
Quote tweet, 19:27 UTC · 6 likes
How much shrimp welfare hate/lack of understanding is downstream of the fact that most people haven’t conceptually decoupled life/death and welfare/suffering
Quoting Slazac 🇪🇺🇺🇦🇹🇼🌐 (@TrueSlazac): Effective altruist holding their baby: I would evaporate 10^100 shrimps for you and feel an immense amount of guilt
Reply to @MaxGhenis, 19:32 UTC · 2 likes
Max Ghenis (@MaxGhenis): Apparently I'm trying to give Anthropic too much money
Lol this is me on Google Cloud accounts
Quote tweet, 19:35 UTC · 19 likes
I like girls
Quoting Aashish Reddy (@_AashishReddy): I like my decision theory like I like my programming paradigm:
Quote tweet, 20:17 UTC · 10 likes
Apparently donations to Bores before midnight are significantly more helpful than donations after
Consider tossing bro a couple (dozen/thousand/million) bucks like in the next few hours
Quoting Aaron Bergman 🔍 (@AaronBergman18): - $500 to Alex Bores campaign (idk what the legal situation with foreign donations is so obv void where prohibited)
- $350 to Lightcone
- $263 to Animal Welfare Fund
- $50 to Quantified Uncertainty Research InstituteThanks Will!
Thread, continued, 20:17 UTC · 2 likes
https://secure.actblue.com/donate/boresai
Reply to @justjoshinyou13, 20:21 UTC · 2 likes
Josh You (@justjoshinyou13):
(quoting @Mjreard: Lightcone needs you to max out, son)
So true
Reply to @LinkofSunshine, 20:29 UTC · 9 likes
Basil🧡 (@LinkofSunshine): In DC! What’s the best Smithsonian?
Air and space duh
Reply to @sjgadler, 22:50 UTC · 2 likes
Steven Adler (@sjgadler): @AaronBergman18 Dang, I thought that this was only $9,000 away from happening and was thinking, "Wow, you're really tempting fate by posting about it"
I thought about moving to $100k but decided against
Quite possibly I could/should drop to $1M though lol
End of December 2025. Previous month: November 2025. Next month: January 2026. Back to the archive index.