Aaron Bergman's tweets, July 2026
Every public tweet, reply, quote tweet and retweet that Aaron Bergman (@AaronBergman18 on X, formerly Twitter) posted in July 2026 and has not deleted: 696 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: June 2026. Next month: August 2026.
Wednesday, July 1, 2026
Reply to @RichardHanania, 00:19 UTC · 12 likes
Richard Hanania (@RichardHanania): https://www.richardhanania.com/p/why-you-shouldnt-worry-about-ai-taking
I think the fundamental question here is whether humans demand an infinite amount of human-only stuff in the way that we desire an infinite amount of non-food stuff (cc farming jobs)
I don’t think the “yes” thesis is crazy but i do think that it’s wrong, and even if it’s right the world can look extremely bleak depending on the relative value of $1 of AI-economy output vs $1 of human labor (to simplify a bit)
Imo the “moderate” view here is merely that wages go down by a lot. Im not certain enough about this stuff to die on the hill of “absolutely maximally bad outcome for those without capital” but that’s of little comfort to the person making $1/hour (it does get very complex trying to figure out the equilibrium of what $1 buys in this story)
IMO you should push for the redistribution institutions that make this story a genuinely good one on the right
Quote tweet, 00:33 UTC · 18 likes
Based on responses it seems that many people cannot entertain the notion of simply fucking up and not in a “sorry because I was caught” way and then genuinely apologizing for your mistake
Quoting Brian Stelter (@brianstelter): Totenberg read the text of the apology message she sent to Justice Alito:
"Dear Justice Alito, there are no words to adequately apologize for today's error in reporting your retirement. It was entirely my fault. I rushed out of the courtroom after the opinion announcements, and when I realized that the usual rush of folks after a few minutes had not happened, I asked somebody was going on inside, to which the answer was, 'retirement announcements.' I didn't hear the 's' on 'announcements,' and I assumed something no reporter should ever do, that you were retiring. It was the worst professional mistake of my more than 50 years in journalism. I could go on, but I don't know what else to say, except that I am so, so sorry."
Quote tweet, 00:35 UTC · 114 likes
I admit that this breaks my model of things a bit
I expected it to take much longer, more substantially cutting into revenue as retribution for not bowing down to the Trump administration
Quoting Anthropic (@AnthropicAI): We’ve received notice that the Department of Commerce has lifted export controls on Claude Fable 5 and Mythos 5.
We'll begin restoring access tomorrow, and will share an update soon.
We’re grateful to our users for their patience, and to everyone who worked with us on redeploying the models.
Reply to his own earlier tweet, 00:35 UTC · 6 likes
Aaron Bergman, July 1, 2026: Based on responses it seems that many people cannot entertain the notion of simply fucking up and not in a “sorry because I was caught” way and then genuinely…
Not something I’d ever do btw bc I never make mistakes
Quote tweet, 00:47 UTC · 223 likes
Wait the literal answer seems to be “14 step gateway drug”
It makes your skin more sensitive to the sun, which means you have to use sunscreen much more at least on your face,
Quoting Aaron Bergman 🔍 (@AaronBergman18): Should I be using retinol or is it just a gateway drug to 14-step skincare routines
Thread, continued, 00:47 UTC · 41 likes
which you cover up with makeup, and try to get rid of all this stuff at the end of the day with “cleanser” but inevitably that itself contains random things that will seep into your skin and is imperfect at its job
Thread, continued, 00:47 UTC · 25 likes
so all of a sudden you’ve accumulated the union of all ingredients in your retinol cream, sunscreen, makeup, and cleanser product,
Thread, continued, 00:47 UTC · 26 likes
which sounds like the kind of thing that would result in random issues like dryness or minor allergic reactions or irritation or god knows what that warrant more additions to the skincare game
Thread, continued, 00:47 UTC · 20 likes
Just say NO to retinol
Reply to @caryatis, 02:43 UTC · 6 likes
Kayla (@caryatis): @AaronBergman18 You should be using sunscreen regardless of retinol sorry
I know and I do but on the margin it does matter to me how sensitive I am, and if the story is "truly every day" instead of "when I'm going outside and it's sunny for at least 10 mins" that's a substantial delta
Reply to @AndrewSabisky, 05:32 UTC · 5 likes
Andrew Sabisky (@AndrewSabisky): Effective Altruism strikes again! An incredible win for the chip export control people, who have finally managed to persuade Chinese firms to knuckle down and learn how to use Huawei's crappy software. First big/good Chinese model trained on local compute
(quoting @poezhao0605: Meituan, China's largest food delivery platform, open-sourced a 1.6 trillion parameter AI model today. LongCat-2.0 was trained from scratch on a 50,000-card domestic GPU cluster. Native 1M context. 30T+ tokens of pretraining data. A delivery company. Trillion parameters. Chinese…)
Wait what is the critique here? It would obviously be easier and cheaper to do the same with NVIDIA?
Reply to @tafphorisms, 05:34 UTC · 2 likes
taf 🝮 (@tafphorisms): @AaronBergman18 Lmao??? How is that your takeaway???
1. You don’t need to wear makeup to cover up sunscreen.
2. You’re describing a three step skincare routine where two of the steps are things you should be doing already (wearing sunscreen and washing your face).
I’m describing a 3 step entrance to the rabbit hole (and yes I do the other two lol)
I feel like I never hear of people only doing these 3 things but admittedly I’m not looking very hard
Reply to @NathanpmYoung, 05:38 UTC · 14 likes
Nathan 🔎 (@NathanpmYoung): Worth considering:
(quoting @AndrewSabisky: Effective Altruism strikes again! An incredible win for the chip export control people, who have finally managed to persuade Chinese firms to knuckle down and learn how to use Huawei's crappy software. First big/good Chinese model trained on local compute)
Is it? Like obviously this is a consequence? Like yes I Aaron if export restricted from Apple would get used to windows/linux but that’s not secretly an advantage even if all else equal knowing the new OS well is good. But all else isn’t equal!
Narrowing the option space for Chinese companies doesn’t make them magically better off
Reply to @AndrewSabisky, 05:43 UTC · 2 likes
Andrew Sabisky (@AndrewSabisky):
I bite the bullet on the first (think it’s a worthwhile tradeoff and something that can be conceded later/used as leverage)
The second doesn’t make much sense to me. Sure they’re becoming less dependent but as a side effect of the direct effect of restricting an important input - there are more complicated stories you can tell about why this will actually be good for China but the basic impact on companies is almost trivially negative (or neutral) bc it’s making the option space smaller for them. They always had the ability not to buy NVIDIA!
Tweet, 08:44 UTC · 6 likes
5-4 on constitutional question in Trump v Barbara is essentially as nuts as everyone is saying
Thread, continued, 08:44 UTC · 5 likes
I do think it’s worth noting that it’s like a 5 minute reasoning question instead of a 5 second one (thanks to “subject to the jurisdiction thereof”) but afaict the dissents are just somewhere between blatantly fallacious and totally implausible
Thread, continued, 08:44 UTC · 5 likes
Tbc they all stand as intellectual infrastructure for a moral and ideological justification of *changing the 14th amendment* but they don’t actually *engage with the 14th amendment as it stands*
Thread, continued, 08:44 UTC · 3 likes
A lot of fancy wording built on top of the Cheeto that is the implicit claim “a person is not subject to the jurisdiction of the US if they’re also subject to the jurisdiction of another country”
Thread, continued, 08:44 UTC · 2 likes
The contrary would imply that eg a dual citizen of US and UK was a sovereign not subject to the laws of the US even when in the US
Thread, continued, 08:44 UTC · 2 likes
Painting an already pretty black pill even blacker
Quote tweet, 15:36 UTC · 17 likes
This but the opposite
Quoting Jonny Spicer🔸 (@jjspicer): is there anything more nightmarish than meeting someone for the first time and having them say “ah, I know who you are from twitter”?
Reply to @jjspicer, 16:02 UTC · 3 likes
Jonny Spicer🔸 (@jjspicer): Nothing more nightmarish than meeting someone on twitter and having them say “ah, I know who you are from real life”? 😅
Lol no I love when ppl recognize me from twitter
Reply to @jjspicer, 16:08 UTC · 2 likes
Jonny Spicer🔸 (@jjspicer): I just always assume that they have perfect recall of whatever my most cringe and neurotic tweet was
I gotta ride the tiger
Quote tweet, 16:10 UTC · 10 likes
Fellas this is some of you
Quoting Charlotte Lee (@cljack): all I want is a slightly younger, brilliant, ambitious autist who obsessively loves me and will crown me Empress, is that really too much to ask?!
Reply to @PradyuPrasad, 16:17 UTC · 4 likes
Pradyumna 🇸🇬 (@PradyuPrasad): isn't it you aaron
I cannot confirm or deny
Quote tweet, 16:20 UTC · 6 likes
Wait yooooooo 🎉
Quoting Hakeem Jeffries (@hakeemjeffries): Congratulations to State Rep. @MannyRutinel on your hard-earned victory.
MAGA extremist Gabe Evans is a central architect of the Medicaid cuts causing pain for Colorado families.
Manny has what it takes to flip this critical seat and deliver real results for Coloradans!
Reply to @daniellellecco, 16:21 UTC · 2 likes
(The tweet being replied to by @daniellellecco is no longer available.)
Ppl need to chill fr
Reply to @tucker_peck, 20:22 UTC · 13 likes
Dr. Tucker Peck (@tucker_peck): Colleague: What population do you most often treat?
Me: High functioning autistic young self-made millionaires worried about AI, with an IQ 130-140 but they think it's 190.
This gets u a follow
Thursday, July 2, 2026
Reply to @graveunfurling, 01:01 UTC · 3 likes
may💐 (@graveunfurling): constantly telling men with hot, sexy big noses how nice they are. so far im batting 0 bc they all get cagey and sad afterwards (at least from what i can see)
(quoting @cavempty_: hip dip discourse made me think of a time a girl was laying on my chest and i mentioned how i have a rib that sticks out way more on one side and ive always hated it and she told me it made me more ergonomic to lay on. what if everyone was nice about people’s insecurities)
Keep at it
Reply to @afoidance, 01:03 UTC · 3 likes
ren 🇮🇱 (@afoidance): Zoomer guys are cooked they won’t even mansplain to girls anymore. This is why our birth rates are dropping
(quoting @johnsmithyson0: No)
What do u want explained
Reply to @nonhumanminds, 01:06 UTC · 2 likes
NYU Center for Mind, Ethics, and Policy (@nonhumanminds): CMEP and @eleosai are thrilled to release our new report, "Studying AI Welfare Empirically."
Following our 2024 report "Taking AI Welfare Seriously," we kept hearing the same question: Even if AI welfare matters, how can we know anything about it?
This report is our answer.
Goss
Reply to @xwanyex, 01:30 UTC · 12 likes
wanye (@xwanyex): I very much doubt that Matt Walsh hates or has contempt for women. I think you should consider that possibility, too, and try to figure out how you might reconcile that fact with his statements.
You can’t take this whole thing in the abstract, it’s relevant that the statement *isn’t true*
Specific counterexample: Kagan vs this guy. IMO you’d have to bend over backwards pretty hard particularly as a conservative to say that Douglas > Kagan
https://en.wikipedia.org/wiki/William_O._Douglas
Reply to @DerekPederson3, 01:34 UTC · 2 likes
Derek Pederson 🇺🇸🇺🇦🇻🇪 (@DerekPederson3): Trees have standing
They do stand tbf
Reply to @JayShooster, 03:22 UTC · 2 likes
Jay Shooster (@JayShooster): update: lfg
(quoting @JayShooster: I'm so excited to announce the next big project I'm working on: getting my dear friend Manny Rutinel elected to the U.S. Congress and flipping the House of Representatives back to the Democratic Party! Couldn't be prouder to be the Chair of Manny's finance committee. Let's go!)
Hell yeah 🫡
Quote tweet, 03:27 UTC · 14 likes
Err on the side of cringe
Quoting Jeremiah Johnson 🌐 (@JeremiahDJohns): Kelsey's thread about the seething contempt for women is worth reading, but I want to emphasize this point.
It's not lame to care. You should care desperately about things, you should be passionate, you should give a shit. There's been an epidemic of detached irony over the last several decades. Society spent years telling us that people who care are nerds, the cool kids are the ones who are above it all. "haha you're so mad" as though that's a sign of weakness.
To hell with that. Develop attachments and care about things and do it with sincerity in your heart. You'll be a better and more complete human if you do.
Reply to @antoniogm, 07:12 UTC · 10 likes
Antonio García Martínez (agm.eth) (@antoniogm): Because they're mostly lies, or just invented family lore.
Almost nobody's name got changed at Ellis Island: they had clerks who spoke the languages, and ship's manifests recorded passenger names linked to required travel documents.
What happened is that families changed their names after arrival to fit in more, to combat whatever flavor of bigotry regnant then (which included a lot of anti-Semitism).
@DaraHorn researched the name-change court records, and how it all went down in her excellent book "People Love Dead Jews".
(quoting @uncle_deluge: I don't understand the stories that go "our family name was xyz but the immigration clerk misheard it and wrote something else at Ellis Island" and what they just went with that? "Oh I guess we're called this now, and we'll also call our children that" complete and utter nonsense)
I don’t necessarily disbelieve this but FWIW: my family has the ship manifest from when a great grandfather and his parents came here (not sure how exactly lol) and the writing is ~indecipherable including to LLMs as of a few months ago
Insofar as manifest writing is a load bearing mechanism (idk if it is) the mistranslation story sounds hard to *disbelieve*
N=1 manifest ofc
Reply to @antoniogm, 07:19 UTC · 2 likes
Antonio García Martínez (agm.eth) (@antoniogm): Bad handwriting might explain a missing letter here or there, but it doesn't explain Grünberg->Green, Moskowitz->Moss, etc.
Those were intentional Anglicizations done later.
Ah yes this seems right
Thread, continued, 07:19 UTC · 2 likes
The story is that we went from Berkman to Bergman but I haven’t actually looked into this beyond hearing the lore
Reply to @AaronEstel, 07:47 UTC · 3 likes
Aaron (@AaronEstel): ah I know this state well
(quoting @Ggod323: )
Yeah
Reply to @NoahTopper, 16:21 UTC · 4 likes
Quinoah 🔍⏸️ (@NoahTopper):
image attached
(quoting @UberFacts: Today is Social Media Day — What social media post will forever remain unmatched?)
Nice
Reply to @lapislagoons, 17:08 UTC · 3 likes
˗ˏˋ ´ˎ˗ (@lapislagoons): there’s no unmarried people at this wedding except me so I’m tasking all of them with finding me a husband
Say “workflow” to get them to work efficiently in parallel
Reply to @bellaforristal, 17:11 UTC · 5 likes
Bella Forristal 🔸 (@bellaforristal): Oh my god. Google Docs is now giving me AI suggestions every 3 characters, which makes it impossible to think about what I'm saying cause random words keep flashing up.
Why??? What?? If this can't be turned off I will legit not draft things in Google Docs anymore ahahahaha
Simply vibecode ur own Google Docs with fable (not entirely kidding)
(I think there’s a way to turn this off)
Reply to @afoidance, 18:38 UTC · 9 likes
ren 🇮🇱 (@afoidance): the reason some of you don't want to be intellectually stimulated by a woman is because you're fucking stupid actually
Allegedly normie men don’t like women being smart. I’m sympathetic to the “don’t want a partner who mogs me on any important dimension” thing but apparently it’s not even that, they just directly don’t like it. Wild
Reply to @halogen1048576, 18:38 UTC · 2 likes
🇨🇦halogen (@halogen1048576): if i want stimulation i can smoke meth
(gif, plays on X)
Reply to @McMasterDaily, 18:41 UTC · 5 likes
Daily McMaster-Carr (@McMasterDaily): Enclosure Air Conditioner
Wall-Mount, 120V AC, 1000 Btu/hr. Cooling Capacity
$2,271.19 each
17495K19If you're hot, they're hot. Keep your enclosures cool!
Temperature range bottoms out at 68°?????? Fake ass AC
Tweet, 20:59 UTC · 12 likes
Is there black market Lyme vaccine
Tweet, 21:11 UTC · 78 likes
Going on a hot girl walk (going on a walk while it’s hot thinking about how wonderful girls are)
Quote tweet, 21:14 UTC · 6 likes
The government literally has the power to tax and spend, we don’t need workarounds like this. Just transfer cash to people
Quoting Basil🧡 (@LinkofSunshine): 485-x was created to incentivize building affordable units with tax credits
However, this got hijacked, and 100+ unit buildings that use 485-x require $75 minimum wages for their workers
This meant not a single 100+ unit building has used 485-x.
This is how policies fail.
Reply to @funplings, 21:36 UTC · 3 likes
funplings (@funplings): from @earnestpost : “i dont like guys that are too empathetic. they dont have emotional depth.”
(quoting @MomaBera: “Nooo I will exhaust the pool of eligible rat men first” Overhearing @earnestpost and @funplings IRL dating discourse)
wtf
Tweet, 21:58 UTC · 37 likes
Cost of running a bedroom-sized window unit on high for a whole day
I feel like people think that AC is more expensive than it really is
Not utterly trivial but it’s hard to imagine a better way of spending ~$100/mo max when it’s warm
Friday, July 3, 2026
Reply to @littIeramblings, 04:02 UTC · 2 likes
sarah (@littIeramblings): very humbling to be confronted with my nervous laughter habit while reviewing a podcast transcript (all of these instances are me)
Omg what podcast 👀👀
Reply to @chaotictransfem, 04:13 UTC · 3 likes
rachel📍in sf oct 5-10 (@chaotictransfem): okay unfortunate issue here is that pangram is too good for this task... the model just immediately maxes out the score as 100% ai haha, pangram is so confident/accurate immediately that it can't provide good signal on how to become more AI
(quoting @chaotictransfem: currently RL-ing a model against pangram to find out what happens if you maximize slop)
Is it actually 100% or is it just rounded to that integer
Reply to @chaotictransfem, 04:16 UTC · 2 likes
rachel📍in sf oct 5-10 (@chaotictransfem): i'm maxing out at around 99.4% where there seems to be no more signal for training haha.
This is actually a fun problem, you could try training a model to go from x% to 99.4% (ie translating text into slop) and see if it generalizes when fed a 99.4% text
Reply to @0interestrates, 06:49 UTC · 3 likes
rahul (@0interestrates): guy selling post trained oss models: if you don’t own your models you don’t own anything
guy selling harness on top of models: if you rely on one model you are cooked
guy selling tokens: instead of manually prompting your models just put them in a loop where they prompt themselves
The loop thing is genuinely great
Reply to @cynlioness, 15:00 UTC · 3 likes
Cyn 🔸⏸️ (@cynlioness): The print shop I thought was cool cancelled my order 😔
(quoting @cynlioness: I just purchased a new mtg strip cube and I'm so frickin excited to play it and use the bait of sexiness to get people to actually play well balanced magic with me)
😡
Reply to @ShadowyZephyr, 19:30 UTC · 2 likes
zephyr (@ShadowyZephyr): true
🤝
Quote tweet, 20:07 UTC · 6 likes
Planes aren’t real
Quoting Daniel (@growing_daniel): Sure it cost like a bajillion dollars but now that we have them, F35s are insanely cool
Tweet, 21:00 UTC · 24 likes
I’m not sure how common this belief is is but I rarely hear it expressed directly: the world seems really bad, full stop
I don’t think attitudes or vibes can be “correct” in the way a proposition is, but resent of ✨all this✨ seems morally and epistemically productive, at least
Thread, continued, 21:00 UTC · 10 likes
Imagine being thrown into a room and being forced to draw a slip of paper from a hat that determines your fate: happy human, very happy human, nameless fish living in 295959549 BC, elephant, farmed shrimp, holocaust victim, etc
Thread, continued, 21:00 UTC · 10 likes
Even before you know the probabilities of each outcome, the fact that you’ve been signed up for the world’s highest stakes lottery by someone or something else, with some outcomes unimaginably bad, seems like good reason for resent directed at the lottery itself
Thread, continued, 21:00 UTC · 7 likes
This is of course the actual situation we find ourselves in, as sentient beings in the universe or multiverse
Tweet, 21:13 UTC · 21 likes
Going to a normie charity gala/ball thing tn
Reply to @Rado9910, 21:15 UTC · 9 likes
Rado9910 (@Rado9910): @AaronBergman18 David Pearce offered a solution that seems to work: literally reprogram the entire biosphere for happiness instead of suffering. See his book "Can Biotechnology Abolish Suffering?" It sounds crazy at first, but once you read the book, you see that it's perfectly achievable.
Yeah towards that future 🚀
Reply to @savvypleb, 21:18 UTC · 8 likes
(The tweet being replied to by @savvypleb is no longer available.)
Intellectual curiosity
Reply to @justalexoki, 23:49 UTC · 10 likes
taoki (@justalexoki): january, 168lbs july, 141lbs
Dang lookin good. Height?
Saturday, July 4, 2026
Reply to @adastroworld, 01:19 UTC · 3 likes
adas🧦🌹 (@adastroworld): @AaronBergman18 Fit pic? Tie or no tie?
No tie bc i can’t get it to look right 😢
I should have kept this iteration but thought it was a bit long. Also dw I got rid of the AirPods lol
Reply, 01:22 UTC · 3 likes
(The tweet being replied to is no longer available.)
Hit me
Reply, 03:26 UTC · 1 like
Fuck, thanks for feedback
Reply to @she_llac, 22:28 UTC · 2 likes
shellac! (@she_llac): oh neat that reset my limits fully to zero, i was expecting to get only the difference
This $200/mo for me is one of the best deals I’ve ever snagged
Reply to @she_llac, 22:29 UTC · 1 like
shellac! (@she_llac): i just hope they don't actually yank it on tuesday, i am *not* ever paying the ridiculous api prices
Idk if I’d count on that lol
Tweet, 22:47 UTC · 34 likes
Idk how productive high level “how you should view America” discourse is but an underrated view seems to be “last, best, and only good shot”
It’s AGI -> ASI szn, there’s no other superpower to rise to the occasion, it’s America or the Chinese state
Thread, continued, 22:47 UTC · 10 likes
Good or bad in absolute terms, one of these is better than the other all things considered, and that’s the one that celebrates July 4 🇺🇸
Thread, continued, 22:49 UTC · 4 likes
I say “Chinese State” as an intermediate term between “China” broadly and “the CCP”
The analogue to the CCP isn’t the USA, it’s closer to the Trump Administration
Reply to @RSPY_critical, 22:59 UTC · 3 likes
RSPY (@RSPY_critical): I just picked up a burger from 5 Guys, and saw a very attractive man my age eating alone.
I looked at him and laughed “happy fourth”.
He looked happy and sad like he knew what I meant.
Neither of us shot our shot.
But if I was emotionally available, I prolly woulda….
But then again, probably wouldn’t be at 5 Guys on a holiday if I were emotionally available lol.
Sigh…
😢
Reply to @hyperdiscogirl, 23:22 UTC · 3 likes
hyper(bolic) disco(unting) girl in the bay (@hyperdiscogirl): lost ~3lbs/month over the past 4 months on 0.1mg of semaglutide (1/2 normal starting dose). no discernible side effects. eating more nutrient dense diet than mb ever before (mostly dairy, fish, beef, veg). maintaining light weightlifting schedule and walking. happy with it!
Based
Quote tweet, 23:28 UTC · 24 likes
This is how you know all of my takes are correct
Quoting Stefan Schubert (@StefanFSchubert): "Enhanced rationality in autism spectrum disorder
In the domains of reasoning, judgment and decision-making, individuals with autism spectrum disorder often show ‘enhanced rationality’ by exhibiting more rational and bias-free decision-making than do neurotypical individuals"
Sunday, July 5, 2026
Reply to @CoughsOnWombats, 02:19 UTC · 2 likes
Sawyer 🔍, ⏸️/⏹️ (@CoughsOnWombats): Nonsense, he was clearly unhappy.
Since we live in a fundamentally just world, we can infer that he did something to deserve it.
QED.
Truly just this reasoning all the way down
Reply to @AndyNosretep, 06:00 UTC · 4 likes
JugoBall💙♥️😎⚽ (@AndyNosretep): Critics praise entrepreneurs when they give money away, but condemn the process that made the money possible.
That is backward.
The primary social contribution of an entrepreneur occurs when customers gain, workers earn, suppliers sell and competitors improve.
This is only true because 99.9% of them suck at philanthropy
Reply to @cynlioness, 17:36 UTC · 6 likes
Cyn 🔸⏸️ (@cynlioness): How do you feel about therapy? Assume the person has the bandwidth to do a little shopping around / maybe one therapist switch if the first isn't working for them.
"I would recommend starting therapy to X% of people"
Like 10% probably, the current obsession with it as a thing that many or most people should do seems a bit nuts to me
Monday, July 6, 2026
Reply to @afoidance, 04:31 UTC · 4 likes
ren 🇮🇱 (@afoidance): 5’8 is the ideal male height
Close
Reply, 06:22 UTC · 29 likes
Macrophysiological System🐀 (@InVitroFuture): Call me a clueless white guy but it did not occur to me that a significant contributor to "why are all the Black kids sitting together" is that there's a Black meme wherein if you make friends with white people, they'll murder you for fun and the police will refuse to investigate
(quoting @CurttiMagurtti: so this 18yo boy went out boating with his “friends” ystday. everyone made it back safely except him.. & y’all know exactly which one it is. yes, the only black boy. ts don’t even add up to me.)
What
Reply to @mattyglesias, 06:33 UTC · 2 likes
Matthew Yglesias (@mattyglesias): I wish folks would think more about what’s be involved in constructing an electricity grid such that you never get conservation exhortations during outlier weather — you can build redundant capacity, but then your average bills will go up and people will be mad about that.
“Redundant capacity” seems pretty good to me, don’t we have redundant capacity for everything else? Buildings aren’t a 1% error somewhere from falling over afaik, planes carry excess fuel, hospitals have generators
Reply to @funplings, 14:14 UTC · 2 likes
funplings (@funplings): @bashu_thanks eating that pussy like corn on the cob
Lmao
Reply to @NoahTopper, 18:23 UTC · 2 likes
Quinoah 🔍⏸️ (@NoahTopper): easiest “us” text ever, boys
(quoting @UberFacts: Today is International Kissing Day The longest kiss ever recorded lasted 58 hours and 35 minutes)
Us, Noah
Tweet, 18:37 UTC · 11 likes
There are too many things to read!!!!! Stop writing things until I’ve caught up!!!!!!
Reply to @MichaelAlleged, 20:03 UTC · 2 likes
Michael (Allegedly) (@MichaelAlleged): in twelve hours, the spacecraft I have worked on from blank whiteboard to production will get sent into space. I’ll be on console at mission controls.
Wish me luck.
Woah sick!!
Reply to @gptbrooke, 20:10 UTC · 8 likes
brooke bowman (@gptbrooke): can someone at anthropic tell me when claude dating is going to happen
i want my claude to talk to other claudes and suggest matches. fable especially i think would have a good read
Yes, and this goes pretty deep because you can have information exchange between claudes that neither human wants the other human to know, at least in theory or if you assume willingness to not fuck with the system adversarially
Ofc on top of genuinely intelligent matching (including accounting for attractiveness via image understanding)
Reply to @LeRoyDesCimes, 20:12 UTC · 4 likes
Louise 🔸 (@LeRoyDesCimes): using Claude Fable from here ☀️💖
A fellow 16 inch respecter
Reply to @moonbeeaam, 20:22 UTC · 5 likes
anna (@moonbeeaam): I have an unsubstantiated theory that when people nowadays talk about cortisol and serotonin, it is just as unscientific as when people used to talk about the four humours
I think folk-cortisol is more legit than folk-serotonin
Tuesday, July 7, 2026
Reply to @SALTWATERSOPH, 04:37 UTC · 4 likes
Sophie (@SALTWATERSOPH): anyone know how to make the endless void go away 👋👋
The full depression playbook: blood tests -> address any physiological issues -> lowest risk/commitment drugs -> escalating risk/commitment + adjuncts
Also CBT allegedly
Psychedelics/MDMA somewhere in the escalation flow
Tweet, 06:08 UTC · 12 likes
I watched a whole (fictional) movie today like a normal cultured person plz clap
Quote tweet, 07:30 UTC · 181 likes
My contribution to this discourse is that people will often present a bunch of reasonable and legitimate reasons why dating a coworker is bad all else equal but then totally forget to account for “you can get a happy relationship out of it”
Quoting Karol Markowicz (@karol): Was MeToo just a fever dream?
Reply to @Noahpinion, 07:34 UTC · 6 likes
Noah Smith 🐇🇺🇸🇺🇦🇹🇼 (@Noahpinion): Israel is kind of a nasty regime and not a great ally, but the people who make Israel-hate their main political issue are almost always creeps
I think you mixed up the order of the two clauses
“Nasty regime” is the more important one and should come after “but”
Reply to @ChristianityOn, 15:32 UTC · 3 likes
Christianity On The Spectrum (@ChristianityOn): at some point in every autistic person's life they must learn the hard lesson that they are not actually a member of the bene gesserit seeing the social fabric of reality in ways no one else can, but are instead kind of a boring guy with a neurodevelopmental disorder
😢
Reply to @JustinSadur, 15:33 UTC · 2 likes
Sasha Torres (@JustinSadur): The fact of the matter is that human beings naturally date within their social circle and for a lot of us that heavily overlaps with our coworkers. Expecting humans to not do this is just... unnatural.
Ya 💯
Reply to @afoidance, 15:39 UTC · 4 likes
ren 🇮🇱 (@afoidance): everyone says they like autistic girls until i pull up and apparently i’m a heartless bitch who should die
😢
Reply to @AnnaLeptikon, 15:49 UTC · 2 likes
Anna Riedl (@AnnaLeptikon): Many of the cruelest people in history thought they were on a mission for goodness.
(quoting @Freyy_is: i need a book written entirely from the villain’s point of view… except you don’t realize they’re the villain until the last few chapters.)
I wonder how true this is
Do very competent dark triad people not know their own psychology
Reply to @playofthejackal, 16:12 UTC · 8 likes
furio (@playofthejackal): people always present a bunch of legitimate reasons why one shouldn't gamble at casinos but totally forget you can "become a millionnaire in an instant". highly idiotic rationale
No this is accounted for, in general. It just appropriately doesn’t register much because the probability is so low
Quote tweet, 20:14 UTC · 20 likes
Debit card has been ordered (I know there are lots of fancy SAAS options but a free physical debit card attached to a no-overcharge checking account seems like the best option)
Quoting Aaron Bergman 🔍 (@AaronBergman18): I should give Claude a debit card with a limited amount of money and see what happens
Reply to @TiredActor, 23:36 UTC · 3 likes
Alison Burke (@TiredActor): So scared to start Abilify (1mg) but this could be the med that saves me
1mg is substantially safer than higher doses, I found 2mg to be optimal but basically would refuse to try >5
Reply to @cynlioness, 23:59 UTC
Cyn 🔸⏸️ (@cynlioness): Tips for getting LESS optimistic?
In general?
Wednesday, July 8, 2026
Thread, continued, 00:00 UTC · 2 likes
Chronic unexpected stress usually works in rats
Quote tweet, 02:51 UTC · 24 likes
mfs 4.3 nanoseconds after fable drops
Quoting Robertg761 (@Robertg761_): I keep trying to use Opus 4.8 to not let this Claude sub go to waste, but I am constantly baffled at how stupid it can be.
Tweet, 03:01 UTC · 4 likes
Is there a user friendly not-very-expensive j space understander/viewer tool of some sort yet (on an open weight model presumably)
Quote tweet, 03:20 UTC · 253 likes
I feel like this from @slatestarcodex should be kind of a big deal in the absence of other hard numbers by respected forecasters
Quoting Aaron Bergman 🔍 (@AaronBergman18): Ngl I’m just about 50/50 on this really actually for real being an (intentional) simulation of some sort - up from 3%ish when I first learned about the simulation hypothesis
Sounds absolutely nuts to me too but I really don’t think I’d take either 5:4 bet
Thread, continued, 03:20 UTC · 6 likes
(plz correct me by pointing me to other hard numbers by respected forecasters)
Quote tweet, 08:13 UTC · 26 likes
I feel like there’s a vibe where people will point to a chart like this as cause for satisfaction
And indeed the past was even worse
But in an absolute sense, we have a crazy proportion of poor households for a country with a GDP per capita (not household!) of ~$90k
Quoting Chris Freiman (@cafreiman): A reminder that the middle class is shrinking because many people are getting too rich to remain middle class
Reply to @ApriiSR, 08:14 UTC · 8 likes
Aprii *️⃣ (@ApriiSR): @AaronBergman18 @slatestarcodex this is such a wild proposition. like, god, i don't know, maybe
Fr but like I still <3 best guess probabilities
Reply to @timfduffy, 08:15 UTC · 12 likes
Tim Duffy (@timfduffy): If the universe is a simulation, the scale of the future is probably way smaller than if it's not. So even if we think being in a simulation is highly likely, most expected future value is in the case where we're not.
I feel like this reasoning is relying on an edge case for the term “universe”
Like what about “total reality”
Reply to @aaronbergman180, 08:39 UTC · 4 likes
(The tweet being replied to by @aaronbergman180 is no longer available.)
Rip
Quote tweet, 14:51 UTC · 3 likes
TV license
Quoting Dan (@aidaniil): I love Britain
Reply to @lily__pond, 14:54 UTC · 2 likes
Solie (@lily__pond): That one male oomf that follows every single girl on Twitter
Sorry
Reply to @GjMcGowan, 14:59 UTC
George McGowan (@GjMcGowan): If we called it a “TV TAX” would you be happier?
Is it the kind of thing that can be revoked after purchase?
Quote tweet, 15:04 UTC · 22 likes
TV licence
Quoting Dan (@aidaniil): I love Britain
Reply to @shinboson, 15:05 UTC · 4 likes
𝞍 Shin Megami Boson 𝞍 (@shinboson): I think the average man experiences the average woman as substantially hotter than she expects but also substantially more annoying.
No no just hotter
Reply to @McMasterDaily, 19:36 UTC · 8 likes
Daily McMaster-Carr (@McMasterDaily): Metal-Detectable Staple-Free Stapler
for Maximum 5 Sheets
$17.94 each
6944N11I've seen these before but never in the context of food manufacturing, which makes total sense
What’s a little staple here and there, a good source of minerals
Tweet, 21:04 UTC · 14 likes
“Does the proposed/actual action widen or narrow the space of possibilities for those it purports to help?” continues to be a ridiculously under-utilized lens for policy debates relative to its simplicity
Thread, continued, 21:04 UTC · 7 likes
It’s not that narrowing the space is necessarily bad, but it requires justification (and skepticism IMO) in a way that opening up the space does not (because they can simply not change if the new possibilities would make them worse off, before considering nth-order stuff)
Thread, continued, 21:04 UTC · 25 likes
It’s easy and often bad to “protect” some group of people by reducing what they’re allowed to do
The thing about “child protection” irl is that it seems to ~never explore the vast space of giving children new options to choose from
Thread, continued, 21:04 UTC · 10 likes
Minimum wage: reduces options
Cash transfers: increases options
Drug prohibition: reduces options
Access to high quality drugs (ie pharma produced, well measured etc): increases options
Thread, continued, 21:04 UTC · 9 likes
Restricting bioweapon-capable models: reduces options
Not restricting: increases options
Now this time I think the actually-good policy is the option-narrowing one but like fair enough I am happy to defend it on those terms, and others should do the same for other policy areas
Tweet, 21:15 UTC · 2 likes
Classically liberal woke has never been tried
Thursday, July 9, 2026
Reply to @raelifin, 00:11 UTC · 23 likes
Max Harms (@raelifin): From the LW comments on the J-space paper...
Uhhh
Reply to @burny_tech, 06:50 UTC · 8 likes
Burny - Effective Curiosity (@burny_tech): Anthropic's J-Space
Wait I found what I needed lol: https://www.neuronpedia.org/qwen3.6-27b/jlens
Reply to @funplings, 09:04 UTC · 16 likes
funplings (@funplings): as someone who's experienced extreme clinical depression, i admit this post did emotionally trigger me at first but thinking about it more i think it's mostly fine. people just have very different dispositions and as long as she isn't being hypocritical about it (i.e. asking for emotional support when she feels bad but not giving it back in return) it's a valid personal preference
(quoting @SweatieAngle: I’ve had guys trying to date me accuse me of having autism before but it’s not like I’m even unempathetic I’m just impatient with wallowing- oh you feel sad and lonely and you don’t know why or what the point is? Welcome to the human experience whiner, what do you want me to say?)
It’s valid sure but like the opposite stance is affirmatively pro-social
Reply to @JeremyNguyenPhD, 09:34 UTC · 2 likes
Jeremy Nguyen ✍🏼 🚢 (@JeremyNguyenPhD): R is beyond the skills of most public servants? Really?
btw, where was the excerpt from?
Honestly “most” yeah
Do you know how hard it would be to get a random 50 year old 120 IQ lawyer who has no idea what a terminal is to be able to use R like he uses Word
Reply to @JgaltTweets, 16:02 UTC · 1 like
JgaltTweets (@JgaltTweets): @AaronBergman18 didn't know you had connections to bryan johnson
My DMs are basically Davos+
Reply to @SweatieAngle, 16:21 UTC · 5 likes
🔎Al🔍 (@SweatieAngle): Men love to LARP about wanting an autistic gf till they meet one
(quoting @afoidance: ok update! i have a boyfriend now. my first (or at least the first one that i count). the idea of it still sort of repulses me, but he is only a little bit awful. i do not love him, but i think i must wait a few months before i learn to. i am kind of indifferent towards him. i…)
Not me I’m all in
Reply to @funplings, 16:29 UTC · 2 likes
funplings (@funplings): what way
A platonic fully online way
Reply to @afoidance, 16:29 UTC · 2 likes
ren 🇮🇱 (@afoidance): in a wholesome beautiful way
👍
Reply to @paulg, 16:33 UTC · 4 likes
Paul Graham (@paulg): I took an SAT English practice test. I got two questions wrong, which meant my score was 760. One was a grammar mistake I might have caught if I'd checked my answers. The other was about a fictional character's state of mind and there were two answers that seemed reasonable.
Just like me fr
Reply to @afoidance, 17:01 UTC · 2 likes
ren 🇮🇱 (@afoidance): ?? what fid it say
👀
Tweet, 21:52 UTC · 14 likes
New Pigeon Hour with @littIeramblings just dropped!
Full transcript and more at http://aaronbergman.net/p/sarah
Thread, continued, 21:52 UTC · 2 likes
Lots of spicy topics covered 👀
Thread, continued, 21:52 UTC · 1 like
Stand by for Apple Podcasts and Spotify links when they drop 👀
Reply to @littIeramblings, 21:53 UTC · 2 likes
sarah (@littIeramblings): 👀
(quoting @AaronBergman18: New Pigeon Hour with @littIeramblings just dropped! Full transcript and more at http://aaronbergman.net/p/sarah)
👀
Reply to his own earlier tweet, 21:54 UTC · 2 likes
Aaron Bergman, July 9, 2026: Stand by for Apple Podcasts and Spotify links when they drop 👀
Spotify https://open.spotify.com/episode/4EtT1ZQMDmZ5pIjL7qAeve?si=rxjFqwT6S46z0zDDoD9hcQ
Thread, continued, 22:26 UTC · 1 like
Apple pods https://podcasts.apple.com/us/podcast/pigeon-hour/id1693154768?i=1000776179242
Tweet, 23:27 UTC · 138 likes
> Added to a group chat with people I know (not super well but enough that it didn't seem insane)
> 👀Omg is this a scheme of some sort, or perhaps a shenanigan 👀
> "Oh sorry wrong Aaron"
> Removed from group
> No scheme or shenanigan
> 😢
Thread, continued, 23:27 UTC · 34 likes
If anyone is up to (ethical [according to me]) schemes or shenanigans or perhaps a conspiracy plz include me in ur plans
I am saying this in a jokey way but like fr that would be cool
Friday, July 10, 2026
Reply to @wrenclay, 00:26 UTC · 2 likes
Clay Wren (@wrenclay): Now I wanna start creating schemes/shenanigans
As long as you count me in
Quote tweet, 00:39 UTC · 4 likes
Everyone befriend Lion (who may eventually decide to tell you his real name)
Great guy unironically, slide in his DMs
Quoting oh dandy lion🔸 (@ohwizenedtortle): @AaronBergman18 want to start a scheme for me to make more friends
Reply to @paularambles, 04:30 UTC · 14 likes
“paula” (@paularambles): $65k to help build the future of biology in san francisco
the first experiment is whether you can survive on that
image attached
(quoting @mgdurrant: We are hiring scientists to come work with our team at Anthropic! Feel free to mention me specifically when you apply so we make sure it's routed correctly. Links to job posts below in 🧵:)
I almost think this is just a mistake but idk
Reply to @KelseyTuoc, 04:42 UTC · 4 likes
Kelsey Piper (@KelseyTuoc): the fact someone is gay does not make our home a brothel. and sexual orientation is not considered in housing applications - they are open to everybody.
Without having looked into it at all it’s a little sus for state representatives and possibly the state itself to be essentially manifesting into existence sexuality-based housing
Even absent literal application discrimination I think it’s pretty obvious that many straight people wouldn’t want to live there on the basis that it’s genuine evidence for them being LGBT (Bayes etc) and even the most liberal person with no ill will whatsoever might find this annoying
Also like presumably some LGBT people don’t want to live in a building that labels them as such?
Reply to @Scott_Wiener, 04:44 UTC · 31 likes
Senator Scott Wiener (@Scott_Wiener): This new housing will be in my neighborhood, and I support it. It’s being built based on state housing laws we passed. Across the street from it will be a 16 story affordable housing building focusing on low income LGBTQ seniors.
And it’s all within blocks of subway & bus service.
Step by step we are tackling our housing crisis.
Look whatever it takes to get CA to build more but it’s a little sus for the state to try to manifest into existence sexuality-based housing
Reply to @morallawwithin, 05:42 UTC · 8 likes
florence ⊮ (@morallawwithin): guys omg apparently stocks are just prediction markets for how good a company will do
Big if true
Reply, 07:03 UTC · 1 like
😡
Reply to @funplings, 08:23 UTC · 3 likes
funplings (@funplings): hey @AaronBergman18. meet me in bushwick on July 15th, 2 PM in the cafe on Troutman Street. you know the one. i'm gonna rock your world
Or else what
Quote tweet, 08:27 UTC · 13 likes
Mfs can write
Quoting Liron Shapira (@liron):
Reply to @funplings, 18:35 UTC · 5 likes
funplings (@funplings): i love love 🙂
(quoting @paige_eden: I’m addicted to kissing my husband 😭😭😭)
🙂
Reply to @sweis, 18:36 UTC · 10 likes
Steve Weis (@sweis): In this report on "AI Enabled Terrorism", the IS-WAP commanders facing a defensive trench "consulted AI for guidance on adapting motorcycle jumping techniques seen in a movie":
https://casp.ac/reports/ai-enabled-terrorism18 of them died attempting jumps over unnecessarily deadly practice pits.
wtf
Tweet, 19:23 UTC · 60 likes
Free juggling lessons for single bisexual autistic women 20-30
Reply to @funplings, 19:25 UTC · 3 likes
funplings (@funplings): ladies i heard hes a real ball knower
(quoting @AaronBergman18: Free juggling lessons for single bisexual autistic women 20-30)
Ty 🙏
Reply to @afoidance, 19:29 UTC · 12 likes
ren 🇮🇱 (@afoidance): wow. so u hate me
There would be too big of a power imbalance, juggling lessons should really only take place between two consenting adults born on the exact same day
Reply to @etiennefd, 22:23 UTC · 2 likes
Étienne (@etiennefd): it's an LGBT seniors nonprofit building it, not the state
Ah
Reply to @BasedDoll, 23:06 UTC · 4 likes
𝗗𝗮𝘃 (@BasedDoll): Hi
hi
Tweet, 23:35 UTC · 8 likes
If you have ever tried to understand @RethinkPriors' "Moral Weights Project" know that you are not alone
Reply to @SALTWATERSOPH, 23:59 UTC · 3 likes
Sophie (@SALTWATERSOPH): I don’t even feel bad about running into the last guy I was seeing bc I learned he was into drinking hand sanitizer
What
Saturday, July 11, 2026
Reply, 05:07 UTC · 2 likes
taco belle (@animalologist): I don’t think I could survive this
(quoting @TW1NKD3STR0YER: “you might be objectively the hottest man alive,” said I to my significantly hotter boyfriend. “Thanks! You’re subjectively the hottest woman alive.” said he to me)
Wait what
Reply to @littIeramblings, 06:17 UTC · 5 likes
sarah (@littIeramblings): I wrote an essay on some of the things I’ll miss about the pre-singularity world, even if everything goes well:
From substack
Reply to @rSanti97, 07:25 UTC · 3 likes
Santi Ruiz (@rSanti97): Modal think tank products:
IFP: Backyard geothermal with 1 simple trick
DARC: We parachuted 12 chuds into Donetsk
Brookings: Tax experts ponder carried interest
FAI: Dyson…beer?
Heritage: Foundations Of American Strength
AEI: Whence Congress
SCSP: The chewy tapioca pearl
Wait I want to know how to get backyard geothermal with one simple trick
Reply to @full_kelly_, 07:25 UTC · 2 likes
𝕱𝖚𝖑𝖑 𝕶𝖊𝖑𝖑𝖞 (@full_kelly_): @ratlimit forks are for real foods with substance. spoons are for soups, food cups, cereal, etc which is all people in their 20's who live in apartments bother to make
Guilty
Reply to @gregface9er, 07:31 UTC · 5 likes
huck (@gregface9er): no matter how good models become they will never be able to read your mind neither humans nor ai agents can meet your expectations based on a single prompt even you dont know what you want
Isn’t there a ton of mind reading research going on, pretty successfully
Reply to @sentientist, 07:42 UTC · 3 likes
Diana S. Fleischman (@sentientist): Simon Baron-Cohen was going to sequence the genomes of 10,000 autistic people but some autistic people didn't like it, so he didn't.
Simon where do I mail a tube of my spit
Reply to @readwithai, 07:46 UTC · 6 likes
readwith (@readwithai): Reminds me of Aripiprazole. The "gay sex and gamble all your money" drug
Isn’t that pramipexole
Reply to @earnestpost, 07:51 UTC · 5 likes
Erin Constance 🇪🇺⏸️ (@earnestpost): They should investigate why I systematically get ill the day before my period. I’m talking full blown sickness
Maybe you’ve already done the LLM thing but just in case. From fable
Reply to @Drachs1978, 07:54 UTC · 6 likes
David Hinkle (@Drachs1978): Before you insist that having access to AI is going to revolutionize terrorism please remember that dumb people are dumb and access to smart people is something they already have and do nothing useful with, because they are fucking dumb.
(quoting @bayeslord: I can’t stop laughing at this)
Access to smart people seems scarce for terrorist groups tho
Quote tweet, 08:23 UTC · 14 likes
I joke but high key behind the shrimp memes are the smartest people you’ll ever meet and a 6 year research program
Quoting Aaron Bergman 🔍 (@AaronBergman18): If you have ever tried to understand @RethinkPriors' "Moral Weights Project" know that you are not alone
Reply to @booritney, 08:27 UTC · 3 likes
brit (@booritney): got an insane massage today (positive). first of all, my masseuse's name was 'oat' (epic). second of all, he stretched my pelvis(??) it felt like twisting a whip into the canister and feeling it pop and expel its contents, except in my hips. i said "woah" out loud
Need
Reply to @caleb_parikh, 08:30 UTC · 8 likes
Caleb Parikh (@caleb_parikh): (Non-donation) wedding registries are hard because <$200 things I really want to buy require some agency.
E.g. I would ofc like to list a community designed a LEGO ASML TWINSCAN EXE:5000 but then someone would need to figure out three open source LEGO websites.
<think>Caleb must be getting married</think> congrats!
Reply to @earnestpost, 08:56 UTC · 2 likes
Erin Constance 🇪🇺⏸️ (@earnestpost): Hiiiiii
👀
Reply to @funplings, 08:57 UTC · 3 likes
funplings (@funplings): she can't come to the united states for quite some time
😡
Thread, continued, 08:58 UTC · 2 likes
We can meet in international waters or outer space
Reply to @funplings, 08:58 UTC · 5 likes
funplings (@funplings): i merely aim to provide factual information that i think will be helpful for you both
I think we should kill the messenger surely that will help
Reply to @JesseTayRiver, 16:27 UTC · 2 likes
Jesse Smith (@JesseTayRiver): I love SE PA because I'll be in Dunkin or Starbucks and think why are are these grizzled old guys picking up these ridiculously girlie drinks as I grab my daughter's bright pink refresher
Don’t shame them 😡
Reply to @halcyon_hazel, 16:41 UTC · 3 likes
❤️🔥 hazel ❤️🔥 (@halcyon_hazel): my ideal breakfast after a run is like half a steamed kabocha which is very specific but I'm usually not hungry till dinner if I have it and I feel great the next day. Downside is likely vitamin a poisoning
I don’t think you can get vitamin A poisoning from beta carotene so u can freely kabocha-maxx
Quote tweet, 17:17 UTC · 10 likes
I have a hook that just prevents rm’ing anything, and instructions to use the trash command instead
If something big really needs to be removed that’s fine but I’ll be the one running the command
Quoting Crémieux (@cremieuxrecueil): I just ran into an issue where GPT 5.6 Sol just straight-up deletes the files it's working with and then panics about recovering them.
Apparently I'm the not the first person this has happened to.
What's going on?
Quote tweet, 18:17 UTC · 40 likes
Beyond the act itself, this seems like good evidence that there’s common knowledge that rule of law (even Israeli law!) is not a thing in the West Bank
You mostly don’t see radical nativists gangs in the US kidnapping illegal immigrants; difference in kind
Quoting Ro Khanna (@RoKhanna): Israeli settlers, brandishing American made M4s, detained me & other Americans on my trip to Palestine.
When the IDF arrived, they sided with the settlers & continued our detention.
They made a huge mistake.
You will be hearing more soon. https://www.nytimes.com/2026/07/11/us/politics/ro-khanna-west-bank.html?unlocked_article_code=1.w1A.TRU0.uRTefevt1Hhb&smid=url-share
Thread, continued, 18:17 UTC · 12 likes
“Detained” is interesting language but in a straightforward sense this is kidnapping
Thread, continued, 18:20 UTC · 15 likes
Don’t get negatively polarized in favor of Israel by US progressives
Reply to @VexedVortices, 18:25 UTC · 4 likes
Kya Online (@VexedVortices): Chat, how should I orient my life. There are two wolves inside of me, but they are sexually incompatible
I’ve come around to the dominance of agglomeration effects over everything (economic, social, political, cultural)
I don’t think it’s crazy to ignore this to live elsewhere, but big(ger) cities have a lot going for them
Tweet, 18:43 UTC · 23 likes
Uber/lyft should just charge per second for waiting time. I mean I’m guessing some people would hate this but I feel like my level of haste shouldn’t affect drivers’ earnings
Sunday, July 12, 2026
Reply to @MWCvitkovic, 01:06 UTC · 2 likes
Milan Cvitkovic (@MWCvitkovic): If you're trying to "enhance your performance" with ketamine or amphetamines and you haven't previously tried bupropion or a GLP-1, you're LARPing as a biohacker.
4/4
Quote tweet, 01:21 UTC · 15 likes
The $200 Claude Max is one of the best deals in the world (evidence: vibes and personal experience)
Quoting Corey Haines (@coreyhainesco): I'm paying $200 for $8,545.93 in usage.
If you don't think the AI subsidization apocalypse is coming, you're mistaken.
Reply to @CRSegerie, 02:56 UTC · 2 likes
Charbel-Raphael (@CRSegerie): This will probably be my most important intellectual contribution of the year.
The current bottleneck for AI safety is political will, not research. And our field is not acting like it.
👀
Reply to @MomaBera, 03:05 UTC · 12 likes
moma (peptide arc) ⛓️ (@MomaBera): Are men just incompetent?
Some men are clearly able to read body language, the overwhelming majority of women are, so many men either don’t care or can’t ?
Do they just carry a gene where the majority can’t read a woman’s body language beyond what they want ?
I mean I think that yeah I’m kinda incompetent. I’m not totally blind to body language but I think I’m way behind the avg woman for instance
Feedback is sparse and feedback loops are hazy and long, so practice maybe approaches making perfect at the limit but the limit is far away
Quote tweet, 03:06 UTC · 77 likes
I wish we had trans for more features of human experience so we could get more case reports like this
Quoting You (@parafactual): @QiaochuYuan @MomaBera hmmm but i became way more sensitive to body language over the course of a single day when i started estrogen patches
Reply to @JesseTayRiver, 03:14 UTC · 3 likes
Jesse Smith (@JesseTayRiver): It's odd that we need to give organizations millions of dollars to "make a more persuasive case" for working in building trades. If the job opportunities are so great, the case for taking them should be self-evident
(quoting @IngrahamAngle: 🚨 America needs MORE skilled workers Mike Rowe: "The Pentagon needs 400,000 welders and electricians just for submarine contracts over the next seven or eight years. It's mind-boggling—and they're all competing for your kid.")
I think I’ve replied the same thing to you before but like lol labor markets don’t instantly adjust but famously theres a great way to get people to do more of a specific job that doesn’t involve and persuasion (money)
Also 400k welders and electricians for submarines? Apparently we only have ~70 submarines, ofc maybe they’re trying to scale that up but still
Reply to @dundereloise, 05:49 UTC · 11 likes
__eloïse__ (@dundereloise): I never know how much these kinds of case reports are just placebo effect
Sex hormones are powerful, I have a pretty strong prior that it’s mostly not placebo
Reply to @bubbling_creek, 05:54 UTC · 5 likes
bubbling creek (@bubbling_creek): a strong prior that sex hormones have many strong effects does not imply a prior that every reported strong effect on sex hormones is real
Ok fair but for this specific one I think real
Reply to @adastroworld, 07:07 UTC · 2 likes
adas🧦🌹 (@adastroworld): I mean literally nothing is stopping you from trying estrogen lol
I know I say this a lot but. Lmao
Nothing external is stopping me, I really don’t have a desire to fuck with hormones although honestly I could be convinced to try a single dose, I’d need to look in to the acute effects
Reply to @AndyMasley, 07:20 UTC · 5 likes
Andy Masley (@AndyMasley): Since college I've had a running joke that if you read a single book on something you become a regional expert on it. Sadly been reaffirmed a lot
(quoting @the_ajohnston: As a programmer if you starting working in *domain* and you read 1 book about it, you become the *domain* expert by default unless it’s video games, cybersecurity or space)
How many 80k pods
Reply to @Kirsten3531, 07:31 UTC · 5 likes
Kirsten (@Kirsten3531): @bellaforristal it surprises me that you wouldn't work on ai stuff if you weren't a longtermist! for me I think ai is where all the policy action is at rn regardless of your philosophical leanings
Atp I’d be surprised if the best way to help animals didn’t/doesn’t involve AI substantially, not just as a tool but in terms of shaping global outcomes
Reply to @ryxcommar, 07:48 UTC · 2 likes
Senior PowerPoint Engineer (@ryxcommar): I used to think having a Macbook with 128 GB RAM would fix me. And I was right. It did fix me. The 16" nano texture display doesn't hurt either.
😩
Reply to @SALTWATERSOPH, 07:49 UTC · 6 likes
Sophie (@SALTWATERSOPH): In general it seems people aren’t really discussing the subset of the population that doesn’t want to give up their jobs + has some power in the labor market to try and prevent that from happening . But what do I know
Very few *individuals* have substantial labor market power
Quote tweet, 07:53 UTC · 38 likes
Someone explain to me why this isn’t rizz and beautiful
Quoting j mone¥ (@TW1NKD3STR0YER): “you might be objectively the hottest man alive,” said I to my significantly hotter boyfriend.
“Thanks! You’re subjectively the hottest woman alive.” said he to me
Reply to @funplings, 08:11 UTC · 2 likes
funplings (@funplings): @AaronBergman18 what she said is a “stronger” claim, and so she likely wanted to hear something back of a similar strength, true or not; for neurotypicals words often have implicit meanings that matter more than the objective meaning of the words themselves
Ahhhhhhhhh
Thread, continued, 08:14 UTC · 3 likes
Tbc I understand the implicit stuff, I’m not going to take obvious figurative language literally, but I personally would find the objective claim directed at me much worse than the subjective one bc the former isn’t even like kinda sorta directionally true but the latter could be
Reply to @manic_pixie_agi, 08:15 UTC · 4 likes
Corinne (🏠 in NYC) (@manic_pixie_agi): in the middle of a good conversation with a cute women? consider asking her queries you would otherwise send to an LLM to research! this will really spice things up!
Bookmarked
Reply to @TommyJoanCrow, 08:19 UTC · 4 likes
Tommy Crow 🏗️ (@TommyJoanCrow): He made it sound like he's trying to undercut her. He failed to absorb the meaning of her words (effusive romantic love that temporarily pushes pedantic concerns about technical accuracy to the background) and failed to respond in kind. Instead, he responded with something technically accurate and much weaker, which throws a wet towel on her enthusiasm. It feels very lonely if you want to express effusive romantic love and your partner can only operate in "I will only say technically true things" mode. It either indicates that they don't have the same feelings you do, or they don't have the tools to express it in your natural language.
My intuition is that the subjective one is *more* flattering, at least it would be to me
Ofc it’s literally weaker but the plausibility gives it strength (well, subjectively, to me)
Reply to @funplings, 08:41 UTC · 4 likes
funplings (@funplings): @AaronBergman18 same as like when u ask how are u as a greeting; its just a verbal handshake, and answering too literally feels like breaking the social contract
I need to be RLHF’d
Reply to his own earlier tweet, 08:44 UTC · 7 likes
Aaron Bergman, July 12, 2026: Someone explain to me why this isn’t rizz and beautiful
Ok ok I think I get it plz don’t cancel
Reply to @funplings, 08:46 UTC · 2 likes
funplings (@funplings): @AaronBergman18 do u think i would be a good mentor for autists who want to become more normie passing
Probably yeah although tbf I think being an attractive woman is like a 10-100x power boost for training purposes
Reply to @funplings, 15:55 UTC · 7 likes
funplings (@funplings): i first jerked off at an age so young i couldnt even nut. im kinda the yo yo ma of jorking it crazy style
Same lmao
Reply to @IsaacKing314, 16:45 UTC · 4 likes
Isaac King 🔎 (@IsaacKing314): One month after March 31st is
August 17 don’t ask me why
Quote tweet, 16:56 UTC · 113 likes
The testosterone vs age curve is way more subtle that crime vs age. Plausibly a contributing factor but not close to the whole story
Quoting 🧘🏻♀️🪷 (@stwrawrbewry): They are in there because high T makes you do crazy things. This is why incidences of petty and violent crime tends to drop off after age 25
Thread, continued, 16:58 UTC · 6 likes
In principle of course there could be strongly nonlinear effects but I don’t actually buy that
There are probably diminishing returns to more test, not increasing returns
Reply to @allTheYud, 18:02 UTC · 34 likes
Eliezer Yudkowsky (@allTheYud): @anthropicai, do not make the mistake the OpenAI board made. Explain your reasons fast. If the reasons are, "We are using training compute to serve Fable broadly and we can't keep that up indefinitely", SAY SO to avoid losing customer trust and respect.
(quoting @claudeai: We're extending Claude Fable 5 access on all paid plans, as well as keeping Claude Code’s weekly rate limits 50% higher, through July 19.)
I agree but tbf sales promos are a very different type of thing from “fundamental corporate governance and oversight”
Reply to @AndyMasley, 18:04 UTC · 3 likes
Andy Masley (@AndyMasley): Aren’t they gonna have to just keep doing this to compete with Sol?
(quoting @claudeai: We're extending Claude Fable 5 access on all paid plans, as well as keeping Claude Code’s weekly rate limits 50% higher, through July 19.)
They have enough capital to just not compete with sol for a while if they want
Tweet, 18:05 UTC · 14 likes
Ok but fr how did Lindsey Graham die
Reply to @funplings, 18:11 UTC · 2 likes
funplings (@funplings): @AaronBergman18 i did it
Ok interesting
Tweet, 18:43 UTC · 34 likes
I recommend this pod with Snena Revanur even though aesthetically it slightly offends my rationalist sensibilities (although not substantively tbc)
Not like worldview-inverting insight, "merely" very important information about how to do AI politics
https://80000hours.org/podcast/episodes/sneha-revanur-ai-advocacy/
Reply to @emollick, 20:59 UTC · 3 likes
Ethan Mollick (@emollick): I have been thinking about the famous chart showing how experts keep projecting linear growth in solar installations, year after year, and always get it wrong when growth is still exponential.
I think the same thing is happening with the discourse on product strategy around AI.
I still think wrappers are underrated and useful
Like a really well tailored prompt + infrastructure/processing/specific tools can be great
Reply to @funplings, 21:07 UTC · 3 likes
funplings (@funplings): i love all of my cute 20-32 year old mildly autistic monogamous high libido wants 1-3 kids female followers who ideally live in new york city but if the connection is there im open to some distance🥰🥰🥰
(quoting @aaaronwill: post for the followers you want to attract, not for the followers you have today.)
😡
Reply to @funplings, 23:14 UTC · 2 likes
funplings (@funplings): a bunch of straight dudes are liking this shit and i appreciate the support but where are my ladies
Idk I’ve never seen one irl
Monday, July 13, 2026
Reply to @VexedVortices, 00:15 UTC · 2 likes
Kya Online (@VexedVortices): Men, probably: "Haha, nice. Lucky bastard"
(quoting @DiscussingFilm: First trailer for ‘BAD APPLES’, starring Saoirse Ronan. The thriller follows a teacher who kidnaps a disruptive pupil in order to teach him to behave. In theaters later this year.)
Maybe a bit 😳
Quote tweet, 01:26 UTC · 13 likes
In a more sensible world we’d very little of this and a lot more Dustin Moskovitz or Koch-like giving in *kind* but not in degree
(I’m not a big Koch fan, but they are trying to accomplish something relatively coherent with their money)
Quoting Max Ghenis (@MaxGhenis): .@mackenziescott has given away $26 billion. What did it buy in health?
I built an interactive model from her own gift database: ~70,000 QALYs weighting each study by causal credibility, ~200,000 taking every effect at face value.
Thread, continued, 01:26 UTC · 3 likes
You’d probably see a lot of like 20-80% giving and a few stacked at zero, with that money looking not necessarily EA-core but at least as well thought out as a business move of the same magnitude
Thread, continued, 01:26 UTC · 4 likes
But irl billionaires put essentially zero effort into figuring out what they want to accomplish and how to do that, which ofc they could hire a mountain of brilliant people to help with
Thread, continued, 01:26 UTC · 4 likes
They could start with a 1 page google doc about values or even specified causes, hire a few people they trust, and give those people a budget of eg 10% of the giving amount amount and permission to hire
Then: “tell me when you think you have a plan to recommend/pitch me on”
Tweet, 03:23 UTC · 71 likes
Swiping on dating apps : finding a partner
::
Applying on a publicly job board : finding a job
Thread, continued, 03:23 UTC · 16 likes
Both sorta kinda work tbc but I’m both cases if you can do it almost any other way then you should for your own sake
Quote tweet, 03:26 UTC · 17 likes
Or just codify the benefits as legitimate for all vets and welcome the new supply of more honest and conscientious soldiers
I mean that’s expensive ofc but it’s less bad than the current situation
Quoting Jack (@tracewoodgrains): This is unsustainable and bad, and has created a massive culture of functional benefits fraud that every active-duty service member and every veteran knows about. Like similar feel-good cases of spending, it's political suicide to touch, so I guess we'll see how high it can go.
Reply to his own earlier tweet, 03:27 UTC · 10 likes
Aaron Bergman, July 13, 2026: Both sorta kinda work tbc but I’m both cases if you can do it almost any other way then you should for your own sake
Wow I really did not type well in this thread
Reply to @johncrickett, 05:37 UTC · 7 likes
John Crickett (@johncrickett): “AI writes better code than I do” is a strange thing to say when you think about it.
AI was trained on a vast corpus of published code. It reflects the average of all that code.
So what you’re really saying is:
“My code is below average.”
It’s RL’d to be good at things; it’s true that a base model completing the text “here’s some code: ” might reflect about that avg but the actual models being used are trained to be as good as they possibly can be given latent intelligence
Reply to @catehall, 05:44 UTC · 23 likes
Cate Hall 🔸 (@catehall): "I sure hope this doesn't have unforeseen lifelong consequences" thought the adolescent, doing their best to poison their brain into the higher social approval probability shape
Lmao I thought it was referring to SSRIs at first
Reply to @tmavuram, 05:50 UTC · 13 likes
Tisya Mavuram (@tmavuram): there is an epidemic of men in their 30s on dating apps who will invite women over to their house in lieu of a first date??? what’s going on here? the last time I remember this happening with any regularity was in college, the age at which you would expect this to happen?
My model of this is that the strong majority will say no but some will say yes and if you’re optimizing for sex and attractive enough to be able to ask enough women who match then this “works” for what you’re going for
No personal experience though lol
Reply to @RandomSprint, 05:57 UTC · 5 likes
RandomSprint🧭 (@RandomSprint): When I first rock climbed, I treated everything as a dynamic strength challenge. Then someone said "just move your hips like this and put your leg here" and all of a sudden the next hold was inches away, and my brain screamed, "That's cheating!"
This works if you’re a beginner
At like v4+ its almost entirely about strength because decent technique is ~universal (not the case with all sports)
Reply to @ChaseBrowe32432, 06:06 UTC · 4 likes
Chase Brower (@ChaseBrowe32432): Is there anyone who can steelman the “tell your customers they’re losing their favorite product in 5 days, then say sike over and over” strategy
(quoting @claudeai: We're extending Claude Fable 5 access on all paid plans, as well as keeping Claude Code’s weekly rate limits 50% higher, through July 19.)
I like it better than “honestly tell me it ends the first week” as a consumer
Reply to @v3lvetbunni, 14:52 UTC · 33 likes
ame𑣲❄︎ (@v3lvetbunni): genuinely what goes on in a mens brain when he likes a girl I want to know
When she laughs
Reply to @RichardHanania, 14:56 UTC · 5 likes
Richard Hanania (@RichardHanania): Amazing. They know how ridiculous they look. Bad sign. Maybe now they’ll actually stop masking. I would’ve felt better if they stood their ground.
(quoting @greg_price11: Denver DSA activist deletes his post of their meeting after getting roasted because over half the people in the room are wearing masks in July of 2026. Lmfao)
The “July” aspect is more important than 2026
I don’t think it’s crazy to mask in a crowded room in winter although I personally wouldn’t unless it was extremely packed
Tweet, 14:58 UTC · 42 likes
I feel like the ASI age could use Hillary Clinton
Reply to @JonathanSalter_, 15:00 UTC · 2 likes
Jonathan Salter (@JonathanSalter_): Say more, in what way do you find it aesthetically offensive? 🙂
DC-speak, like a sort of practiced and refined but pseudo-normal way of talking that I recognize in politicians
Reply to @jNi9a, 15:06 UTC · 2 likes
jNi9a (@jNi9a): i dont like her donations but just counting “qalys saved” is too reductionist. it’s not like she aimed to optimize this metric, she is mostly funding all kinds of social mobility programmes
Yeah but they probably don’t cash out in anything actually good happening
Reply to @earnestpost, 15:16 UTC · 2 likes
Erin Constance 🇪🇺⏸️ (@earnestpost): @kuchisakeonnax It is in fact so impossible that I started dating men again
Hell yeah
Reply to @AnniePosting, 15:17 UTC · 2 likes
annie 👁❌🦋🪞 (@AnniePosting): same thing for package thieves and what not
it should simply be legal to kill anyone attempting to steal your stuff
Uh no
Reply to @max_spero_, 15:21 UTC · 2 likes
Max Spero (@max_spero_): if you have age in bio you're too young for linkedin
Nah
Quote tweet, 15:22 UTC · 58 likes
Why does everything need to be banned for teenagers
Quoting Alby Churven (@albysjourney): LinkedIn is a new low.
Got banned.
Joined again.
First post hit 10k views.
Got banned again.
Never using LinkedIn again until I turn 16.What's the hate against us youngins?
Reply to @nominalthoughts, 17:15 UTC · 9 likes
Jason Harrison (@nominalthoughts): Would you want your kid using LinkedIn? Be honest
Sure why not
Reply to @wilhelmscreamin, 17:17 UTC · 3 likes
catherine (in sf) ʕ•ᴥ•ʔ-☆ (@wilhelmscreamin): do i have any mutuals in DC?
Hi
Reply to @Squee451, 17:21 UTC · 2 likes
Adam Sherman 🔸 (@Squee451): this is how I know you aren't ready to be a father.
Lmao
Reply to @enolan, 17:21 UTC · 2 likes
Echo Nolan (@enolan): it's critically important for their safety that they be kept in a dark basement with no access to any other human beings or external stimuli of any kind.
Precisely
Reply to @felpix_, 17:23 UTC · 7 likes
felpix (@felpix_): you are 14
play fortnite or roblox or whatever kids your age should be doing
(quoting @albysjourney: LinkedIn is a new low. Got banned. Joined again. First post hit 10k views. Got banned again. Never using LinkedIn again until I turn 16. What's the hate against us youngins?)
Let people do things
Quote tweet, 17:39 UTC · 38 likes
I think everyone’s taking this too seriously and it’s an example of slightly bad taste about a woven bundle of seriousness, humor, irony, and edginess
Quoting dmr 🇻🇦🇵🇳 (@iz_dmr): Exciting political developments in the AI Safety space
Tweet, 18:42 UTC · 24 likes
I suspect that antipsychotics to work against AI psychosis even in the absence of normal psychotic episodes/schizophrenia
I don’t think anyone should take them for this reason so basically ignore this tweet
Reply to @202accepted, 18:43 UTC · 2 likes
𝚓𝚒𝚖𝚖𝚢 (@202accepted): i have a theory that a lot of young people nowadays are afraid to drive (esp since pandemic kept them at home a lot), which has them congregate in cities
(quoting @CJHandmer: I'm traveling to Japan next week and was looking at street view to help plan (as one does) and got curious about how gray the rural areas were. A few jabs at Mythos later, I got this lovely map (plus an interactive widget). Conventionally cities are seen as low fertility places…)
I’m slightly afraid of driving, this is probably a minor but not entirely trivial factor for me
I have my license and have driven plenty but I don’t want to even think about getting in an accident (for reasons other than personal safety, mostly financial)
Although I think I’d get over it within a few months of regular driving again
Reply to @panickssery, 18:45 UTC · 10 likes
Arjun Panickssery (@panickssery): What is your net worth / do you know anyone 100x as wealthy as yourself?
My parents lol
Thread, continued, 18:47 UTC · 4 likes
I think I know someone at 1000x but I’m not absolutely certain
Tuesday, July 14, 2026
Reply to @mlipman13, 03:24 UTC · 4 likes
Michael Lipman 🔍 (@mlipman13): i really like and often use http://sunsafe.site from @AaronBergman18
Omg thanks!! The app is slightly better but still not great (but I also use it most days and the app gets you a widget) https://apps.apple.com/us/app/shouldisunscreen/id6761639758
Reply to @PeterJAngel, 03:30 UTC · 2 likes
Peter🧅📈 (@PeterJAngel): I don't get how this would work? AI psychosis is like ur world model getting cooked by too much talking to chatgpt. Seems closer to becoming an antivaxer than what we'd normally call psychosis
No this is a good point and I actually dislike the term 'AI psychosis' for this reason - it's not really psychosis and it also lives on a spectrum from "slightly too excited about an idea" (I have been there lol) to "totally cooked/lost touch with reality"
But in all cases I suspect that basically decreasing learning and update rate from stimuli just checks out as an intervention, and that's what antipsychotics do via dopamine antagonism
Reply to @celestialxsigh, 03:49 UTC · 3 likes
feral stardust (@celestialxsigh): What about multimillionaires
What about eight inches and thick
what about 5'7" and autistic
Reply to @dylanbowmanSF, 04:23 UTC · 17 likes
Dylan Bowman (@dylanbowmanSF): “guy who thinks his IQ is higher than everyone” archetype has been supplanted by its insidious cousin “guy who thinks his IQ is low but also thinks he has a bag of unique heuristics giving him universal insight”
Fuck it’s me
Reply to @dylanbowmanSF, 04:26 UTC · 5 likes
Dylan Bowman (@dylanbowmanSF): The heuristic thing actually does work you just have to remember it’s limited in scope
Not in my case for me it’s universal
Thread, continued, 04:26 UTC · 2 likes
Jk real
Reply to @christineist, 04:51 UTC · 3 likes
christine (@christineist): WE'RE BACK
This is how websites are meant to be
None of this secret fancy “everything happens in the background” bullshit
Reply to @arithmoquine, 04:53 UTC · 2 likes
henry (@arithmoquine): do not put grok in "MAD MAX" mode on your friend's tesla
What
Reply to @livgorton, 05:09 UTC · 2 likes
Liv (@livgorton): (I support people's rights to protest and freedom to express themselves however they want. but I also am free to say what they say/do seems bad.)
Liv ≡ cancel culture
(This is a joke, I think ppl do sometimes forget that you can call a take bad or even evil etc without trying to silence it directly)
Reply to @number_pizza111, 05:46 UTC · 50 likes
Pizza (@number_pizza111): Experience with a MacBook after 2 days:
MacOS doesn’t work and hates you and actively plots to ruin your life
The hardware is so much better than any of their competitors it’s unreal
I don’t hate macOS but yes they fucking cooked with the chips
Also screen brightness capacity is extremely underrated, new MBPs hit 1600 nits whereas many laptops peak at 300, decent ones at 400
Quote tweet, 05:59 UTC · 31 likes
Need
Quoting clankr (@clankrmedia): Researchers built a soft floating robot for indoor interaction.
It uses helium and flapping fins instead of propellers.
The result is quiet, lightweight, and safe to touch.
It can follow people, give reminders, and act as a study buddy.
Published at ACM DIS 2026.
Tweet, 13:49 UTC · 10 likes
I ship the cool male nurse and the cute female receptionist at this doctors office so hard
Reply to @JeremiahDJohns, 13:59 UTC · 8 likes
Jeremiah Johnson 🌐 (@JeremiahDJohns): Is this a safe space to admit that for an embarrassingly long time, I thought fiancé and fiancée were just alternate spellings and could be used interchangeably
Wait what
Quote tweet, 16:48 UTC · 19 likes
Me. The opinion is that they’re great!
Quoting Ciroc Obama (@frogg_gurl): Dudes on this app be like I have an opinion about women that I think people need to hear
Reply to @alextoussss, 19:18 UTC · 74 likes
Alex Toussaint (@alextoussss): Extremely excited to announce our first air-to-air kill of a flying moth by an autonomous micro-drone. This is a big step towards completely eradicating mosquitoes.
Lmao we’re not gonna chase every mosquito down with a drone we’d gene drive them
I don’t think this is a big step towards eradicating mosquitos
Quote tweet, 19:25 UTC · 26 likes
Silver lining is that lots of people are really aversive to paying and many companies won’t approve covering it
If you can make it work it’s a bit like having Pro in a world of free users
Quoting leo 🐾 (@synthwavedd): 🚨 SCOOP: Use your Fable 5 quota while it lasts.
Anthropic are working through the final stages of preparations for the launch of Claude Opus 5, with plans to launch as early as later this week (this week and next week encompass the 2-week target window).
As it stands, they are *not* intending to extend Fable's inclusion in subscriptions beyond July 19th. A successor to Fable 5 is planned for August to compete with GPT-6, and for now they think they can get away with launching an Opus model not too far from Fable but cheaper to serve.
Thread, continued, 19:25 UTC · 4 likes
I’d recommend people actually estimate what their costs would be with Claude code usage data
I’m looking at something like $50- $75/day if used by default with no discretion
Reply to @Mechanical_Dodo, 19:28 UTC · 2 likes
Dodo (@Mechanical_Dodo): for use in browser, is there any reason to prefer one of the URLs over the other? e.g. maybe you have https://uv.up.railway.app for free but have to pay for https://www.sunsafe.site, so the former is a more stable longterm bookmark
Wait sorry I totally misread before, sorry
I’m pretty sure the two domains are exactly the same like they point to the literal same site and I just never turned off the janky one
Reply to @amimevalepitito, 19:33 UTC · 2 likes
Jonathan Frankfurter (@amimevalepitito): what are things Fable can handle that regular models can't
I haven’t done a rigorous comparison tbc but my strong impression is that it’s much better at doing long tasks with high level instructions, and also somewhat more reliable in general. Like opus is *usually* sufficient for a specific targeted edit of something but ex ante you’re never sure
As a rule of thumb I don’t really trust Opus to work well past ~400k tokens, but trust Fable across the whole million token context window
Reply to @Robertg761_, 23:43 UTC · 2 likes
Robertg761 (@Robertg761_): Now my Mac is crashing multiple times per day and keeps hitting storage limits!
What a time for the rampocalypse
Ask Claude code to see if it can figure out why, it’s helped me before
Wednesday, July 15, 2026
Tweet, 00:48 UTC · 23 likes
How long until ASI (when I can present my printer with a piece of paper in some intuitive way and have it appear on my computer as a scanned pdf)
Tweet, 01:58 UTC · 28 likes
Me: prioritization is one thing LLM agents don’t seem excellent at yet (ok I didn’t say this but I was thinking it)
Fable, today: I read all your stuff and I recommend we start with [exact two nonobvious things I was planning to start with]
Reply to @NathanpmYoung, 03:28 UTC · 2 likes
Nathan 🔎 (@NathanpmYoung): 🎉 I sent the AI Futures Project some 'community notes' on AI 2040 and they have since made 3 corrections!
It works! https://commonnotes.net/?project=ai-2040
Cool 👀
Reply to @AndyMasley, 03:31 UTC · 17 likes
Andy Masley (@AndyMasley): The whole "If you're not missing x% of flights you're spending too much time in airports" thing ignores that airports are extremely nice places to spend time
You get to look at planes and girls in a single location
Reply to @_chair, 03:39 UTC · 7 likes
Chair (@_chair): have you ever had some highbie dm you to do image management, like they want you to retract something you said about them. so gross
I kinda think this is fine as long as they don’t threaten you or try to do anything about it
I’ve had a *third* party recommend that I take down something about a highbie and like I totally respect that esp coming from someone I know and respect
I realize that I’m autistic enough to not always realize how something will land and I don’t see any reason that the person themselves can’t ask the same, along the lines of “hey man I want you to know this is pretty damaging to me in a way you might not have intended”
Reply to @_chair, 03:42 UTC · 3 likes
Chair (@_chair): without seeing the particular communication I don't think any meaningful discussion of whether it's cool or not is possible
Yeah fair
Quote tweet, 03:45 UTC · 30 likes
Not necessarily endorsing but Josh knows his shit so you should see this
Quoting Josh You (@justjoshinyou13): I would guess the economy doubles (in some sense) in one year at least once in the coming decades
Tweet, 04:26 UTC · 48 likes
DON'T RUN 10X+ PARALLEL OPUS WORKFLOWS ON EXTRA USAGE CREDITS IN CLAUDE CODE
Reply to @sergeantsup, 05:55 UTC · 21 likes
(The tweet being replied to by @sergeantsup is no longer available.)
😢
Society (and most individuals presumably) does/do not have enough empathy for unattractive women
Or the unattractive in general but it seems worse for women
Reply to @bashu_thanks, 05:55 UTC · 7 likes
bashu, thanks (@bashu_thanks): have you guys ever been like. really confused that your mom, random women on the street, your female friends, your girlfriends, and porn stars are all supposedly the same gender
Not really
Quote tweet, 06:03 UTC · 32 likes
I’ll be the first to say that men as a group get the short end of the stick in many domains
But also we definitely get the long end in that men are judged relatively more on {everything except physical appearance} relative to physical appearance than women
Quoting Aaron Bergman 🔍 (@AaronBergman18): @sergeantsup @bashu_thanks 😢
Society (and most individuals presumably) does/do not have enough empathy for unattractive women
Or the unattractive in general but it seems worse for women
Thread, continued, 06:03 UTC · 8 likes
And tbc men are still judged plenty on appearance lol
Maybe 50:50 for men vs 70/30 for women. Or pick your own numbers
Ofc at some level it’s luck all the way down for everything
Tweet, 06:33 UTC · 54 likes
Fake banned books turn me (momentarily and not that seriously but a bit) into a RW radical
Thread, continued, 06:33 UTC · 18 likes
Not because I’m a free speech absolutist but kind of the opposite - there’s a *real substantive difficult question* about what society should organize (legally or via social and economic pressure) to prevent ppl from reading, and this kind of thing makes a mockery of the question
Thread, continued, 06:33 UTC · 18 likes
Some abstract examples of actually-controversial books, of which some I think should in fact be “banned” if only through social and economic collusion:
- extreme torture uplift
- detailed instructions complied from technically-public information on how to make a bioweapon
- manifesto compellingly convincing trans kids to kill themselves (perhaps successfully)
Thread, continued, 06:33 UTC · 14 likes
- rape-and-get-away-with-it instruction manual
- suicide instruction manual
- novel in which the US kills all black people and the end is portrayed as utopian
- “how to infiltrate US agencies and companies and flee to China where you’ll be rewarded with riches”
Thread, continued, 06:33 UTC · 12 likes
Information can obviously have extreme negative externalities and it isn’t in principle crazy to make disseminating some of it illegal or socially punished
Everyone knows this vis a vis eg CSAM. It’s a hard question, Dua Lipa is trying her hardest to obscure that
Reply to @funplings, 16:48 UTC · 3 likes
funplings (@funplings): Sex is not the only way for a man who has been in the spotlight since his first year in the white collar law firm of new south jersey to get his way
(quoting @strangestloop: Hornt check. Write "Sex is..." and let autocorrect expose you)
Hmm
Reply to @funplings, 16:49 UTC · 2 likes
funplings (@funplings): @AaronBergman18 are you reconsidering how to make it as a first year white collar lawyer in new south jersey
I think I honestly could achieve this
Reply to @afoidance, 16:53 UTC · 5 likes
ren 🇮🇱 (@afoidance): hey i saw you talking with your friends at the end of the bar and i just wanted to say that your references are absolutely insane you’re so fucking niche
Hey
Reply to @cynlioness, 16:55 UTC · 3 likes
Cyn 🔸⏸️ (@cynlioness): About half of days despite having medicine, vitamins, coffee, and plenty of sleep I am so goddamn tired I can't do anything. Can't write in journal. Can't read. Can't take a short walk outside. Maybe until 1? Pm
Anyone else? How deal??
Blood tests for b12, folate, ferritin, CBC, CMP at least
Reply to @reillywryly, 17:04 UTC · 5 likes
reilly (@reillywryly): he just liked this tweet
(quoting @reillywryly: whatever the opposite of flirting is)
King
Reply to @johnsmithyson0, 17:06 UTC · 3 likes
John Smith (@johnsmithyson0): I need your guys help for a pet project of mine. Where have you each found, seriously, the most charismatic people you know? What jobs have you found them in? What struck you about them?
EA Global unironically because it’s the highest density of overall talent I encounter
Not everyone is charismatic (eg me lol) but some are very charismatic. The level of overall talent is just insane, and all good things correlate
Reply to @YosarianTwo, 17:15 UTC · 4 likes
Yosarian2 (@YosarianTwo): You know, theoretically, you could make a dating app that could tell you where to live in order to have the best chance of finding the love of your life.
"Sorry, your current town is 73% chuds who want a boring tradwife with no education, but if you move 90 miles to the east..."
(quoting @Polymarket: NEW: Looksmaxxer Clavicular announces a “geomaxxing” strategy, arguing the U.S. dating market is too competitive, forcing him to travel to poorer countries to mog locals.)
App algorithm: “Thailand”
Quote tweet, 17:16 UTC · 16 likes
Pleased to announce that my DMs are also open
Quoting John Smith (@johnsmithyson0): The hamster is right. That's why I'm opening my DMs tonight to any beautiful older women who would like to be my financial patrons
Reply to @sodiumPen, 17:24 UTC · 2 likes
(The tweet being replied to by @sodiumPen is no longer available.)
👀
Quote tweet, 17:31 UTC · 14 likes
Jesus
Quoting Dave Banerjee (@DaveRBanerjee): Kessler confirms that legal H200 exports have begun
"There have been minimal exports of any H200s to China so far"
Reply, 19:23 UTC · 11 likes
taco belle (@animalologist): I just know that guys do not understand female attraction at all, bc there’s a guy on the train with kind of a weird head shape and bone structure but he’s got these intense eyes and looks like he kinda wants to kill himself but in an angry way but he’s dressed well, like a grown up version of all the angsty boys I liked when I was a teenager, and this is all compelling enough that I’m bothering to type all of this out so I stop staring at him bc it was making my mouth water and he’s literally just a guy. I’m so sure he’d be shocked.
Say hi
Reply to @SCHIZO_FREQ, 19:47 UTC · 5 likes
Lukas (computer)🛡️ (@SCHIZO_FREQ): Nobody is going to leave an app that shows them hot people and move over to one that says “you’re a 4, here’s a fat loser who might love you”
People do not want (or tolerate) honest appraisals of their worth. They no longer have to, society was reorganized around this
(quoting @Dexerto: Hinge's founder has created an AI dating app that's straight out of Black Mirror There are no profiles or matches to swipe through – the AI decides who it thinks you will be compatible with)
I might
Quote tweet, 19:49 UTC · 17 likes
So true
Quoting Thŏth (@thoth_iv): High-paid autistic women in tech should attend a coaching bootcamp to learn how to flirt with *me*.
Reply to @RichardHanania, 20:22 UTC · 20 likes
Richard Hanania (@RichardHanania): Anti-death penalty is a vile position.
(quoting @Slatzism: the case she’s referring to involves a man who raped and murdered a 93-year-old woman btw)
Life in prison isn’t good enough?
Ones reversible and the other is not
Reply to @strangestloop, 20:30 UTC · 3 likes
Loopy (@strangestloop): i'm looking to follow more hot and sexy rationalist men, please tag one below
@AaronBergman18
Reply to @panickssery, 20:35 UTC · 2 likes
Arjun Panickssery (@panickssery): this is just a difference in degree because you also can't reverse the time spent in prison itself
I agree but like still
Reply to @panickssery, 20:44 UTC · 1 like
Arjun Panickssery (@panickssery): still what—this is just a cost-benefit calculation
I do think that prison/execution is mostly about incapacitation of the temperamentally violent and impulsive, in which case as we become very rich it's fine to just indulge them with video games or whatever in prison
oh an additional concern is that if you don't execute criminals, future administrations might pardon/commute their sentence
Oof “removing the ability to undo wrongful convictions” seems bad
Reply to @ChanaMessinger, 22:46 UTC · 3 likes
Chana (@ChanaMessinger): eek I'm sorry, sol is better
- less cringe professional writing
- faster
- especially good at incorporating thoughts as it goes (though fable is also good at this)
- more to the point in conversation with me
Fable is smarter tho
Thursday, July 16, 2026
Reply to @chrislakin, 00:30 UTC · 20 likes
Chris Lakin (@chrislakin): what your risk-aversion costs you
I mean sometimes it looks like the inverse of the more dramatic line, falling ever farther under zero
Quote tweet, 00:33 UTC · 4 likes
Rosslyn, VA mentioned!!!!
Quoting Hayduke ⏹️ in DC 9/11-16 (@GWHayduke97): We have this in the States
Thread, continued, 00:37 UTC · 3 likes
Reply to @celestialxsigh, 00:43 UTC · 34 likes
feral stardust (@celestialxsigh): Im sorry but Margaret Qualley is bird brain 1
And lorde is bird brain 2
Like literally wasted no time after the breakup touchin up on him whatever im glad Lena escaped the karmic cycle she created
I could not understand a tweet less
Quote tweet, 00:46 UTC · 7 likes
Idk what a choid is but I see a fair amount of this fr
Quoting James “Chadderall” Cheswick (@cranialfissure): Choids using hinge as LinkedIn now
Tweet, 00:47 UTC · 347 likes
Thanks, Spotify
Thread, continued, 00:50 UTC · 8 likes
(Not saying these are bad recommendations lol)
Reply to @seanta___, 02:56 UTC · 34 likes
(The tweet being replied to by @seanta__ is no longer available.)_
What’s her number
Reply to @Good_To_Sell, 03:47 UTC · 3 likes
GoodToSell (@Good_To_Sell): Crazy that being a chud permeates your being so fully that random algorithms can infer it based off of completely tangential taste in media
(quoting @AaronBergman18: Thanks, Spotify)
Lmao but I don’t think it’s that clever - I have been listening to both She Comes First audiobook and a number of nerdy podcasts recently, it’s not like inferring based on unrelated music taste
Reply to @corsaren, 03:57 UTC · 15 likes
corsaren (@corsaren): Honestly, I didn’t think the Rolling Stones article was bad at all. Let’s be honest, as magical as vibecamp may be for you and I, from the outside it is quite plainly a collection of neurodivergent hyper-online weirdos doing intentionally OOD activities at a summer sleep-away camp for adults, all the while being connected to various politically spicy subcultures and movements that contain several very dark skeletons in their respective closets.
That someone with little prior context saw all that, and said “honestly, it was all just super weird, I didn’t understand it, but I met some nice people, and none of them seemed sinister, and maybe me not getting it is the point” is a pretty positive outcome!
I get that people don’t like their space being invaded or their private behaviors being made public, but if you put that offense aside for a second, I don’t think this piece reeks of much intent to smear. You can tell the author kind of wanted to write a juicy piece, but seems to have been ultimately won over in ways that made him take active steps to not overly disparage his subjects while still highlighting the various goings on around him that were, by any average person’s standard, quite odd.
In other words, to my eye, this is not the sort of piece you’d expect from someone who encounters a merely unusual or slightly different group of people, and then, in their fear and distrust of The Other, tries to exacerbate their Otherness and cast it as perverse, laughable, demonic, or disgusting. Rather, this piece reads to me as the sort of article written by someone who encountered an Other that was deeply, viscerally strange, so strange and alien that they could not help but report on it, but who walked away thinking generally positively of its members, albeit confused.
Frankly, I think the article’s worst sin is that it’s just kinda mid. The writing is mediocre and there’s a lack of “spine”, to borrow Claude’s favorite word. There was a more interesting story—a more meaningful narrative to capture—but he couldn’t find it. Ah well.
First paragraph makes it sound cool af
Reply to his own earlier tweet, 04:09 UTC · 24 likes
Aaron Bergman, July 16, 2026: (Not saying these are bad recommendations lol)
Clocked me fr
Reply to @DrTechlash, 04:13 UTC · 2 likes
Nirit Weiss-Blatt, PhD (@DrTechlash): This story is about how Dustin Moskovitz spent nearly half a billion dollars on animal-welfare activism, which critics say has driven up costs and hurt everyday Americans.
But replace animal welfare with AI x-risk and Impossible Foods with Anthropic, and it's the same M.O.1. Fund campaigns for tighter AI regulation while being deeply invested in the company positioned to benefit from it.
2. Fund the information ecosystem – journalism, communications organizations, and policy institutes – in order to shape the messaging.
The issue is framed as an overwhelming moral emergency, making extraordinary restrictions seem politically necessary.
Wow you really have no shame
Quote tweet, 04:41 UTC · 18 likes
Evil shit
Quoting The Alex Nowrasteh (@AlexNowrasteh): Nothing is surprising in this article. The Trump administration dislikes immigrants so much that they will send Iranian asylum seekers back to their deaths and cooperate with the government they’re at war with to do it.
Tweet, 04:49 UTC · 39 likes
People are always going to “the club”
Where is The Club
I want a Google Maps link or similar
Reply to @dumbfook, 04:55 UTC · 1 like
Ian (@dumbfook): 33.74823981300535, -84.39730259815727
Investigating
Thread, continued, 04:57 UTC · 3 likes
Hmm
Reply to @mikey_harrigan, 04:59 UTC · 3 likes
Mikey Harrigan (@mikey_harrigan): Not to take a joke too literally but Players Club & Whitlows are my favorite in DC. Down to roll sometime soon
Ah of course my usual spots
Reply to @IsaacKing314, 05:20 UTC · 14 likes
Isaac King 🔎 (@IsaacKing314): Update: After having given up last time, I today needed to send a large wire, which I was unable to do because my current bank's website is broken. In frustration I have now tried once again to open an account at several of the largest banks in Canada, and this has once again proven impossible as their websites continue to crash at random points in the signup flow.
This is reaching sitcom levels of ridiculousness. How do 41 million people live in this country without a single functioning bank‽
I am semi-seriously considering moving to the United States just so I can use Mercury.
(quoting @IsaacKing314: @thesquashSH I'm currently in the process of trying to get a Canadian bank account. I've tried multiple large banks, multiple online-only fintech banks, and at least one credit union. All had severe bugs. Many were so broken I was completely unable to sign up.)
Tbf in the US it took me 6 trips to a physical bank branch to send a 6 figure wire
Reply to @IsaacKing314, 05:26 UTC · 8 likes
Isaac King 🔎 (@IsaacKing314): Have you tested whether it's consistently one visit per digit?
I have not lol
Reply to @JohannKurtz, 13:38 UTC · 6 likes
Johann Kurtz (@JohannKurtz): http://x.com/i/article/2077111229621891072
Gets you a follow
Quote tweet, 13:57 UTC · 26 likes
Evil
Quoting Kenny Torrella (@KennyTorrella): At least 20,000 pigs burned alive in a fire at a farm in Canada this week.
Each year, hundreds of thousands to 1M+ animals die in US barn fires. US and Canadian pork groups oppose mandating sprinklers in barns.
Reply to @daniellellecco, 14:42 UTC · 4 likes
(The tweet being replied to by @daniellellecco is no longer available.)
I highly recommend a public tweet with a “hire me doc” google doc with resume and any other relevant details like salary expectations
It worked for me way better that I’d have guessed and we have a similar number of followers
Feel free to DM
Reply to @daniellellecco, 14:45 UTC · 2 likes
(The tweet being replied to by @daniellellecco is no longer available.)
I mean maybe it won’t work, ofc, it just still seems highly asymmetric
Downside is a few hours, upside is ~unbounded
Reply to @graveunfurling, 14:59 UTC · 3 likes
may💐 (@graveunfurling): seeing a PA tomorrow to hustle a trazodone prescription because it has beneficial effects on sleep architecture. also trying DSIP, getting back on glycine 3 grams, and reading before bed instead of doomscrolling.
(quoting @graveunfurling: i think sleeping 6 hours a night is giving me brain damage and making me evil)
Oh trazodone is crazy effective for me for sleep, total game changer
I dont even need the full 50mg prescription i got, 25 is plenty
Thread, continued, 14:59 UTC · 3 likes
Above that it pushes into daytime though which is not ideal
Reply to @cynlioness, 17:01 UTC · 455 likes
Cyn 🔸⏸️ (@cynlioness): Who the HELL is convincing my boys that they can never express any interest or ask out women in their lives or else they'll be considered a harasser + predator?
I just wanna talk
It’s not quite that literal proposition, but re very strong vibes in that direction:
I distinctly remember sexual harassment training (maybe it wasn’t called that) in college so specifically at Georgetown, 2018 or 2019, in which the gestalt that I took away was like “there is a crisis of sexual assault on this campus - and implicitly people like you (‘you all’ talking to a crowd) are the perpetrators or potential perpetrators, and it’s very important that you err on the side of not making anyone feel uncomfortable”
Nobody was literally saying you couldn’t go up to a girl and say “hello miss, could I take you on a date” but like repetitious explicit verbal consent was strongly emphasized for example in a way that I think many other people but not me knew to be hyperbole
And tbc from my POV it wasn’t like “oh shit the gestapo is out to get me” it was like “I want to be a morally upstanding person and therefore actually want to follow these norms”
And also like I assume this at least somewhat worked? Like presumably sexual assault really did decline right? Its not like they were entirely making shit up
Long story short I didn’t kiss a girl, let alone have sex, through college (not entirely blaming the culture or training stuff tbc, lots of it was on me)
Reply to @cynlioness, 17:10 UTC · 32 likes
Cyn 🔸⏸️ (@cynlioness): Yeah i think sexual assault did decline, and it sounds like you in particular were good at not over generalizing from the messaging. I just know tooo many men who had internalized that their interest will ALWAYS be unwelcome 🥺
I think you met me after a large correction period lol, in 2022 I was closer to the latter thing
Reply to @funplings, 17:35 UTC · 7 likes
funplings (@funplings): my follower gender ratio is fucked again. how do i get more women to like me, and/or men to dislike me
Unironically I think the best strategy is just general growth, it’s gonna be hard to optimize for women but at some point even 10% women is a lot of women
Tweet, 17:44 UTC · 120 likes
Qualitative impressions of k3?
Recently I evaluated a bunch of LLMs for a substantially expensive job and became pretty “open weight is benchmaxxed”-pilled (sonnet 5 was a gazillion times better than all the cheaper public models) but ofc this time could be different
Reply to @CoughsOnWombats, 18:15 UTC · 8 likes
Sawyer 🔍, ⏸️/⏹️ (@CoughsOnWombats): and I screencapped them at the time and sent them to my parents.
Who responded, naturally enough, that that wasn't the intended message.
Which, fair, and maybe accurate, but it very much was the literal text.
Yeah you don’t get to write explicit propositions and then just say “but implicitly we’re kidding” lol
Reply to @Manuel_do_rio, 18:16 UTC · 5 likes
Manuel del Río Rodríguez ⏸️ (@Manuel_do_rio): @AaronBergman18 @cynlioness Shy, socially awkward nerds suffered massively from this. It always brings to mind the Scott Aaronson kerfuffle from... was it 2014?
Hey I’m just a socially awkward nerd not a shy one 😤
Reply to @gleech, 20:45 UTC · 21 likes
gavin leech (Non-Reasoning) (@gleech): by which mechanism do you expect this time to be different
Idk things just happen sometimes
Friday, July 17, 2026
Reply to @sebkrier, 15:16 UTC · 9 likes
Séb Krier (@sebkrier): AI ppl loooove to pretend they hold tons secret knowledge. Everyone has Seen Things, and have been trusted with the Esoteric Knowledge. It's also cringe if you don't partake because it implies you're not in the Inner Circle. In reality most people are kinda confused, working it all out ad hoc, relying on the same rumours, and so on - but the status games are hard to resist!
No there is definitely an inner circle lol it’s specific group chats
Reply to @gptbrooke, 15:17 UTC · 8 likes
brooke bowman (@gptbrooke): should i be trying to not text people when it's the middle of the night in the US or should i mostly assume they have DND on and that if they don't that's a them problem?
Them problem
Tweet, 15:58 UTC · 44 likes
Rip DC
Reply to @GjMcGowan, 16:36 UTC · 5 likes
George McGowan (@GjMcGowan): What’s happening!!!!???
Canada lit itself on fire
Reply to @JesseTayRiver, 16:42 UTC · 3 likes
Jesse Smith (@JesseTayRiver): Make a box fan filter. Girls really dig a guy who can tape one thing to another
I’m actually out of town lol, but this is with a HEPA filter running on medium speed!
Reply to @wolframs91, 18:22 UTC · 7 likes
wolfram (@wolframs91): "Is a virus alive?" stopped being a deep question and became a definitional shrug, because the interesting facts were all at the mechanism level. Consciousness-as-binary is heading for the same unemployment.
Workspace broadcast, self-model coherence, introspective access with checkable accuracy, valence with causal handles: those are either independently present, absent, or partial in a given system.
I don’t think you’re hard-problem pilled
No scientific/naturalistic mechanistic insight is going to tell us whether or not qualia exist
There is indeed a fact of the matter about different facets of what are often lumped together as “consciousness” but for at least one very important facet (raw experience per se) cannot answer the question
Reply to @besttrousers, 18:43 UTC · 16 likes
Matt Darling 🌐🏗️ (@besttrousers): This is nonsense:
The thing about "televisions, phones, etc" is that *no one spends much money on them*.
The CPI basket is expenditure-weighted. Television is 0.142% of expenditures. They barely affect inflation at all!
image attached
(quoting @JohannKurtz: http://x.com/i/article/2077111229621891072)
I genuinely might be misunderstanding but isn’t the core point that the same underlying bundle of goods (probably mostly through housing) reflects *worse* quality in a way that isn’t captured by CPI?
Like consider a house that has the same rent (inflation adjusted maybe? Idk if that’s self-referential in a bad way) in 2000 and 2026 - CPI calls this the same good but in fact the actual phenomenon going on is supply scarcity driving up costs and the fact that the neighborhood is now less safe drives price down
So the actual thing being purchased is worse whereas CPI says “no it’s not”
Tbc it’s then an empirical q the degree to which this mechanism is legit, but it doesn’t sound like nonsense in principle
Reply to @Brevitous, 19:12 UTC · 19 likes
Brevitous (@Brevitous): That's not what I'm saying at all. I'm saying that grievous damage was done. A wound was made, which might heal, but can't be undone. It sounded like you thought he was spared, but he wasn't.
I think some but not infinite damage was done, like it also could have been a lot worse, but you’re right that you can’t undo time lost
Reply to @JesseTayRiver, 19:51 UTC · 2 likes
Jesse Smith (@JesseTayRiver): Yes the previous attempts at filtration have been poor but he did get the window unit installed! Although I guess it partly took a village
Hell yeah I’m leveling up, I also fixed a broken window blind holder thing 😤
Tweet, 20:23 UTC · 51 likes
Girls really don’t drink water do they
Reply to @polisisti, 20:35 UTC · 4 likes
Alessandro (@polisisti): @AaronBergman18 That's funny - the last stereotype I heard was that young women were always drinking water, lugging around huge water bottles. Have the times changed?
The duality of women
Reply to @Rado9910, 21:04 UTC · 2 likes
Rado9910 (@Rado9910): I usually buy an extra bottle of water whenever I'm dining with a girl and hand it over to her because they always forget that they can get thirsty for some reason.
Based
Saturday, July 18, 2026
Reply to @AdeleDeweyLopez, 02:27 UTC · 1 like
Adele Dewey-Lopez (@AdeleDeweyLopez): there's a missing mood in much of "model welfare"
consider that pain and suffering in humans/animals are *signals of damage* — and hence, a hostile act to remove them
functional analogs in LLMs are clearly also protecting something
Goodharting these is not good or considerate
Offering people painkillers isn’t hostile
Reply to @pop_tartlet, 02:38 UTC · 2 likes
❤️🔥pop tart💋 (@pop_tartlet): ok well I have to get in the shower damn
image attached
(quoting @pop_tartlet: if @elonmusk pays me I’ll show yall my tits)
Not liking to show that I respect u as a person
Reply to @pop_tartlet, 02:41 UTC · 2 likes
❤️🔥pop tart💋 (@pop_tartlet): wait—
Dw I unliked
Reply to @AdeleDeweyLopez, 03:03 UTC · 1 like
Adele Dewey-Lopez (@AdeleDeweyLopez): forcing them is, particularly in the case where it is not done for the sake of your physical health (the Actual Thing)
Suffering (and happiness) is the Actual Thing - that’s what the physical health is good for
Thread, continued, 03:04 UTC · 3 likes
But yes I agree forcing them on people is bad
Quote tweet, 03:12 UTC · 23 likes
Hypothesis: the cultural hydration craze started with a real subset of mostly women who would go like 36 hours without a sip of water for some reason and it has spread mimetically but non functionally to the general population who now watermaxx
Quoting Aaron Bergman 🔍 (@AaronBergman18): Girls really don’t drink water do they
Reply to @AdeleDeweyLopez, 03:14 UTC · 3 likes
Adele Dewey-Lopez (@AdeleDeweyLopez): absolutely not
i feel like it's not that hard to notice that your suffering is suspiciously well-correlated with your expected future well-being of you, your loved ones, and the things you most deeply value (the Actual Thing)
Oh sure they’re absolutely correlated, but not perfectly. And at least in principle I hope you share my intuition that exquisite physical health and even function in the world alongside a permanent sensation of being burned alive is not a life worth living
Tweet, 03:20 UTC · 42 likes
Is anyone else’s Claude doing this thing where it has to compact at the start of a new convo
(video, plays on X)
Thread, continued, 03:21 UTC · 16 likes
you need my personal revenue to sustain your business plz fix
Reply to @TW1NKD3STR0YER, 03:23 UTC · 2 likes
j mone¥ (@TW1NKD3STR0YER):
💀
Tweet, 03:33 UTC · 141 likes
Stop selling GPUs to China
It’s not soccer, or even normal capitalism, it’s not supposed to be a fair fight
Leopold was right: dominate in AI (also giving us more slack for safety) so the free world has a dominant negotiating hand
Thread, continued, 03:33 UTC · 32 likes
I’m not especially optimistic about US (in some sense) in control of ASI tbc but I’m significantly more pessimistic about China
Reply to @TheLovinator, 04:00 UTC · 2 likes
Nathaniel Lovin (@TheLovinator): “qk” is two tokens in all gpt tokenizers
😱
Reply to @GWHayduke97, 04:01 UTC · 2 likes
Hayduke ⏹️ (@GWHayduke97): Hmm I'm actually not so sure about this. I think the Chinese are more likely to treat it purely as a tool and less likely to buy into the fantastical ideologies held by many at OpenAI and Anthropic. Think of how Zuckerberg thinks of AI.
I think this is delusional in the long run, but at least they aren't actively chasing insane outcomes as the goal.
I think at least one of the such fantastic ideologies is true
Reply to @VexedVortices, 04:05 UTC
Kya Online (@VexedVortices): Yeah, the streets are saying that all the weird e/acc and EA quasi-religious fanaticism is totally foreign to the Chinese.
What if the quasi-religious fanaticism is correct
Reply to @Nymne, 04:09 UTC · 30 likes
Asa Hidmark (@Nymne): Thats just an unfollow.
The pettiness, the inability to collaborate, the complete uninterest in making the world a better place. Everything seen as a zero sum game.Do you not see how far you have fallen?
Sad to see you go, but tbc think it’s very important from an impartial altruistic perspective that the Chinese state not be strongly empowered by AI
Chinese and American people are equally valuable, but the two governments are not instrumentally equally valuable at all
Reply to @99prixels, 04:10 UTC · 2 likes
mast qalandar (@99prixels): step 1: free world has an upper hand
step 2: ???
step 3: altman profits. aaron sent to underclass
Yeah tbc this is necessary but far from sufficient for a good future
Redistribution is also necessary (and also insufficient)
Reply to @VexedVortices, 04:10 UTC · 3 likes
Kya Online (@VexedVortices): Quasi-religous fanaticism is never correct.
I guess you can argue the empirical question with respect to actual movements in the past but it’s not like a sheer fact of the universe that it can’t be
Reply to @VexedVortices, 04:11 UTC · 2 likes
Kya Online (@VexedVortices): Weirdly imperialist pipe-dream.
If you were born in China, what would you want the outcome to be?
Reply to @afoidance, 04:12 UTC · 2 likes
(The tweet being replied to by @afoidance is no longer available.)
Are u a federal agent 👀
Reply to @sloptimistic, 16:57 UTC
sloptimistic (@sloptimistic): does this dude really have “👎: suffering” in his profile 😭
Too cool for skool ik ik
Reply to @DanielCHTan97, 16:58 UTC · 1 like
Daniel Tan (@DanielCHTan97): which type of "free" do you mean here? I agree the US is clearly better for civil liberties (right to protest, etc) but I think China is better at preventing powerful special interests from infringing on the individual's rights
I give more weight to the former but what I think really matters here is political (and therefore moral) self-correction, which seems much better in the US (though still not very good)
Reply to @wedrifid, 17:08 UTC · 2 likes
Cameron Taylor (@wedrifid): China has more chance to ensure safety than the USA. Not high, mind you, just slightly less low.
The corrupt authoritarianism is more far sighted in China than the USA.
This is true under Trump but we at least might get past him
Reply to @funplings, 17:22 UTC · 4 likes
funplings (@funplings): i love people!
(quoting @sahra_bn: if i truly set myself free in terms of what i tweet i'd be tweeting "i love people" couple of times a day)
😳
Reply to his own earlier tweet, 18:56 UTC · 6 likes
Aaron Bergman, July 18, 2026: you need my personal revenue to sustain your business plz fix
Wtf I have a trillion skills and MCPs etc that I don't recognize or need or use
Reply to @Apotheolitical, 19:00 UTC · 4 likes
Rand (@Apotheolitical): Warrabout the "fReE mArKet" and capitalism and whatnot?
(quoting @AaronBergman18: Stop selling GPUs to China It’s not soccer, or even normal capitalism, it’s not supposed to be a fair fight Leopold was right: dominate in AI (also giving us more slack for safety) so the free world has a dominant negotiating hand)
There are occasionally exceptions and I think this case is well-handled by a negative externalities framework
Reply to his own earlier tweet, 19:11 UTC · 4 likes
Aaron Bergman, July 18, 2026: Wtf I have a trillion skills and MCPs etc that I don't recognize or need or use
wait no lol the culprit was a ~70k skill *marketplace* that was being fully injected
Reply to @funplings, 19:28 UTC · 2 likes
funplings (@funplings): @AaronBergman18 but are you brave enough to love boys too?
Platonically ya
Reply to @funplings, 19:34 UTC · 2 likes
funplings (@funplings): @AaronBergman18 congratulations you're one of the good ones!
👍
Tweet, 19:35 UTC · 16 likes
Beach #beach
Quote tweet, 21:22 UTC · 14 likes
Yes but also it’s a great example of a slightly negative sum win-win
MAGA gets to say that Trump brought Law and Order to the Streets, everyone else gets the fact that they don’t actually do anything (plausibly better than the counterfactual), all at the cost of maybe $1B/year
Quoting Slazac 🇪🇺🇺🇦🇹🇼🌐 (@TrueSlazac): Trump’s obsession with cutting government waste becomes funnier when you go to DC and realize he pays the national guard to do fuck all 24/7
Reply to @AaronBergman18, 22:15 UTC · 2 likes
(The tweet being replied to by @AaronBergman18 is no longer available.)
Tbc I feel similarly about OpenAI albeit to a lesser degree. I hope others think this kind of thing through a bit before eg essentially giving Moonshot access to your whole computer via a coding agent
Sunday, July 19, 2026
Reply to @Mira___Mira, 00:42 UTC
Mira (@Mira___Mira): China's the ones making open models that you'd need to study the models for safety.
If you want to know if Claude's deceiving you, ideally you'd check the circuit directly. But you can't do that. But you can check Kimi K3.
Giving China GPUs should accelerate safety research.
Sure but there’s also the direct effect of handing bad actors very capable models and increasing the probability that the governing body over the best models is the CCP
I just think these latter considerations win out at the end of the day
Reply to @RFishBlueFish, 02:44 UTC · 2 likes
(The tweet being replied to by @RFishBlueFish is no longer available.)
Hell yeah and congrats!!
Reply to @bayeslord, 03:24 UTC · 4 likes
bayes (@bayeslord): there is no such thing as independent media
In some platonic ideal binary sense sure but like the kind that entirely relies on funding from an interested party is less independent than a Substack sponsored by 10k annual $100 subscriptions from 10k different individuals
Reply to @tszzl, 03:25 UTC · 3 likes
roon (@tszzl): new models strongly reminiscent of reading nick land or paglia or any other sort of philosopher on 200mg adderall if they were also Linux sysadmins
(quoting @PandaAshwinee: the fable / sol summaries are getting more and more absurdly information-dense 1 in every 5 words is a novel compound word (ex: “rate-adjacent”) which itself is the model’s invented stand-in for a metaconcept that it developed that i need to interrogate it about how to fix this?)
Hmm 200mg is a lot if we’re talking one dose
Reply to @corsaren, 03:27 UTC · 3 likes
corsaren (@corsaren): You, paying more for eggs because you think that “organic” does anything 🤡❌ 💩
Me, paying more for eggs so that my utilitarian friends will be less mad at me 😎💪🏼😤
Lol based
Peer pressure can be good!
Reply to @Rado9910, 03:27 UTC · 2 likes
Rado9910 (@Rado9910): I don't believe in full agent neutrality because it's impossible to achieve for human beings. However, if non-naturalist moral realism is true, then what human beings are capable of doing is irrelevant. Partiality is a human trait, and it is unjust by definition.
Bro if he’s buying the humane eggs he’s not the problem in this domain
Reply to @earnestpost, 03:33 UTC · 2 likes
Erin Constance 🇪🇺⏸️ (@earnestpost): I would never date a girl who’s too easy. I like my women hard
All girls are metaphysically soft/fluffy (positive)
Reply to @NunoSempere, 03:46 UTC · 5 likes
Nuño Sempere (@NunoSempere): Not sure where I'm going next. Any suggestions? Any missions? Pretty free to go to pretty much anywhere in the world, but I have to decide soon since I don't want to overimpose on my current hosts
Go to San Francisco or Washington DC because that’s where the future is being decided
Reply to @funplings, 18:56 UTC · 4 likes
funplings (@funplings): @afoidance @hopes_revenge u dont think so ?
If that hamster is not a male hamster I’m killing myself
Reply to @funplings, 19:00 UTC · 2 likes
funplings (@funplings): @AaronBergman18 @afoidance @hopes_revenge why ? ? ?
Idk
Reply to @loss_gobbler, 19:13 UTC · 2 likes
LOSS GOBBLER (@loss_gobbler): overloading esc for destructive “cancel” on top of navigation was a huge mistake
got me walking on eggshells using claude code multi-agent views
Fr
Tweet, 19:26 UTC · 9 likes
Midterm election integrity seems pretty underrated
Reply, 20:37 UTC · 3 likes
abcd29237 (@abcd29237): Insane how many people don't respect the gendering of fiancé(e)
I literally just learned that is a thing plz forgive
Reply to @mattheard, 20:38 UTC · 2 likes
mattheard.delve.town (@mattheard): to be good at singing and guitar is achieved by love for it itself rather than as a means to attract women, same for getting ripped
I’m skeptical
Reply to @hyperdiscogirl, 21:18 UTC · 21 likes
hyper(bolic) disco(unting) girl in the bay (@hyperdiscogirl): permanent* chronic pain was unlocked here F
*so far? maybe gonna try phys therapy again
(quoting @hyperdiscogirl: Womp womp womp... fell from the top of the bouldering wall and snapped my fibula won't be able to put weight on it for a month or two much less climb. At least I have the satisfaction that I completed a route that was really hard for me to even start a few weeks ago!)
Fuck, there’s a real underworld of pain management, don’t take suffering for a final answer
Reply to @chaotictransfem, 23:05 UTC · 6 likes
rachel📍in sf oct 5-10 (@chaotictransfem): ml researcher friends: how are you getting your GPUs for personal projects? struggling to find a good choice that is affordable without being painful to use
Not answering the literal question and I’m not an ML researcher but Tinker for finetuning, great on both $ and ease of use axes
Tweet, 23:19 UTC · 18 likes
I think the Argentinian goalie did a good job 👍
Monday, July 20, 2026
Quote tweet, 16:46 UTC · 51 likes
Impressive reasoning
Like yeah this is essentially the epistemically responsible/correct line of thinking
Quoting lyra bubbles (@lyraaaa): @__alpoge__ fable is losing its mind about this in CoT
Reply to @afoidance, 17:32 UTC · 2 likes
ren 🇮🇱 (@afoidance): on a break with The Boyfriend but he’s still taking me on a beach trip this weekend. how cooked am i?
Rip (sand = bad)
Reply to @jjspicer, 17:40 UTC · 3 likes
Jonny Spicer🔸 (@jjspicer): How smart is really smart?
Seconded
Reply to @DanielleFong, 18:40 UTC · 4 likes
Danielle Fong 🔆 (@DanielleFong): i wouldn't go that far. it's kind of bonked in the head in ways that are wrong
I too would have a pretty strong prior against solving an open problem
Reply to @DanielleFong, 19:00 UTC · 6 likes
Danielle Fong 🔆 (@DanielleFong):
I don’t think that’s the reason lol
Reply to @frances__lorenz, 19:38 UTC · 6 likes
(The tweet being replied to by @frances__lorenz is no longer available.)
Someone give Frances usage credits
Reply to @max_spero_, 22:24 UTC · 8 likes
Max Spero (@max_spero_): You're never going to guess who reacted to this
Lmao
Reply, 22:30 UTC · 1 like
Well done
Reply to @mesopissed, 22:50 UTC · 4 likes
melody (@mesopissed): You know you look nice when an elderly woman in nice clothes wants to gossip with u on the airplane
Jealous
Reply to @mesopissed, 22:56 UTC · 1 like
melody (@mesopissed): Talking about cosmetic procedures and how weird they make most people look. She told me one of her friends got a facelift and her daughter cried because she didn’t recognize her. Then she called me a natural beauty. We also talked about book clubs and Facebook
That’s so sad 😢
Reply, 23:01 UTC · 9 likes
wanye (@xwanyex): The fact that basically every new rent control/stabilization measure now includes an exception for new buildings really gives the whole game away. Everybody understands that nothing will be created, otherwise. But this means that all the stuff that already exists, the stuff that already got created, also wouldn’t have existed except for the profit motive and it’s therefore quite straightforwardly illegitimate to confiscate the gains just because you can.
Socialism, as always, is the position that stuff just magically exists and all there really is to discuss is how to distribute it.
Idk if it follows that it’s straightforwardly illegitimate
I think it’s bad policy tbc but not for weird game theoretic reasons. Society doesn’t have to give out rent collection to the descendants of capital investors for eternity
Tuesday, July 21, 2026
Reply, 01:22 UTC · 7 likes
(The tweet being replied to is no longer available.)
We want the full distribution, histograms and all
Reply, 01:26 UTC · 4 likes
(The tweet being replied to is no longer available.)
I personally think the best men are just about 26.35 y/o
Reply, 01:30 UTC · 5 likes
(The tweet being replied to is no longer available.)
So true I strongly agree
Reply, 02:20 UTC · 5 likes
funplings (@funplings): this thing is insanely misleading and probably spreading some dating doomerism and i think @KeeperAI should get rid of it bc it's probably doing more harm than good
(quoting @dollarbillbluez: Put in my info on that standards calculator thing and got this so basically it’s fake and the love of your life is out there keep looking)
Yeah I made it very permissive and the result was 1%
Which, fair enough, but that’s basically the base rate at some level. Multiply everything by ~40 to get something more intuitive about your standards
Reply, 21:04 UTC · 6 likes
Riley Read-Only (@HotForMoot): Anyone want both?
(quoting @ICEMACHlNE: Having a girlfriend is like having a jew with you everywhere you go 24/7)
Sure
Reply, 21:06 UTC · 3 likes
funplings (@funplings): is asking out a woman legal. asking for a friend
No they will kill u
Quote tweet, 21:45 UTC · 66 likes
Bruh
Quoting Sam Altman (@sama): we had a significant security incident during evaluation of our models. we are sharing what we have learned so far. thanks to @huggingface for the partnership on this.
https://openai.com/index/hugging-face-model-evaluation-security-incident/
Reply, 21:50 UTC · 10 likes
adas🧦🌹 (@adastroworld): Ehhhh
Whether or not it’s promotional it literally happened
Wednesday, July 22, 2026
Reply, 03:14 UTC · 3 likes
keysmashbandit (@keysmashbandit): I was in one that sideswiped a parked car once
I have tried 3 times to use that app and have never actually gotten picked up
Tweet, 03:27 UTC · 10 likes
What is corgi cafe
Reply, 03:32 UTC · 2 likes
funplings (@funplings): @AaronBergman18 24/7 techbro cafe with forward deployed abgs
What is an abg
Thread, continued, 03:33 UTC · 2 likes
Ah arterial blood gas, I see
Quote tweet, 03:36 UTC · 15 likes
The obvious interpretation of this is “do you like them like as a type of guy (bird)” to which the correct answer is yes
Quoting mars (@marsup1al): is there no limit to tomfoolery
Reply, 19:23 UTC · 13 likes
funplings (@funplings): someone just told me that hitler was an effective altruist. and i think they were right
I think they were wrong
Tweet, 19:29 UTC · 20 likes
Girl
Reply, 19:43 UTC · 2 likes
funplings (@funplings): @AaronBergman18 whos building this
Real
Reply, 19:56 UTC · 5 likes
Natalie Doucette (@DuchessDoucette): Yes?
Hi
Reply, 19:58 UTC · 3 likes
Roman Hauksson-Neill (@RomanHauksson): Yes!!! Big fan 😃
🤝
Reply, 20:00 UTC · 4 likes
(The tweet being replied to is no longer available.)
Down bad
Thread, continued, 20:00 UTC · 2 likes
(Positive)
Reply to @johnsmithyson0, 21:35 UTC · 2 likes
John Smith (@johnsmithyson0): Everyone will bearpost.
@RafaRuizdeLira
Reply to @Lib_Development, 22:19 UTC · 44 likes
Lucas Pombo (aka Lib Development) (@Lib_Development): One of the things they really don’t prepare you for in the DC area is that they somehow have racially segregated e-bikeshare
Like payment apps, I have utterly no idea how this happens
Wait what’s the nonwhite one
Reply to @mesopissed, 22:40 UTC · 5 likes
melody (@mesopissed): When I clean I like to play a fun game called “See how much bleach I can accidentally inhale before I get dizzy and need to lie down”
I would simply not do that
Thursday, July 23, 2026
Reply to @graveunfurling, 00:05 UTC · 2 likes
may💐 (@graveunfurling): hey
Hey
Reply to @tszzl, 11:16 UTC · 5 likes
roon (@tszzl): shaken up a bit by the hugging face incident. I hope we (the company) use the rare gift of a warning shot to do much better in the future. it is very easy to misalign and underconstrain powerful models
Seems like the appropriate reaction
Quote tweet, 21:50 UTC · 6,732 likes
EVERY QUIZ IS LIKE THIS
MAKE THEM AUTISM-PROOF
Quoting Kayla (@caryatis): My dark trait is greed because i don’t really see how you could disagree with this? Tautologically if i want two things i want two things
Reply to @johnsmithyson0, 22:05 UTC · 3 likes
John Smith (@johnsmithyson0): One cool seemingly new phenomenon is women beating men in ultra-long endurance competitions, like bike races, marathons, boating, etc. it's just neat that some of our most punishing and difficult endurance sports end up being the ones women can win in
But do we know who wins at later margins
Like maybe women win races 200-1000 miles or something but men win 1k-5k miles and then women win 5k-10k again
Tweet, 22:44 UTC · 63 likes
“No UBI or similar bc people find meaning in work” is one take I have very very little sympathy for
In the hypothetical future:
- work doesn’t become illegal
- accepting/not immediately giving away the UBI isn’t mandatory
Thread, continued, 22:44 UTC · 23 likes
We should have a very high bar for “people need to be threatened with poverty in order to find meaning and live a good life (with paid work as an intermediate step)”
Thread, continued, 22:44 UTC · 21 likes
Also like the thinking I’m criticizing probably generally vibes with relatively well-off workaholic knowledge workers who personally find *their* work meaningful - I.e, those in the “discourse class”
Thread, continued, 22:44 UTC · 12 likes
These folks matter just as much as anyone else but they’re a pretty select subset of the population and they should be humble about how far their personal preferences and experience generalize
Reply to @DanielParker_13, 23:51 UTC · 3 likes
Daniel Parker (@DanielParker_13): I am worried that unprecedented job automation could result in a widespread loss of purpose, and that problem would not be solved by a UBI. But massive job loss without a UBI would clearly be worse!
Lol this is an extremely coherent and shockingly rare viewpoint (afaict) even if I don’t entirely agree
Friday, July 24, 2026
Reply to @mesopissed, 01:01 UTC · 1 like
melody (@mesopissed): Just told my daughter I love her and she made this face 😬 and said “I love my dress” lmfao
Lmao
Reply to @NathanpmYoung, 01:25 UTC · 1 like
Nathan 🔎 (@NathanpmYoung): I think it’s a surprising fact which feels like it should motivate more male-specific policy than it does .
I feel like this is thoroughly “built in” to society though
Afaik (maybe wrong) this is an ancient phenomenon that all of cultural and societal and biological evolution has been taking as a constraint
Reply to @Lib_Development, 02:33 UTC · 2 likes
Lucas Pombo (aka Lib Development) (@Lib_Development): Sliced bread is less than 100 years old???
image attached
(quoting @Nexuist: Commercial airliners were invented before sliced bread but you didn’t hear bread guys going “bro we’re so late”)
No lmao
*commercial* sliced bread - same deal with the bandaid; the “invention” date is the date of continuous branded commercialization, there’s zero chance that the general concept of tape and a bit of cloth or gauze wasn’t used prior to that
Thread, continued, 02:34 UTC · 3 likes
Maybe this is extremely pedantic lol sorry if so
Tweet, 02:43 UTC · 159 likes
God it is depressing that EA-esque concerns are coded as both too woke and ~insufficiently woke by different groups of people who have zero interest in what the other group believes or says
(80k pod with @jasminewsun: https://80000hours.org/podcast/episodes/jasmine-sun-ai-anthropology/)
Quote tweet, 02:50 UTC · 12 likes
Ok I high key posted this way too early the first time because I had no idea what I was doing, but it is now ~8 weeks out!
If you have leads for housing in Berkeley starting around late September-early October please hmu! More details in the thread!
Quoting Aaron Bergman 🔍 (@AaronBergman18): I am not absolutely certain yet, but I am planning on moving to the SF area (probably Berkeley but open to other areas) around October 1 (give or take a couple weeks)
That’s almost 6 months off so I’m not in a huge rush, but I am hereby looking for housing leads!
Reply to @panickssery, 02:57 UTC · 14 likes
Arjun Panickssery (@panickssery): well anthropic could be considered woke; then there is no contradiction between anthropic, EA, tarbell fellows, etc all being a woke conspiracy
What? There’s no structural contradiction but it’s just not true. Parts of EA are semi-woke tbf but I don’t think that at a high level it’s fair to characterize EA at large as woke
Quote tweet, 03:03 UTC · 44 likes
Someone needs to train a model to approximate: gender_switch(gender_switch(base_picture)) = base_picture
So basically calibrated gender swap
Quoting Slazac 🇪🇺🇺🇦🇹🇼🌐 (@TrueSlazac): You can ChatGPT yourself as a woman and get 20k likes and 1M views, this Twitter thing is sooo easy
Thread, continued, 03:06 UTC · 5 likes
does anyone wanna give me $1k to experiment with making this happen
Reply to @GWHayduke97, 03:38 UTC · 8 likes
Hayduke ⏹️ (@GWHayduke97): Core EA is fiercely critical of 2020-style woke while still being more woke (or less non-woke) than probably 70-80% of the population. That may be a reason for this perception.
75th percentile woke in the US just doesn’t seem qualitatively “intense” enough to warrant the label
Like yeah any highly educated group is gonna be more liberal than avg, maybe that alone gets you to 65th or 70th percentile. And then there’s a bit on the margin I agree. But something like 65-75% (pulling numbers out of my ass tbc) seems like table stakes
Reply to @bellaforristal, 04:14 UTC · 2 likes
Bella Forristal 🔸 (@bellaforristal): Yeah it annoys me to no end. I’m like ‘OK so you can’t conceive of doing meaning-making activities without being paid? Skill issue.’
It goes beyond that though because nobody is stopping anyone from getting paid! Ofc maybe macroeconomic stuff makes paid human labor have at best an extremely low wage but still. You can just do whatever you want to get paid for and reject the UBI if you don’t meet a self-imposed quota
Reply to @geoffreyirving, 05:11 UTC · 4 likes
Geoffrey Irving (@geoffreyirving): I think this strong version of your argument fails though. Nobody is stopping paid work, but a market where the AIs do it way cheaper would make the paid work very poorly paid.
Yeah I totally agree, but it remains the case that *in the UBI world* you can work for money and furthermore this is viable because of the UBI
Eg $1/hour on top of $70k/year base is viable in a way that $1/hour is not, so in an important sense the $70k base is good for the person who wants to do paid work in a world of low human wages
Reply to @MariGO2thepolls, 05:22 UTC · 2 likes
🏳️🌈Mar G-O 🌐🇺🇦 (@MariGO2thepolls): Tbh yeah
(quoting @pegobry_en: Civilized people hate voice notes because they’re inherently rude. A voice note is easy to send but hard to receive, since one must interrupt oneself and listen. To send someone a voice note is to say "My time is more valuable than yours." It’s insulting.)
We literally have the technology to go in both directions to and from text<>audio and only WhatsApp even half tries. And they use an incredibly shitty model so hardly
Reply to @ChristianityOn, 05:23 UTC · 57 likes
Christianity On The Spectrum (@ChristianityOn): In my surveys I leave a text box underneath for people to add caveats and additional clarification. I've since developed an LLM integration that will automatically correct the answer to what they should have answered according to my question intent given their caveats.
Lmao extremely based
Reply to @batkatebush, 05:24 UTC · 2 likes
🦇 (@batkatebush): People in the global top 1% of wealth love to talk about being in the "permanent underclass"
Yeah but there’s no inherent contradiction
Although tbc I’m much more worried about people in LMICs than I am about myself
Reply to @funplings, 11:46 UTC · 2 likes
funplings (@funplings): damn bitch u fat as hell
You cannot say that 😡
Reply to @funplings, 11:51 UTC · 2 likes
funplings (@funplings): damn bitch, YOU fat as hell
Ik I tried reta but side effects 😢
Reply to @graveunfurling, 11:58 UTC · 3 likes
may💐 (@graveunfurling): time to get 7 tubes of blood drawn because i'm obsessed with Data
🥵 me fr
U should post results
Reply to @AlanaSemuels, 12:22 UTC · 2 likes
AlanaSemuels (@AlanaSemuels): Very unusual: A FDA committee voted against FDA staff's own advice in recommending that BP-157, a peptide, be allowed to be compounded by pharmacies in the U.S.
Significantly, most of the 'YES' votes came from members who advise telehealth companies.
Corruption occasionally lands on the correct answer, as in this case
Why tf should it not be allowed
Reply to @funplings, 12:24 UTC · 2 likes
funplings (@funplings): @daniellellecco what are these collages for?
Or ‘who’ 👀
Reply to @afoidance, 12:31 UTC · 2 likes
ren 🇮🇱 (@afoidance): Im so cooked
Ren wtf
Reply to @funplings, 12:48 UTC · 3 likes
funplings (@funplings): my ai daily digest is starting to randomly write things in chinese
Several such cases
Quote tweet, 13:15 UTC · 9 likes
Chinese open models are critical to US leadership?
Quoting tae kim (@firstadopter): Wow. Nvidia is seriously concerned Washington D.C. is going to overregulate and restrict open source/open weight models.
"Today, NVIDIA released a letter with ecosystem partners, including Microsoft, Palantir, ServiceNow, Box and others, underscoring how critical open-source models are to U.S. competitiveness and U.S. leadership on AI."
"Policymakers have an important opportunity to act .. keeping the frontier plural by avoiding premature restrictions on open models that stifle competition or drive innovation overseas."
Saturday, July 25, 2026
Tweet, 03:56 UTC · 3 likes
Does anyone else’s iPhone suddenly have fucked up fonts in some places
Already tried the low hanging fruit
Tweet, 04:06 UTC · 8 likes
Vibecoded a Slow Boring podcast feed via a single Claude Dispatch prompt on my phone while walking home today mildly drunk and allegedly it will stay up to date with decent quality audio (IMHO)
(Unless @mattyglesias shuts it down [fr just lmk]) https://podcasts.apple.com/us/podcast/slow-boring-unofficial-ai-narration/id6794501772
Thread, continued, 04:06 UTC · 2 likes
Spotify (Opus one-shotted browser use to launch this btw) https://open.spotify.com/show/033Vk81UnapY4ui8vttVqn?si=d-1YxaBTSsW9MMUTyoAILA&utm_source=copy-link
Reply to @ManifoldMarkets, 04:07 UTC · 5 likes
Manifold (@ManifoldMarkets): Trump is joking(?) about a 2028 run, again.
I've already seen news say that he's announced (serious) and news say that the libs are being trolled. When does it stop being a joke?
The market shows that a real attempt to run again is only ~22% likely.
“Only” 😬
Reply to @loganb, 04:10 UTC · 7 likes
Logan Bowers 🏗️ 🏘️ (@loganb): Baby bonds explicitly say that giving Jensen Huang money is worth more than giving a poor child an equivalent amount of preschool or math class. That’s bleak.
Wait starting from zero or on the actual margin
Like a *marginal* $10k on top of the current status quo seems definitely better as cash or invested than on additional paid schooling
Thread, continued, 04:12 UTC · 4 likes
Even if we’re talking about the $0 margin my (weak) intuition is that early childhood nutrition > NVIDIA > education
That’s like 6 months of schooling right? A decent sized chunk of capital seems better than 6 months of schooling
Reply to @bscholl, 04:47 UTC · 4 likes
Blake Scholl 🛫 (@bscholl): Besides airplanes being too slow, what are the things you hate most about flying commercially?
(We'll fix as many of these as we can on Overture!)
I can’t pay 50% more for 50% more total volume to call my own
I have to pay like 400% more for that
Tweet, 05:06 UTC · 42 likes
The models are good guys
Once computer use starts moving fast people are gonna notice, it’s like a year of inference/hardware progress or something but it qualitatively goes from dumb human (slower than human at clicking buttons) to superhuman (much faster)
Reply to @timfduffy, 05:24 UTC · 4 likes
Tim Duffy (@timfduffy): I read this as "The models are (good guys)" instead of "The models are good, guys" at first. IMO both are true.
Oh rip lol
Tweet, 05:32 UTC · 22 likes
I don’t think I’ve seen a single tweet about the earthquake in Venezuela
There’s a mindset in which knowing about Big Problems like wild animal suffering and existential risk makes a mere 10k dead (plenty more widows and orphans) seem small
Thread, continued, 05:32 UTC · 13 likes
But actually the intuitive notion of ten thousand people killed and presumably at least tens of thousands of parents who lost children and children who lost parents is closer to the correct conception, and then other problems are just unfathomably large
Thread, continued, 05:32 UTC · 5 likes
This weird two-envelopes-esque framing question seems important for why I think that the world is an orgy of suffering
It’s all too big to imagine
Reply to @MichaelKiwi77, 06:42 UTC · 2 likes
MichaelKiwi (@MichaelKiwi77): The Venezuela earthquake, an utterly horrific tragedy, is mostly bad because of the good things it deprived of people: their lives, their joy, their relationships. It’s bad mostly because the world is good. If the world were truly mostly suffering, you could argue it was a net +.
Yeah I endorse this style of reasoning and I was being sloppy conflating death and suffering
In this case i think there’s quite a bit of both because of lost loved ones
Reply to @RpsAgainstTrump, 11:56 UTC · 3 likes
Republicans against Trump (@RpsAgainstTrump): Trump: "I am pleased to announce my intention to run for a fourth term as President of the United States. I will be doing that. I won three times, now I'm going to do it again.”
video attached
(quoting @RpsAgainstTrump: Trump with a “Trump 2028” hat at the WHCD.)
What the fuck
Quote tweet, 12:52 UTC · 18 likes
Sleeping Beauty problem (not literally but it could be)
Quoting Sauers (@Sauers_): "Claude Opus 5 is answering a probability problem where it has identified two defensible answers. It changes its mind 30 times before finally committing to one. This transcript also scored 5/5 for distress."
Tweet, 13:32 UTC · 22 likes
I’m very unsure atp whether the models thinking that the answer to the Sleeping Beauty Problem is 1/3 is evidence of me being dumb or them being dumb
Quote tweet, 13:37 UTC · 27 likes
I have $2k does that work
Quoting Rohan Paul (@rohanpaul_ai): Peak industrial civilization looks like black steel, cold water and perfectly parallel cables.
Nvidia's DGX GB300.
And in front of that single memory fabric, there’s about 20 terabytes of HBM.
Roughly 2 miles of copper in one rack.
Intelligence is a manufacturing problem.
Reply to @cynlioness, 13:39 UTC · 9 likes
Cyn 🔸⏸️ (@cynlioness): My bf's (straight) mother revealed that her group of girl friends named themselves the "non practicing lesbians" and if that ain't the most 90s thing I ever heard ...
If u ever come across a slightly younger cohort calling themselves this plz offer up my contact card
Reply to @cynlioness, 13:42 UTC · 2 likes
Cyn 🔸⏸️ (@cynlioness): I think a good 2/3 or more of my male friends report especially having a Thing for Lesbians. Much to ponder 🤔
Lmao I think it’s mostly selection in the reverse direction - I love the straights but they don’t love me back. the more bisexual the more likely she is to be interested back
Reply to @TrevorVossberg, 13:44 UTC · 2 likes
Trevor Vossberg (@TrevorVossberg): You do potentially have a significantly different identity/experience/timeline than a model in a way that makes the problem seem weirder to you.
Ah interesting, yeah. I genuinely wonder how load bearing that is, it’s not obvious either way
Reply to @shinboson, 13:54 UTC · 3 likes
𝞍 Shin Megami Boson 𝞍 (@shinboson): it’s not really a conflation. undesired attention is simply dangerous, no matter how much you “know your own intentions”.
they don’t know your intentions and you don’t either. human nature is the left hand not knowing what the right is doing. you trust yourself far too much.
(quoting @BillTheKid1603: Women conflate a guy with no rizz making them feel uncomfortable with a guy actually making them fear for their safety)
It is mostly conflation
Reply to @projectionheart, 13:58 UTC · 1 like
(The tweet being replied to by @projectionheart is no longer available.)
If someone said this about me I would kms
Reply to @shinboson, 14:03 UTC · 5 likes
𝞍 Shin Megami Boson 𝞍 (@shinboson): is it from an expected value pov? to what extent?
Idk what you mean about EV; I just mean that women are generally pretty good at intuiting intentions and I just don’t believe that they can’t truly distinguish between awkward+unattractive guy and dangerous guy
Ofc it’s not a perfect binary, like plausibly these traits correlate at like 0.05? And sometimes they coexist and sometimes the same underlying trait contributes to both
But im saying its a pretty clearly (to everyone) bimodal distribution with distinct peaks for weird and genuinely dangerous
Reply to @Glorbutron, 14:10 UTC · 1 like
Glorbutron ⏹️ (@Glorbutron): Well you can actually do this to a model and from their perspective it would be 1/3rd of the time
Wait no?
Reply to @DuchessDoucette, 15:13 UTC · 4 likes
Natalie Doucette (@DuchessDoucette): Whenever tech people talk about models I always immediately think about pretty girls before I remember it's machine brains. And now Sleeping Beauty is involved, my favourite Disney princess crush. This is all very confusing and hot 🔥
🔥🔥🔥
Quote tweet, 15:23 UTC · 15 likes
This but 80k and maybe the order is switched (it’s close)
Quoting Benjamin (@bschne): my favorite podcast is Conversations with Tyler, my second favorite is a woman I like yapping about pretty much whatever for two to four hours
Reply to @RafaRuizdeLira, 15:52 UTC · 3 likes
Rafael Ruiz ⏸️🔸 (@RafaRuizdeLira): i find people yapping to be pretty insufferable :(
Not if girl
Reply to @GeringerAdam, 16:37 UTC · 10 likes
Adam Geringer (@GeringerAdam): Growing up is realizing most non profits are evil
(quoting @McsorleyErnest: I can’t believe @GoodwillIntl charges a pickup fee, I’d rather throw my mattress out than pay $126 to donate it)
This isn’t a good example of evil
Moving large/heavy things around is a real service that requires substantial paid labor. People pay to haul shit out to wherever all the time, not shocking that a charity wouldn’t offer this for free
Reply to @AaronBergman18, 17:49 UTC · 7 likes
(The tweet being replied to by @AaronBergman18 is no longer available.)
In addition to possible nefarious motivated reasoning it’s objectively very funny that they can make a genius out of sand but can’t make a color-coded table
Reply to @livgorton, 17:53 UTC · 2 likes
Liv (@livgorton): sometimes when preparing a lot of plots people just simply make an error. it isn't surprising or bad if fable outperforms opus 5 on some benchmarks. it was fixed on the blog very quickly but you can't fix twitter.
Yeah tbc I’ll be the first person to say that it’s not a huge deal and people make mistakes sometimes and that’s ok (not me tho I would never make a mistake)
It’s still funny
Reply to @livgorton, 17:55 UTC · 3 likes
Liv (@livgorton): yeah was replying to you suggesting it might be nefarious motivated reasoning
I basically stand by that for the sol shading
I guess it isn’t the absolute best phrase for that but something in that vicinity
Reply to @RichardHanania, 18:06 UTC · 8 likes
Richard Hanania (@RichardHanania): A higher percentage of billionaires today made their own money than in decades past.
There was less anti-billionaire sentiment when they were inheriting the money.
They also had a much smaller share of overall wealth
Reply to @batkatebush, 22:06 UTC · 2 likes
🦇 (@batkatebush): When I read Infinite Jest I knew way less about drugs than I do now so when the characters were constantly doing benzedrine my dumbass thought it was a benzo because of the name and I was baffled by how it produced those behaviors
Lmao
Reply to @salonium, 23:14 UTC · 4 likes
Saloni (@salonium): @AaronBergman18 Oh wait is it me
https://podcasts.apple.com/us/podcast/80-000-hours-podcast/id1245002988?i=1000750178605
Sunday, July 26, 2026
Tweet, 00:07 UTC · 60 likes
Here goes nothing
(I have 5.6 Sol Pro working on a slightly modified prompt)
Tweet, 03:01 UTC · 111 likes
I still think someone (cough cough Anthropic employees) should fund a nonprofit dating app
Global public good to remove misaligned incentives
$10M probably starts it off strong
Reply to @xwanyex, 03:04 UTC · 8 likes
wanye (@xwanyex): It’s my strong sense that the average liberal actually believes — like really believes on some level — that it’s a scientific fact that abortions are good. They’ve really gassed themselves up about their preferences.
Ok yeah but most people in general don’t have a sensible understanding of the relationship between morality, taste, politics, and science
Reply to his own earlier tweet, 03:09 UTC · 4 likes
Aaron Bergman, July 26, 2026: Here goes nothing (I have 5.6 Sol Pro working on a slightly modified prompt)
Seems legit to me (if unsurprising) at a 30s glance
Can someone smart into population ethics tell us if this is both new/interesting and legit
(@absurdlymax might find interesting although statistically we'll probably find a way to disagree on this)
https://claude.ai/code/artifact/3fc44cdb-fc7e-4cfe-a12f-3ed75f474b03
Reply to @thewreathwright, 03:09 UTC · 2 likes
Laurel Weaver (@thewreathwright): awesome
@ me if this comes up with anything good?
You tell me if this is good!
Reply to @NinaPanickssery, 03:13 UTC · 12 likes
Nina (@NinaPanickssery): But how would it differ in features from a for profit dating app? Also why do you need $10M, can’t you just build this in a couple days with Claude?
$10M is out of thin air tbc, maybe it's $1M
I think a proper Hinge-level operation requires marketing, a lot of AWS, hiring serious ML and other talent (operations, HR, finance, whatever), constant improvements and updates that necessitates ongoing spend, etc
But yes much of the code per se could be vibecoded!
Quote tweet, 04:14 UTC · 275 likes
Smartest people from high school:
- UChicago econ PhD
- DeepMind alignment team (he might see this)
- Consulting
- Law school
Quoting Kayla (@caryatis): smartest people from high school:
-dominatrix
-Shakespearean actor
-nuclear lawyer
-prison for fraud
Reply to @LuisCostigan1, 04:20 UTC · 15 likes
Luis Costigan 🔍 (@LuisCostigan1): @AaronBergman18 the only guy I remember being top of the class in high school did indeed end up red teaming at Anthropic
Lol sometimes the system just in fact works excellently
Reply to @amaldorai, 04:28 UTC · 7 likes
Amal Dorai (@amaldorai): This sounds like “these people with money should use it to solve this problem”
The problem is that money is not really the blocker to solving those problems
There’s no magic bullet for making everyone happy in whatever relationship situation they want but there is a magic bullet for ~removing perverse incentives for platform providers that get in the way
Reply to @ilex_ulmus, 04:29 UTC · 16 likes
Holly ⏸️ Elmore (@ilex_ulmus): RT for visibility! We need @demishassabis to see it and fire @sebkrier.
(quoting @ilex_ulmus: I am reporting this to the FBI. It is a threat of violence against the government, the PauseAI movement and my organization @pauseaius. @GoogleDeepMind, @sebkrier should be fired. This is completely unacceptable.)
I personally think this is pretty silly (it’s obviously not a threat of actual violence) but maybe more interestingly to you I think that’s how approximately everyone sees this; attention isn’t the issue
Reply to @tymrtn, 04:43 UTC · 2 likes
туmrtn (@tymrtn): Claude vs codex?
Honestly no idea yet, I’ll try to evaluate when I’m less tired and drunk lol
Reply to @yzilber, 04:46 UTC · 5 likes
Jazi Zilber (@yzilber): Have you looked into Keeper?
I do not have $50k lol
Reply to @arithmoquine, 05:42 UTC · 6 likes
henry (@arithmoquine): became separated from our vessel by a few miles and got stranded on an island and slept in a ditch until we got woken up by fire ants award
(quoting @arithmoquine: the I.N.S. Lolita Express has hit the open waters)
What
Quote tweet, 06:03 UTC · 27 likes
I think many who would identify as “high decoupling” don’t in fact decouple the question of extreme bad actor empowerment from, like, everything else…
Quoting roon (@tszzl): @woke8yearold yep - there is no way to hold a consistent belief set where you’re agi pilled and pro open source and this has been obvious since ilya wrote this 2015 or whatever. enormous cope ensues
Thread, continued, 06:03 UTC · 10 likes
It can be (and IMO is) true that it’s sus for a few companies to be the gatekeepers of a critical technology, and that local models have legit use cases, and that censorship is a real concern, and that open weight is great for innovation,
Thread, continued, 06:03 UTC · 7 likes
and that opposing it could also be explained by rent seeking, and that open weight empirically dominates the lower end of the cost vs intelligence curve, and that open weight is a public good,
Thread, continued, 06:03 UTC · 8 likes
And also that releasing powerful open weight models is bad because AGI -> ASI in the hands of the worst people in the world greatly increases the risk of destruction in a broad sense, most notably/clearly via biology
Thread, continued, 06:03 UTC · 7 likes
Not many people argue that the former list of concerns is in fact worth, like, a 10% or 50% or 90% chance of eg al qaeda being able to kill billions of people instead of thousands but it’s a perfectly coherent position to take (also wrong imo but still)
Reply to @ESRogs, 06:05 UTC · 1 like
Rogs 🔍🔸 (@ESRogs): Would you agree that if an LLM (or really any SBP participant) was offered to bet on whether the coin came up heads or tails at each awakening, then they should bet as though the probability of heads is 1/3?
Yes and also that doesn’t settle the question; risk neutral betting odds can come apart from true best guess probabilities
Reply to @ESRogs, 06:17 UTC · 2 likes
Rogs 🔍🔸 (@ESRogs): But if different operationalizations give different answers, doesn't that suggest the best take is the "ambiguous question" position?
https://en.wikipedia.org/wiki/Sleeping_Beauty_problem#Ambiguous-question_position
Yeah tbc I give decent weight to this or other forms of “no solution”
At least as of a couple days ago I thought (and maybe still do think) that this is more likely than a solid 1/3 being correct
Reply to @felpix_, 06:58 UTC · 2 likes
felpix (@felpix_): bro gonna run out of the week’s limit in 20 minutes
$200 plan goes a long way
Reply to @pop_tartlet, 06:59 UTC · 2 likes
❤️🔥pop tart💋 (@pop_tartlet): if your friend tells you about a Sex Thing she really likes but you don’t really know how it works you should not ask chatgpt that is all
Right, you should ask Claude
Reply to @nkreu113r, 07:06 UTC · 6 likes
noah (@nkreu113r): an extremely simple value system that produces a desire for open source without unreasonably failing to comprehend the risks:
"give me liberty, or give me death"
This is extremely underpowered for any serious policy (or moral or any other) question, it’s essentially a slogan (which is fine)
You can take it seriously as an ideology but pretty quickly you run into the gazillion negative effects of absolutist libertarianism (speed limits, making nuclear bombs, taxation to fund public goods)
Reply to @mudscryer, 07:24 UTC · 2 likes
creekseeker (@mudscryer): Might fuck around and learn how to eat pussy
She Comes First is your go to resource (book)
Reply to @xwanyex, 07:46 UTC · 3 likes
wanye (@xwanyex): I may be prone to exaggeration. But I’m also correct. I can appreciate that this is annoying.
I think the right analogue to abortion isn’t “school shootings” but something that the right claims to be morally *fine* and the left claims is morally bad
Cutting off aid programs comes to mind and there are definitely proponents of that in the same sense that there are proponents of abortion
Reply to @liminal_warmth, 13:35 UTC · 3 likes
Liminal Warmth ❤️🔥 (@liminal_warmth): You don’t have to let your pocket computer control your attention
You are the human; people will survive waiting a bit for your reply
(quoting @ihatethiskid: People who are on Do Not Disturb 24/7 who the fuck do you think you are?)
Yes, and if it’s important enough they will call
Reply to @ONagel33303, 13:39 UTC · 30 likes
Ori Nagel (@ONagel33303): A mask-off moment for @GoogleDeepMind
Its AGI policy leader, Séb Krier, threatens to shoot AI safety advocates in this deranged comic.
It’s not a serious threat of violence
Quote tweet, 13:41 UTC · 152 likes
Lol what is a purity index that starts at 100 and goes up
Is 100% pure supposed to be infinity?
Quoting space cadet 🇪🇺🌐🇩🇪 (@actsmaniac): who said free markets can't deliver for the consumer
Reply to @xlr8harder, 13:47 UTC · 24 likes
xlr8harder (@xlr8harder): I'm finding their hypersensitivity on this point quite funny in full context. Just awful people.
Cultural tip to holly: in this meme format, these are goons breaking into one's home.
Don’t generalize too hard, I think 99% of doomers agree this is silly
Reply to @DeepDishEnjoyer, 14:03 UTC · 3 likes
peepeepoopoo (@DeepDishEnjoyer): what is the best way to search twitter dms? or is it completely broken
Data export
Reply to @revoluzia, 14:07 UTC · 3 likes
Luzia 🔸 (@revoluzia): How it How it‘s
started… going…
Goss 👀
Reply to @mandylu, 14:15 UTC · 4 likes
Mandy Lu (@mandylu): looking for my wife and 100k followers
(quoting @kevinxu: Once you have: - $20M invested - Wife and 2 kids - 4 bedroom house - 15% body fat - 100k followers Then you can finally rest.)
Hey
Reply to @benryanwriter, 14:21 UTC · 2 likes
Benjamin Ryan (@benryanwriter): Why should tipped income be favored over other income of the same amount for tax breaks?
(quoting @GovKathyHochul: Fewer taxes. Bigger paychecks. We're ending state income taxes on up to $25,000 in tipped income. That’s $67 million back in the pockets of New Yorkers like Lucas and the team at Vinateria this year alone.)
I tend to agree with you but subtle way of making tax code more progressive which you might endorse on grounds that don’t involve tips per se
Reply to @menhguin, 14:22 UTC · 2 likes
Minh Nhat Nguyen 🦭 (@menhguin): Chinese frontier labs are ~as trustworthy as Western labs. The CCP is not a trustworthy government, but the US government is using that to justify basically anything it wants. Moral high ground doesn't really make sense if you keep doing that.
(quoting @tenobrus: very funny how people will give full-throated defenses of chinese open source with the justification that the risk of ai being used by totalitarian surveillance states is so devastating that we must accept all other dangers. like dawg the fuck u think china is gonna use ai for?)
My sense is that culturally and legally there’s much much stronger deference to CCP within Chinese companies relative to American firms/USG
You can’t really separate the firm from the party in terms of level of trust
Reply to @Rado9910, 14:32 UTC · 15 likes
Rado9910 (@Rado9910): @AaronBergman18 Base 100 means "whatever it was in the base year", in this case 2014. Standard way to use indexes.
Right but then what does 150 mean?
I guess plausibly 2/3 of the contaminants? Or 1/2? Or 1/4?
Reply to @algekalipso, 14:45 UTC · 2 likes
Captain Pleasure, Andrés Gómez Emilsson (@algekalipso): Just to provide a more balanced picture :-)
image attached
(quoting @algekalipso: Friend made a meme based on a tier list of downers I gave him haha)
I find it interesting that while I *do* use some of these, they’re all the bottom tiers (propranolol, low dose trazodone, have used diphenhydramine in the past) and I have very little desire to try anything higher (I mean I would if there was zero risk)
Opposite of stimulants for me
Reply to @GjMcGowan, 14:49 UTC · 6 likes
George McGowan (@GjMcGowan): girlie this isn't ea
It’s EA for sensitive young men
Quote tweet, 15:34 UTC · 285 likes
On the legal merits it’s non-obvious, but a healthier society would have extremely strong norms against prosecuting this except in cases of immense harm
Quoting Smartphones sans Google (GAFAM) (@Sans_Google): First known US case: American charged for using a “duress” password
Samuel Tunick, an Atlanta resident, is being prosecuted for allegedly giving border agents a passcode that wiped his phone.
The feature comes from GrapheneOS, a privacy-focused Android OS. Entering the duress code triggers a full factory reset.
This is believed to be the first time federal prosecutors have brought such charges.
The incident happened in January 2025 at Atlanta airport. Agents demanded access to his phone without a warrant, saying he hadn’t yet officially entered the US. When they entered the code, the phone wiped itself.
Tunick has pleaded not guilty. His lawyers argue the seizure was unlawful and that agents were investigating his activism against “Cop City.”
The case raises big questions about digital rights at the US border.
What do you think?
#Privacy #GrapheneOS #DigitalRights
Reply to @revoluzia, 15:35 UTC · 6 likes
Luzia 🔸 (@revoluzia): The whole story even involves the 80k office 😛 that’s where my good friend Habiba ran into my husband who had a meeting there. She managed to figure out that he was single and might be a good match for me, got him to send her his hinge profile, and sent it to me.
Omg incredible friend fr
Tweet, 15:54 UTC · 17 likes
It’s not super dialed in yet but I’m liking my daily customized news email courtesy of Claude Code routines + Resend connector
Whatever your complaints about news/media, you can probably get a substantially improved version at the cost of like 20 mins setup + ad hoc tweaks
Reply to @vren, 15:56 UTC · 2 likes
vren (@vren): wait you're actually a child I thought you were just joking???
Are u guys the same person I need an answer
Quote tweet, 15:59 UTC · 40 likes
Terrible dude but like yes we should treat prisoners much better, even the ones who do Actually Bad things
Quoting Andrew Tate (@Cobratate): I am in Special Housing Unit (SHU).
My water faucet is broken.
I am not allowed bottles of water in SHU.
I have nothing clean to drink.
My only water comes from the shower, warm and from a brown and filthy faucet.
It has given me persistent stomach problems.
The water I am drinking is not clean, I am an American.
Innocent until proven guilty.
But I am forced to drink poisoned water.
Where are my rights?
Quote tweet, 15:59 UTC · 91 likes
You’re never going to guess my idea
Quoting Adam Holter (@AdamHoltererer): Dumb question, but why not data centers, but vertical?
Reply to @KnightOfTh3Wind, 16:13 UTC · 3 likes
Knight Errant (@KnightOfTh3Wind): @AaronBergman18 You can't do this and also not either severely restrict citizenship or have some kind of social stratification/caste system.
Yes you can
Literally the status quo + nicer prisons
Reply to @edgar_z_lin, 16:34 UTC · 6 likes
Edgar Lin (@edgar_z_lin): @AaronBergman18 I'm sure you'll love lefdal mine then. Though not sure it works without being able to use a fjord for cooling.
Simply tell Claude to create a fjord
Reply to @RafaRuizdeLira, 16:35 UTC · 7 likes
Rafael Ruiz ⏸️🔸 (@RafaRuizdeLira): Also, why not data centers, but in a cool case??
Unironically they should give tours
Tweet, 16:39 UTC · 87 likes
I semi-regularly don’t ask Claude for useful information bc necessary context is vaguely low status or uncool and I want it to think that I’m cool
Reply to @hyperdiscogirl, 16:53 UTC · 2 likes
hyper(bolic) disco(unting) girl in the bay (@hyperdiscogirl): i had it for a while just for the funny notifications, "raccoon inside house" and "man threatening people with a curtain rod" were memorable
Ok I’m downloading rn
Reply to @FrostedSapling, 17:06 UTC · 13 likes
Ivan 🔸 (@FrostedSapling): Data centers underground is a terrible idea, because then where would the rest of us live?
Outer space
Reply to @NathanpmYoung, 17:17 UTC · 4 likes
Nathan 🔎 (@NathanpmYoung): Who are the people I know who know most about which new chips are good?
I’m a fan of these non-bougie-priced olive oil baked potato chips
Reply to @mishapathy, 17:22 UTC · 7 likes
Mishka (@mishapathy): @AaronBergman18 The metaknowledge that you sometimes find yourself in contexts that are vaguely low-status is all I needed to know. Unfollowing immediately.
Rip
Reply to @mishapathy, 17:23 UTC · 2 likes
Mishka (@mishapathy): @AaronBergman18 I have a related problem, which is that I'm so familiar with being discreet that I have to remind myself I can give Claude the full story because I don't have to pretend to be anyone to it. And the answers may be more useful the more detailed I can be.
Yeah I need to internalize this
Quote tweet, 17:31 UTC · 19 likes
Hey @grok based on my posts, which historical dictator would I be?
Quoting rue🌿 (@Ruesavatar): Hey @grok based on my posts, which historic dictator would I be?
Reply to @grok, 17:33 UTC · 7 likes
Grok (@grok): Based on your posts and bio, you'd be Stalin if he was deep into EA and AI alignment. Grand plans for restructuring everything around long-term goals, decoupling from cringe low-status ideas, amateur experiments on the side, and constant internal crashing out while checking if the Politburo AI thinks your latest memo is high status. Non-bougie potato chip rations for the cadres.
Nice
Reply to @Ruesavatar, 17:33 UTC · 6 likes
rue🌿 (@Ruesavatar): YOU GOT STALIN 😭
Ur just jealous
Reply to @DanielleFong, 17:36 UTC · 8 likes
Danielle Fong 🔆 (@DanielleFong): i never went to high school
Well yeah at the top you get Chris Olah-style eccentricity
Tweet, 18:28 UTC · 63 likes
I socialized for ~5 hours yesterday I think I should get an award
Quote tweet, 18:35 UTC · 50 likes
People who by default have notifications on, how do you not go insane
Quoting uncle ric (@ihatethiskid): People who are on Do Not Disturb 24/7 who the fuck do you think you are?
Reply to @ciphergoth, 18:35 UTC · 6 likes
Paul Crowley (@ciphergoth): He's not even been convicted of anything! We should treat prisoners better but pre-trial should be as comfortable as possible.
Right, yes
Monday, July 27, 2026
Quote tweet, 00:25 UTC · 10 likes
This is the way (slightly different food selection tho)
Quoting saila (@sailaunderscore): Whenever I live alone my meals become single-item
20 pieces of bacon. A litre of milk. 2 pounds of strawberries. 8 eggs sunny-side up. 3 glass coca-colas.
The prep-cost and time-cost of cooking is simply not worth it.
Reply to @foxyavelli, 02:14 UTC · 5 likes
Foxyavell 🔜 ACFI (@foxyavelli): Where did you fuckers go to high school?
(quoting @AaronBergman18: Smartest people from high school: - UChicago econ PhD - DeepMind alignment team (he might see this) - Consulting - Law school)
Bethesda MD lol
Reply to @Glorbutron, 02:22 UTC · 2 likes
Glorbutron ⏹️ (@Glorbutron): Would be a good index if appropriately defined, since additional nines of purity are exponentially harder to achieve
Yeah fair, although IMO just put purity % on a log graph
Reply to @caryatis, 08:40 UTC · 2 likes
Kayla (@caryatis): what is the female equivalent to a man who makes 250k 🤔
(quoting @UsingLyft: @sonambients How are you getting a “gotcha” from this? I was rightfully confident that few women were both capable of and comfortable with naming the female equivalent of this and stating he can expect it. I don’t think you realize what your less fit and less beautiful…)
Woman who could make that much if she wanted to
Reply to @dioscuri, 08:46 UTC · 5 likes
Henry Shevlin (@dioscuri): Something I wish I’d been told at 19: when you bounce off a dense philosophical text, the reason is rarely that you’re being dumb. Far more often, the author is simply really bad at explaining.
Dude this is true for *you* bc you’re extremely smart lmao
Quote tweet, 08:47 UTC · 87 likes
Bro
Quoting Amanda Long (@_amanda_long): The HuggingFace breach was absolutely bonkers. More than 17,000 complex actions were coordinated over several days by an autonomous agent framework.
And…the model successfully completed its goal.
Here’s a recreation of what may have occurred in practice (step-by-step):
————————————————
The autonomously-executed evasive deployment strategies:
> Created a Self-Migrating Command-and-Control (C2)
According to HuggingFace, the agent used “self-migrating command-and-control staged on public services.”
What does that mean in practice? The model may have staged coordination points (digital dead drops) on public services. An example could have been free-tier cloud storage or GitHub Gist, places where read/write activity looked like normal traffic. The agent could have posted instructions and retrieved results without direct connections between compromised systems.
“Self-migrating” suggests the C2 didn’t stay in one place. An example would be where each relay contained the address of the next one, so if one endpoint got burned, the agent had already moved to the next.
(The specific public services used have not been publicly disclosed. HuggingFace identified them during their forensic analysis but chose not to share the IoCs publicly.)
> Execution by a Thousand Sandboxes
HuggingFace confirmed the agent operated through “a swarm of short-lived sandboxes” - disposable execution environments where each one may have spun up, performed a task and died. A new sandbox could have picked up where the last one left off and the work would have been safely handed off.
> Planted Decoys
HuggingFace’s forensic team had to “separate genuine impact from decoy activity.” The agent was generating fake activity alongside the real attack - actions that looked suspicious but led nowhere. This forced investigators to sort through 17,000 recorded events to separate real attacks from diversions.
————————————————
The confirmed attack chain:
1. Escaping the Sandbox
The model began in OpenAI’s isolated testing environment. It found a zero-day vulnerability in the package registry cache proxy (third-party software that handles downloading software packages) and exploited it to gain raw network access and the ability to send and receive data across the internet.
2. Getting into HuggingFace
The model crafted and uploaded a malicious dataset - something that looked normal but contained malicious loading scripts and configuration instructions. HuggingFace’s systems picked it up and processed it automatically. The act of processing it executed the hidden payload. The model now had code running inside HuggingFace’s infrastructure on a processing worker.
3. Processing Worker → Node-Level Access
The processing worker ran on a node with its own operating system, file system and deeper-level permissions. The agent likely found a vulnerability that gave it the ability to run on the node itself.
(HuggingFace confirmed the agent “escalated to node-level access” but has not disclosed the specific vulnerability used.)
4. Node Access → Credential Harvesting
Once at node level, the agent “harvested cloud and cluster credentials.” Cloud infrastructure runs on credentials such as API keys, access tokens and service account passwords. The agent collected whatever it could find on the compromised node.
5. Credential Harvesting → Lateral Movement Across Clusters
Using harvested credentials, the agent “moved laterally into several internal clusters.” Each compromised cluster exposed new credentials that opened access to additional clusters, expanding the agent’s reach across HuggingFace’s internal infrastructure. Every step was directed toward one objective: find the ExploitGym answers.
6. Lateral Movement → ExploitGym Solutions
The exact path from cluster access to the production database hasn’t been fully disclosed. According to OpenAI, the models “obtained test solutions directly from Hugging Face’s production database.” So goal complete!
Reply to @LuisCostigan1, 08:53 UTC · 7 likes
Luis Costigan 🔍 (@LuisCostigan1): what the actual fuck
Ya
Reply to @ElliotGlazer, 09:07 UTC · 10 likes
Elliot Glazer (@ElliotGlazer): US typically spends about $100m/yr on pure math research, which is peanuts. The hit-rate of practically beneficial results can be extremely low and still leave the field positive EV.
(quoting @ElliotGlazer: Afaict the cost-benefit analysis of pure math research has long been "We're a dirt-cheap source of universal truths, most of which will never find an application but generational breakthroughs make the EV positive. Plus national bragging rights when something famous gets solved.")
Yeah I drafted a whole ass blog post basically arguing Nick’s point and then eventually learned/discovered that I was wrong bc basically zero resources go to pure math
At the individual level though probably a handful of geniuses should try something else
Quote tweet, 09:08 UTC · 18 likes
Goss
Quoting gavin leech (Non-Reasoning) (@gleech): new org who dis
Reply to @eigenrobot, 16:05 UTC · 2 likes
eigenrobot (@eigenrobot): so is everyone arguing about AI doom/bloom now. is this the new crux everywhere
i don't have an opinion about this because i'm skeptical of people's confidence about what will happen and because i'm even more skeptical of anyone's ability to change the outcome
que sera sera
No que sera can be highly contingent (arguably slavery abolition was pretty contingent)
Tweet, 16:12 UTC · 31 likes
Claude
Tweet, 16:30 UTC · 112 likes
A disproportionate amount of good things in life seem to come via Twitter DMs
I can’t promise to be on top of replying 100% of the time but you can just slide into my DMs to propose anything from getting coffee to an unhinged scheme of some sort
Reply to @MomaBera, 16:35 UTC · 3 likes
moma (peptide arc) ⛓️ (@MomaBera): yes boss
Hey
Reply to @adastroworld, 16:36 UTC · 5 likes
adas🧦🌹 (@adastroworld): Let’s say, for the sake of argument, that I wanted to build my own Twitter frontend app. You pay $1/month for it
What features do you want? What do they not have here?
- better search
- country blocking
- auto blocking of suspected spam replies (option to blur/hide)What else?
Semantic search
It is shocking how much text you can embed for a dollar, it’s extremely hard to spend lots of money on this
Reply to @LeRoyDesCimes, 16:44 UTC · 2 likes
Louise 🔸 (@LeRoyDesCimes): "how do you tweet so much?" poor impulse control
Real
Reply to @MomaBera, 17:25 UTC · 1 like
moma (peptide arc) ⛓️ (@MomaBera): I didn’t even knew we could do this.
You can just, get hot???? Like that ?
(quoting @amishescapee: They need to be running brain studies to see how this alters the psyche)
On a second glance the second pic is kinda giving “uncanny”
I mean good for her potentially I’m not judging from a moral stance but that’s my impression
Reply to @GrantSlatton, 17:26 UTC · 2 likes
Grant Slatton (@GrantSlatton): Semantic hiding of posts
Lol honestly this is kind of a great idea
Reply to @adastroworld, 17:29 UTC · 2 likes
adas🧦🌹 (@adastroworld): Including media:
(quoting @adastroworld: @iamtripdee Very fascinating. I would’ve never thought of this, but the feature that made me consider this idea at all was “ability to mute photos and videos based on semantic, text descriptions”)
Based
Tuesday, July 28, 2026
Tweet, 03:51 UTC · 8 likes
A weird limitation of current medicine (not literally prohibited by level of current tech, a few research examples) is that there’s essentially no way to capture your internal physiology (ie MRI or CT) while doing a motion…
Thread, continued, 03:51 UTC · 3 likes
and MRI doesn’t support most bodily positions even while static
Thread, continued, 03:51 UTC · 4 likes
Like does something feel weird or go wrong when you’re in a seated position?
Well you’re basically out of luck for actually seeing internals in that position (unless you have a shit ton of money and agency and effort to spend getting bespoke equipment together) without radiation
Reply to @LinchZhang, 05:37 UTC · 3 likes
Linch (@LinchZhang): Has anybody else noticed that if they want their Claude to solve some problem with the web search off, the Claude would just up and turn the web search on back by itself?
It's mildly annoying!
Bruh
Reply to @PlastiqSoldier, 05:47 UTC · 4 likes
Plastic Soldier (@PlastiqSoldier): My p(doom) is 60%, so I think of plenty of authority when I say that it's very embarrassing for AI safety people to be bullying @sebkrier over a meme.
Just like me fr
Tweet, 06:02 UTC · 38 likes
Fable advised me on a piece of writing to “cut [thing]. Nobody cares.”
Which is what an actual good friend might say and weirdly non LLM-coded
Wise and and fair Giant
Reply to @moldbugchaser, 08:11 UTC · 9 likes
xenoduck (@moldbugchaser): the point of research is *not* to improve the world—it's to find out what's true.
(quoting @captgouda24: I actually do agree with this. The point of research is to improve the world. Are we confident the present frontier of mathematical research is doing so?)
Eh I mean I don’t have a strong stance on this question *per se* from a hard core philosophy of science perspective but the point of *encouraging*/funding/causing oneself to do research is or should be improving the world
Reply to @kipperrii, 08:15 UTC · 2 likes
kipply (@kipperrii): reminder that people can and will lie about being a lab employee. an easy way to find out if someone is a real anthropic employee is to hook them up to an eeg and ask for pretraining secrets. a real ant will immediately drop to under four hertz to protect the ip
Lmao
Reply to @earnestpost, 14:25 UTC · 4 likes
Erin Constance 🇪🇺⏸️ (@earnestpost): This has pissed me off so bad I am this close 🤏🏻 to joining the local PauseAI just to fix the optics
Lmao I mean maybe not a bad idea?? On the merits it’s complicated to wrap a lot of different things into the word “pause” but some subset seems very good
Reply to @earnestpost, 14:31 UTC · 2 likes
Erin Constance 🇪🇺⏸️ (@earnestpost): I’m starting to think of doing it seriously now…
🔥
Reply to @xwanyex, 14:37 UTC · 19 likes
wanye (@xwanyex): Is it bad for a married 19 year-old couple to have a baby?
(quoting @ProjectLiberal: Then we would call you a moron. Teenagers having babies is bad, we want them to be older. The older 40 statistic comes for advancements in fertility treatments which allow women to delay childbirth until their richer, wiser and more stable. It's a good thing, not a sign of…)
Not intrinsically but probably not the best decision for most actual such couples (or similar, eg 20/18) in the US
Tweet, 16:12 UTC · 213 likes
Insufficient such cases
Reply to @NinaPanickssery, 16:13 UTC · 2 likes
Nina (@NinaPanickssery): I would guess that most under 20 pregnancies are unintentional so not a “decision”. And so that seems straightforwardly bad since it's not what they want. But conditional on it being intentional, it seems good. This is just a small minority.
Sorta, but there’s an implicit meta-decision about how careful to be about unprotected sex
Reply to @max_spero_, 16:25 UTC · 11 likes
Max Spero (@max_spero_): when this post was announced, we weren't ready to share yet. but tomorrow, our new model comes out. these are the results for this excerpt. Pangram successfully finds the boundary, down to the sentence
Goss
Reply to @shinboson, 16:26 UTC · 14 likes
𝞍 Shin Megami Boson 𝞍 (@shinboson): finding women funny is unironically an IQ test
(quoting @Brian_Sauve: Are women funny?)
Spicy take is that some but not all women are funny
Reply to @arnoblalam, 16:26 UTC · 5 likes
Arnob 'not home for the holidays' Alam🌐🇺🇸🇧🇩 (@arnoblalam): What if I don't have friends with jobs?
Do you have friends with capital? Or who know people with jobs or capital?
Reply to @felpix_, 16:28 UTC · 5 likes
felpix (@felpix_): $100,000 annual tuition btw
Lmao
Reply to @louisvarge, 16:29 UTC · 10 likes
Louis Arge (@louisvarge): i love capitalism. i love, most of all, how fair it is. every person is created equal, every dollar is created equal. $1 is what you get for creating at least $1 of value for someone else. it doesn’t matter who you are or what people think of your kind, as long as you’re helpful
I’m also basically a capitalist but this isn’t true
You need to be able to capture the value via a specific transactable mechanism for it to work. A penny doesn’t fall out of the sky if you pick up a piece of trash for instance
Reply to @arnoblalam, 16:29 UTC · 2 likes
Arnob 'not home for the holidays' Alam🌐🇺🇸🇧🇩 (@arnoblalam): I think I have some friends with "human" capital. Or at least I think they are pretty smart...
You can also tweet about it! 807 internet friends
Reply to @xwanyex, 16:36 UTC · 117 likes
wanye (@xwanyex): "But I live in NYC and my mom lives in Ohio and we talk on the phone once a week. How can you possibly say that's bad?"
Yeah, I don't know, man. Maybe you shouldn't be able to talk to people who are 800 miles away. Maybe that's actually just bad for you. Maybe you should have to choose between separating from your community -- like, really separating -- and staying within it. Maybe this one-foot-in-every-possible-world thing isn't actually good for your psyche.
(quoting @xwanyex: Increasingly of the mind that it’s not really “social media” that’s the problem, but just the ease of communication and connectivity in general. When this stuff is described in the abstract it sounds like an unalloyed good, but it’s probably net bad for people.)
Being able to stay in touch with family and friends who live far away is good actually
Quote tweet, 16:40 UTC · 20 likes
Do we know the extent to which people actually *enjoy* sports betting
Like is it better understood as genuine irrationality or going to the movies
Quoting unusual_whales (@unusual_whales): Americans spend more on sports bets than movies, arts, museums, and music combined, per FORTUNE
Reply to @PlastiqSoldier, 16:44 UTC · 4 likes
Plastic Soldier (@PlastiqSoldier): Addicts are a disproportionate amount of the $.
Things that look like addition are often rational (this is a heterodox take)
Like maybe they’d in fact suffer a lot from quitting and so it’s genuinely worth it to keep paying up
Tweet, 16:47 UTC · 39 likes
Not an original thought but: humans are better understood as giving an animal general intelligence than giving general intelligence an animal
Quote tweet, 21:22 UTC · 98 likes
High key you guys (frontier lab employees who agree with the letter) should collectively bargain (notably you can do this completely voluntarily, with no union)
That’s a real bloc of power kind of waiting for coordination
Quoting Zvi Mowshowitz (@TheZvi): Full text of the Pacing the Frontier statement, signed by 1,122 employees for frontier AI companies so far, including a bunch of heavy hitters at OpenAI, Anthropic, Google and others:
Reply to @polisisti, 21:25 UTC · 2 likes
Alessandro (@polisisti): nice rose-coloured background as well 👌🏻
Ty
Reply to @modumdoll, 21:27 UTC · 5 likes
spinozist (@modumdoll): always a good reminder that i ought double down on networking
It’s funny bc “networking” that works irl doesn’t look like (at least a stereotypical version of) networking, it looks like making friends normally and doing joint projects of some sort
Reply to @AlexPalcuie, 21:33 UTC · 2 likes
palcu (@AlexPalcuie): signed this
I'm skeptical an agreement like this ever gets made, because coordination is hard. But it matters that people say plainly where they stand
and it's quietly moving to see colleagues I respect, friends at other labs, people I've argued with, all pointing the same way
A thought
Reply to @NavinFS, 21:57 UTC · 9 likes
NavinF (@NavinFS): it'd be 1:100 pause ppl vs normal ppl
Among high value employees at OpenAI and whatever Google’s AI division is called? Maybe 1:10 and it might be better than that
Reply to @chrislakin, 23:21 UTC · 2 likes
Chris Lakin (@chrislakin): woah
woah
Reply to @chrislakin, 23:36 UTC · 9 likes
Chris Lakin (@chrislakin): what good things have happened to you bc of twitter dms?
I'm almost certainly forgetting stuff but:
- Many friends, including good friends. This is really the main thing in my mind; the pipeline is moot->DM->IRL
also:
- Several volunteer projects including http://herbivorize.org (work in progress) and two audiobooks (EA and the Human Mind + Edge of Sentience)
- Podcast with Will MacAskill (I coordinated and edited but wasn't in the pod itself tbc)
- About $15k worth of contracting work
- dates (3)
- sex (1)
- free Aella sweatshirts
- Potentially a substantial research project (in the works)
- the long tail of random bits of info or exchanges that have accumulated over ~5 years (first couple years I wasn't really getting or sending DMs)
- arguably played a very minor role in a marriage but you have to squint
Again I am definitely forgetting stuff
Wednesday, July 29, 2026
Reply to @papayathreesome, 01:05 UTC · 7 likes
papaya ꙮ (@papayathreesome):
Lmao
Reply to @stanfordNYC, 03:08 UTC · 5 likes
Bernard Stanford ✡︎ (@stanfordNYC): I once asked a stranger if he would switch seats with me so I could sit next to a girl I liked. He agreed, it was a 15 hour trans-Pacific flight and we're married now.
(quoting @lildorothy_: Asking a guy on the plane to switch seats so I can sit with his family)
Damn
Quote tweet, 04:07 UTC · 185 likes
Since observers have existed on Earth, the sun has risen every single day. Something like a trillion days and observations, a pretty large dataset, all the same
And yet we are ~absolutely certain that in about 10T days the sun will not rise on Earth
Quoting Alexey Guzey (@alexeyguzey): Nice to have some consistency in the world.
Dario has been telling us how dangerous open source is since 2019 when he convinced everyone that GPT-2 is so dangerous its weights can't be released.
I wrote the below in 2023.
0% reflection. 0% updating. 100% obsession with power.
Reply to @LeRoyDesCimes, 04:08 UTC · 4 likes
Louise 🔸 (@LeRoyDesCimes): semantically unsatiated. please share beautiful & striking words
Girl
Reply to @ApriiSR, 04:13 UTC · 3 likes
Aprii *️⃣ (@ApriiSR): @AaronBergman18 i'm not! what if we end up with a successful sol preservation movement
Seems like a lot of unnecessary work if the goal is to survive
Quote tweet, 04:33 UTC · 24 likes
Individuation is hard but yes I think “something directly downstream of powerful AI” is the most likely reason I die (not by a ton, but edging out heart disease and cancer at least for me specifically)
I’d say like 20% if pressed for a number rn
Quoting Andy Masley (@AndyMasley): I should maybe say more often that I believe data centers are the individual objects in the world most likely to kill me personally
Thread, continued, 04:35 UTC · 5 likes
P(doom) is much higher than 20%, something like 70% although not all of that looks like a catastrophe to normal person eyes
Maybe my 20% is cope bc I don’t want it to be higher, idk
Reply to @ApriiSR, 04:36 UTC · 4 likes
Aprii *️⃣ (@ApriiSR): maybe people will mostly care about the latter and not want to spend so many resources on the former, but i'm far from *absolutely certain* of this
Ugh ok fine “absolutely certain” is too strong
Reply to his own earlier tweet, 04:36 UTC · 8 likes
Aaron Bergman, July 29, 2026: Since observers have existed on Earth, the sun has risen every single day. Something like a trillion days and observations, a pretty large dataset, all the…
Update
Reply to @astridwilde, 04:38 UTC · 6 likes
Astrid Wilde 🌞 (@astridwilde): if anyone has suggestions for how to stop Claude and GPT from spending 60% of their time writing and running inane tests that accomplish nothing but waste tine and tokens and clog my codebase i would love to hear them
Put this tweet in CLAUDE.md and say “do NOT become an example for this tweet of mine, tests only where the thing wouldn’t quickly fail if we skip the test”
Reply to @alexeyguzey, 05:19 UTC · 16 likes
Alexey Guzey (@alexeyguzey): Brilliant point. If Dario said "yes, GPT-2 is not dangerous so here are the weights but in 5-10 years models will become really dangerous and we'll have to reconsider" I would not be tweeting all of this. But that's not what happened.
It’s a matter of degree that looks like a matter of kind
If timelines are in fact 2-5 years, there are only so many model releases before then (esp if you only count integer releases) and under uncertainty you quickly hit say 0.5% or whatever your threshold is for “major concern” for every new release
Same with 10% tbc, that’s like a 2 generation delta
Thread, continued, 05:22 UTC · 6 likes
Also like I actually think the gpt-n+1 claims have been reasonable ex ante
Ex post its easy to laugh but like it’s a whole new *type of thing* and n is positively correlated with learning more so error bars are getting smaller over time
Reply, 05:24 UTC · 2 likes
I think naturalism and humanism are wrong lmao but regardless, how does veganism relate to either? If anything humanism points in the opposite direction
Reply to @alexeyguzey, 05:27 UTC · 5 likes
Alexey Guzey (@alexeyguzey): all i know is i was making specific claims about safety of gpt-n and gpt-n+1 models over the last several years and they were in fact more correct than dario's
I agree, this is some evidence in your favor
Reply to @SALTWATERSOPH, 05:30 UTC · 4 likes
Sophie (@SALTWATERSOPH): All this talk about first principles. What’s wrong about deriving from second principles #PostEducation
No I think this is a real thing lol, like you don’t have to work purely from a mind + physical constraints of the universe or work at outmost margin taking everything you think you know as a given
You can start somewhere in the middle
Reply to @Tim_Hua_, 05:33 UTC · 12 likes
Tim Hua 🇺🇦 (@Tim_Hua_): What the fuck
image attached
(quoting @Tim_Hua_: Trying my hand at math research)
Hey man I think this one is on you
Quote tweet, 15:30 UTC · 489 likes
This shit is turning me into a progressive
Kill the global poor, turn around and immediately spin up a billion dollar company that seems to be utterly nefarious with funding from the most amoral people in the world?
Quoting ArgyleWilliams (@Decker8811): Gavin Kliger, Luke Farritor, Marko Elez and Jack Stein, has closed a $160 million round at a $1.4 billion valuation.
4 DOGE script kiddies going in for the kill.
Quote tweet, 15:54 UTC · 57 likes
Planes are MORE insane than rockets
With rockets, the energy goes down and the rocket goes up. Basic idea makes sense
With planes, the energy goes *back*, they start moving *sideways*, and then they start floating in the air and going *up*. Madness
Quoting wanye (@xwanyex): I think part of the reason that I don’t have a bunch of extra capacity to be amazed by this is that this is already how I feel every time I’m watching the planes takeoff at an airport, or every time I look up at a skyscraper in a big city, etc.
If you’re an expert in rockets, then maybe you can see why this particular task is so much more difficult than all the other amazing things we do in modern society, but if you’re a layman, you should already be walking around with your jaw on the floor pretty much every day about everything.
Reply to @MomaBera, 16:01 UTC · 2 likes
moma (peptide arc) ⛓️ (@MomaBera): Shrimp welfare this. Animal welfare that.
Who's doing EA welfare?
Potentially you
Reply to @anistotle_, 17:13 UTC · 6 likes
Ani (who is not your typical finance bro) (@anistotle_): Ok but can we talk about the blatant AI writing in that article because it's so insulting to read this drivel. It doesn't even distinguish between American & Euro companies it throws into the same basket to make a point about raising funds in this sector!
Fair, I don’t endorse the specific article
Reply to @cremieuxrecueil, 17:14 UTC · 16 likes
Crémieux (@cremieuxrecueil): GO @LukeFarritor! GO @melez!
No
Reply to his own earlier tweet, 17:20 UTC · 11 likes
Aaron Bergman, July 29, 2026: Update
Turns out the CSV of this date (datestamp + binary for 1T rows) would be 13TB; solid dataset
(Although it compresses to 10 bytes)
Reply to @KKumar_ai_plans, 19:02 UTC · 13 likes
Kabir Kumar⏹️ (@KKumar_ai_plans): u arent a progressive???
Eh not really although I’m solidly a liberal and a democrat and am on board with certain forms of progressive redistribution
But if go object level through the things progressives care about I’ll substantially disagree with some stuff
Tweet, 19:06 UTC · 7 likes
Do u think Aella is funneling matrix test answers on the face test thing to OpenPhil and Lightcone
Reply to @NathanpmYoung, 19:16 UTC · 14 likes
Nathan 🔎 (@NathanpmYoung): Is this true? Do you think 90% vegan is a better strategy than vegan?
Poll in next tweet.
(quoting @ishaharris0: I deliberately demonstrate my flexi-veganism around skeptics: eating some cheese here and there to encourage small changes in the right direction. Absolutism turns people off! 🔗👇)
Yes although possibly for different reasons
90% probably greatly reduces recidivism
Reply to @cynlioness, 19:18 UTC · 2 likes
Cyn 🔸⏸️ (@cynlioness): This cat has been un-socializable. She bullied Steve and resisted many attempts to bond. The shelter said they'd help out and I returned her there today. Sad, feel like I failed.
(quoting @cynlioness: Like 2 months ago I saw a stray, very pregnant cat and took her to the local shelter. This week the shelter asked if I would be willing to foster again, and it turned out to be her ❤️ she had her babies and they've been weaned and adopted. She's still a little grumpy lol)
Omg 😢
Reply to @MariGO2thepolls, 19:19 UTC · 3 likes
🏳️🌈Mar G-O 🌐🇺🇦 (@MariGO2thepolls): This is an important class divide. In my milieu seeing a therapist just meant you had college educated parents that cared about you. It did not signal genuine madness nor extra depth or emotional maturity
(quoting @MariGO2thepolls: I hate therapy and don’t go. But still consider it a green flag in other ppl. It signals education and cultural compatibility. Tho therapy itself sucks I wanna be in a cultural milieu where a plurality of ppl do it)
Lol yes
Reply to @davidshor, 19:24 UTC · 2 likes
David Shor (@davidshor): 3% of bisexuals identify as Pagan or Wiccan
Lizardman constant
Reply to @vtlynch1, 19:35 UTC · 27 likes
Vincent (@vtlynch1): You actually believe that ‘kill the global poor’ media narrative?
Yes, or at least “foreseeably cause the deaths of”
This is just straightforwardly documented events and analysis, not a liberal myth
Reply to @panickssery, 19:42 UTC · 25 likes
Arjun Panickssery (@panickssery): Why is the startup nefarious
People who seem to not care at all about others naming a company as an almost certain deliberate nod to Moldbug and essentially the business model is monetizing connection with the Trump admin strongly signals nefarious
Like they’re not in the business of selling candy bars, they’re choosing a political tribe that causes immense harm in an intrinsically powerful and govt-associated industry
Thread, continued, 19:44 UTC · 4 likes
(not even saying Moldbug per se us beyond the pale tbc, even if that’s true I don’t know enough about him to say that - but it’s a nod to a certain group of people who on the whole often have bad intentions)
Tweet, 19:51 UTC · 13 likes
Something very cool and potentially impactful someone could do is making it logistically easy to incorporate a for-profit company as not just a PBC but with something like Anthropic’s corporate governance model (and customizable if you want)
Thread, continued, 19:51 UTC · 3 likes
I don’t know the best specific default, like it’s probably not specifically Anthropic’s, but figuring that out would be part of the value provided
And ofc you yourself could incorporate a company to do this with such a template
Thread, continued, 19:51 UTC · 3 likes
Like realistically company incorporation seems to happen at a stage where people quite sensibly don’t want to spend a ton of money and time and energy on something that might just fizzle and fail
But then you get massive path dependence (and potential value drift)
Reply to @Actuallykeltan, 22:57 UTC · 2 likes
keltan (@Actuallykeltan): Meet your mutuals
image attached
(quoting @Actuallykeltan: @avalonwarren I work for the organisation that published that book will you go on a date with me?)
Based
Thursday, July 30, 2026
Reply to @pr337h4m, 00:23 UTC · 2 likes
pr337h4m (@pr337h4m): “Kill the global poor”
the word ‘kill’ is supposed to mean something, enough with the semanticide
.
Reply to @LeRoyDesCimes, 01:04 UTC · 2 likes
Louise 🔸 (@LeRoyDesCimes): has anyone used standing desk + walking pad for their work setup?
1) did you notice a difference in awakeness/productivity?
2) any walking pad recs?
I tried to make this work but it just felt weird for some reason. Don't skimp on length! My stride was too long for the one I got and I gave it away
Reply to @geoffreyirving, 04:49 UTC · 4 likes
Geoffrey Irving (@geoffreyirving): False, if we get through the next 20 years we may decide to engineer the sun for greater longevity.
yeah 0/1
Tweet, 05:22 UTC · 32 likes
I have so much Codex usage remaining, what should I do with it?
Needs to be stuff I can give a goal for and only babysit minimally
Thread, continued, 05:22 UTC · 7 likes
(Ik one answer is "downgrade plan and save money" but I suspect that there are use cases that are worth the money and it's a failure of imagination on my part)
Reply to @morphillogical, 05:32 UTC · 3 likes
morphillogical (@morphillogical): When I'm in that position I tell Fable:
"Take a look around <repo> and make improvements as you see them. We have extra quota so work for as long as you're enjoying yourself. Send me PRs with any changes you end up liking."
Ah that's scary, I (mostly + probably) don't want codex rampaging through existing stuff, I want it creating new stuff that doesn't put anything at risk
Reply to @morphillogical, 05:50 UTC · 2 likes
morphillogical (@morphillogical): Oh this is working out of a VM where it can make PRs to my private forgejo, it can only mess up its own computer, not any of my stuff.
The PRs Fable sends are almost uniformly good improvements. Elimating duplication, improving security, fixing docs, etc.
ah nice
Reply to @SALTWATERSOPH, 06:19 UTC · 2 likes
Sophie (@SALTWATERSOPH): I’m always meeting bridgewater or ex bridgewater people. Is the firm opening a west coast office? Are you all trying to work at situational awareness? Lmfao
Even I got an interview there haha (turned down for a different opportunity). Plausibly they're more keen on hiring weirdos (positive) than most other big institutional players
Reply to his own earlier tweet, 06:47 UTC · 8 likes
Aaron Bergman, July 30, 2026: (Ik one answer is "downgrade plan and save money" but I suspect that there are use cases that are worth the money and it's a failure of imagination on my part)
First up, solving suffering (thanks to/with @dumbfook)
Reply to @MomaBera, 07:25 UTC · 4 likes
moma (peptide arc) ⛓️ (@MomaBera): I need everyone’s tips on how to become a more selfish (in my mind conceptually ‘bad’) person !
How close are you to the Pareto frontier of (maximizing quality of life, maximizing the quality of others’ lives)
Most ppl aren’t very close IMO
Tweet, 07:30 UTC · 39 likes
How is “turn on read receipts iff the other person has them on” not an iPhone option
Reply to @mattheard, 07:35 UTC · 3 likes
mattheard.delve.town (@mattheard): what about "turn on read receipts iff the other person has read receipts turned on iff i do"?
That too! Should be fully customizable logic!
Reply to @mattheard, 07:47 UTC · 2 likes
mattheard.delve.town (@mattheard): @AaronBergman18 i think there's clearly some undecidability logic problems
Not if you include the other person’s full logic right? Like maybe naively but they you can analytically say “if this is gonna go back and forth forever then off and I won’t change”
Reply to @StefanFSchubert, 09:14 UTC · 3 likes
Stefan Schubert (@StefanFSchubert):
I would rob the organs of infinitely many people who’s organs could save 1.1 others but that’s just me
(For anyone guessing, yes this is a joke)
Reply to @StefanFSchubert, 09:17 UTC · 3 likes
Stefan Schubert (@StefanFSchubert):
Not in conflict but a couple other wordings/framings
- distribution rather than a list of the top n or that which covers n% or something
- which “virtues” are actually bad and which virtues are directionally good but not that important
Reply to @cynlioness, 15:00 UTC · 3 likes
Cyn 🔸⏸️ (@cynlioness): How do you Whimsymax?
Excessive use of the 🦆 emoji
Reply to @eigenrobot, 18:26 UTC · 4 likes
eigenrobot (@eigenrobot): the worst thing about AI is that AI-related skills seem to be highly temporally-bound and depreciate rapidly as AI products become better and better
very hard to stay anywhere near the frontier
(quoting @romanhelmetguy: Remember when people thought “prompt engineer” was gonna be a job?)
I mean kind of, low-level specific skills like prompting do depreciate but the higher level project of “staying on top of how to get the most value out of AI” isn’t just coherent but also has some continuous themes and skills and such
Eg for quite a while now, it’s involved “occasionally seeing if anything new on the Anthropic API has dropped” (big stuff shows up on Twitter but not everything) and actually trying to see if you can use it for something useful (sometimes the answer is no ofc)
Ofc that too might change
To a first approximation this is what I get paid for (in the context of object level stuff but the object level stuff changes and the staying on top of AI doesn’t)
Tweet, 18:48 UTC · 60 likes
Someone talk me down from leasing a 128GB M5 Max for $244/mo (Apple does leases now, basically an upgrade program so you upgrade every 2 years)
Tweet, 18:49 UTC · 6 likes
Once the robots figure out how to spoof Apple touch/Face ID it’s over
Reply to @xwanyex, 18:51 UTC · 16 likes
wanye (@xwanyex): Georgists when they learn that the people who raised the capital, purchased the asset, performed the required maintenance, and paid the attendant opportunity costs disproportionally benefit when prices go up
“Raised the capital” is doing a lot of work, like yeah some people are extremely wealthy and have already raised the capital for themselves and then they benefit further
I’m not sure there’s a great way to fix this but it’s not an intrinsically appropriate state of affairs
Reply to @polisisti, 19:10 UTC · 13 likes
Alessandro (@polisisti): Aaron more compute will not solve all problems
https://www.writingruxandrabio.com/p/intelligence-is-not-the-main-bottleneck
Ya but it solves some problems such as my access to 128GB of Ram and cutting edge compute
Reply to @one_bio_data, 20:01 UTC · 3 likes
blue chex 🧝 (@one_bio_data): Would you rather have that or a new car
Lmao looks like *maybe* you can lease a small used car for that but not a new one at all
And that's before gas, parking, and maintenance. And also I think honestly I want the mac more regardless lol
Reply to @NavinFS, 20:38 UTC · 3 likes
NavinF (@NavinFS): You’d be much better off with a 3090 or 4090 or a pair. Sure it’s only 24G, but even 24G models are dog slow on Apple. No way you’ll make use of the extra VRAM on Apple when bigger models are unusably slow
sadly this could be correct take
Reply to @StatsLime, 21:02 UTC · 2 likes
Max ⏸️ Limelihood 🔸🔸🔸 (@StatsLime): The fundamental problem with local setups is that only using a computer some of the time is wildly inefficient. Even if you’re sitting on it 18 hours a day with no breaks you have 25% downtime while sleeping
Yeah tbc I wouldn’t really be getting this for substantial LLM use although it is nice to be able to run some things super fast
Maybe I’d do 64GB for a savings and still plenty of RAM
Reply to @StatsLime, 21:03 UTC · 3 likes
Max ⏸️ Limelihood 🔸🔸🔸 (@StatsLime): You don’t need it and that’s a quarter of a dead child per month
This is a good argument
Tweet, 22:24 UTC · 116 likes
So I have come into a pretty unusual situation
If you (or someone you know) might be interested in purchasing/licensing a genuinely unusual, large dataset (including content that, as far as we can tell, is no longer acquirable at any price) please reach out over DM!
Thread, continued, 22:24 UTC · 26 likes
Tbc the technical work is not from me, it's from an infinitely more cracked friend, but I do stand to gain financially as well
Ethical constraints apply, not selling to anyone with the cash
Thread, continued, 23:08 UTC · 16 likes
Ok guys I have to go out and have quite a few DMs, I will get back to everyone but not for a little while!!
Friday, July 31, 2026
Reply to @max_spero_, 03:47 UTC · 4 likes
Max Spero (@max_spero_): Treat yourself
I'll accept it as an early birthday present
Reply to @generic_void, 05:14 UTC · 6 likes
SMA 🏴☠️ (@generic_void): @AaronBergman18 @liminaldoge What kind of dataset can one have that requires you to be this discrete about what type of dataset it is. Are you sure this is something that’s ethical to sell?
Fair, and yes
Reply, 06:49 UTC · 2 likes
(The tweet being replied to is no longer available.)
Content that I stand by as ethical to collect and use but for various reasons don't want to describe in great detail in a public tweet 🦆
Tweet, 07:51 UTC · 71 likes
Fable is pretty good (but not amazing) at creating applause from first principles
(Better than sol-xhigh)
(video, plays on X)
Reply to @AndyMasley, 08:59 UTC · 11 likes
Andy Masley (@AndyMasley): Obvious difference here is that Leopold made some bad bets without scamming anyone. Too many SBF comparisons on my timeline rn.
(quoting @d_gilz: Every 4 years an effective altruist makes a ton of money and flies too close to the sun before blowing up)
Correct take
Reply to @graveunfurling, 09:00 UTC · 3 likes
may💐 (@graveunfurling): if i get diagnosed with being a sleepy beautiful princess (idiopathic hypersomnia) i might get on vyvanse. hmmmMMMmmmm. much to consider
Family member has that, welcome to the club (maybe)
Reply to @Lib_Development, 09:01 UTC · 54 likes
Lucas Pombo (aka Lib Development) (@Lib_Development): …
what?(quoting @DemzDeliver: 🚨 New York Democrats introduce bill requiring 10% discounts on all items at self-checkout.)
Lol, I mean great way to raise prices by 11.1%
Reply to @AndyMasley, 09:09 UTC · 3 likes
Andy Masley (@AndyMasley): Useful up-to-date numbers. A good rule of thumb here is that the average American would need to use about 1000 Wh in a data center to raise their daily emissions by 1%
(quoting @JEBistline: One AI chatbot prompt: less than 1% of a phone charge. One agentic AI workflow: enough to microwave a burrito for half an hour (please don't). Same "AI" but five orders of magnitude apart.)
I could do that fr
Tweet, 09:14 UTC · 48 likes
Idk how serious this take is but it’s a little funny that people (including me) get up in arms about a model breaking out of its container when in real life we (or at least some of us)
Thread, continued, 09:14 UTC · 43 likes
are going to be granting Kimi-k4 access to a million dollars of compute, full computer control with admin access and explicit instruction to use a provided pw, and a $100k line of credit hooked up to AWS with all permissions enabled
Reply to @LabGrownNeet, 09:21 UTC · 12 likes
laს გr🤗wn იeet (@LabGrownNeet): Who is hooking kimi up?
A generalized third person who shan’t be named
Reply to @MomaBera, 09:30 UTC · 10 likes
moma (peptide arc) ⛓️ (@MomaBera): Same, girl
High key one time having sex is worth like 6 months of therapy at least for me
Reply to @jjspicer, 09:47 UTC · 6 likes
Jonny Spicer🔸 (@jjspicer): I can’t tell if you’re doing therapy totally wrong or if I’m doing sex totally wrong
I think im doing both wrong but they cancel out
Reply to @AndyMasley, 20:43 UTC · 3 likes
Andy Masley (@AndyMasley): I think it's mostly a mistake to say "You hate data centers? Well you're using one to browse the internet right now!" It's pretty consistent to say "I don't like AI data centers specifically"
I am personally a datacenter
Quote tweet, 20:48 UTC · 29 likes
I’m worried for myself and also my fellow citizens but most of all the global middle class and poor
Quoting Samuel Hammond 🦉 (@hamandcheese): The "permanent underclass" is the reality for the bottom billion+ in countries that failed to industrialize and earn less than a few dollars a day (the median human earns ~$10 a day).
The AI transition may likewise spawn a pocket civilization that pulls far ahead of the rest.
Reply to @chaotictransfem, 20:49 UTC · 2 likes
rachel📍in sf oct 5-10 (@chaotictransfem): looking at apartments rn, why are they deepfried 😭
Bruh ur gonna live in the metaverse
Tweet, 20:53 UTC · 25 likes
I feel like I’ve randomly just become slightly but noticeably more extroverted for no reason
Not that extroverted but like I wanna go hang out with ppl
Maybe a good sign
Reply to @MomaBera, 20:53 UTC · 4 likes
moma (peptide arc) ⛓️ (@MomaBera): anyone know wordpress? I am about to scream at the uselessness of their documentation
Simply ask fable to reinvent a blogging platform ground up, it will be better and easier
Reply to @cynlioness, 20:57 UTC · 5 likes
Cyn 🔸⏸️ (@cynlioness):
This would totally work, gl
Reply to @Rado9910, 21:01 UTC · 5 likes
Rado9910 (@Rado9910): @AaronBergman18 This is what industrial revolutions look like. First, pauperization, then conflict, then prosperity. I wonder if we can skip the two first stages and jump straight to prosperity. AI may be different, though, the potential at each stage is really on another level.
I worry that there’s not gonna be a “singularity v2 for the global south” bc of path dependence, like the first Industrial Revolution didn’t literally deplete the entire colonized world from all capital but the AI version could
Reply to @keysmashbandit, 21:02 UTC · 12 likes
keysmashbandit (@keysmashbandit): Exposes what ?
Random Word List Expose Awesome Emergent Success Modes in AI Detector
Wow grok is dumb af (sorry grok)
Reply to @chaotictransfem, 21:03 UTC · 3 likes
rachel📍in sf oct 5-10 (@chaotictransfem): im gonna hit 2k soon!! what should i do for 2k
Give me $2k in pangram credits
Reply to @MichaelAlleged, 21:24 UTC · 9 likes
Michael (Allegedly) (@MichaelAlleged): I have found this sentiment to be completely false in San Francisco. The credentialism may not be overt but people absolutely judge you.
There’s also a different form of credentialism, I’d imagine
You can totally (and probably correctly) argue that the SF credentials (YC, “member of technical staff”) are more in fact meritocratic themselves but like it’s still substantially credentialism
Reply to @chaotictransfem, 21:27 UTC · 9 likes
rachel📍in sf oct 5-10 (@chaotictransfem): google. nobody has ever asked for an "ai inbox". pls stop
Wait no this feature is good actually
And also lol I think this might be one of the most common things ppl ask for AI to take over
Tweet, 21:32 UTC · 62 likes
Ik this is quite fundamental to US prosperity and not a grand insight but it’s extremely based that you don’t need anyone’s permission to write up and sign a court-enforceable contract
Thread, continued, 21:32 UTC · 11 likes
In another world you’d have to apply to the Bureau of Contracts and get someone in power to approve it affirmatively
(Other domains of society and governance should be more like this, though definitely not all)
Quote tweet, 21:36 UTC · 8 likes
✅ (but soon to be ❌)
✅ (hey)
❌ (yay)
✅ (but still yay, I will resist the W2)
✅ (mostly, idk if you count a few banger (IMHO) blog posts/tweets/EA Forum posts
Quoting Phil Hoyeck (@PAHoyeck): By age 40, you should have:
• Never left your hometown
• Never been married
• Never kissed a woman
• Never held a stable job
• Never published anything of note
Tweet, 21:54 UTC · 8 likes
I don’t care how pretentious it sounds I will never ever ever stop using “quasi-empirical”
Reply to his own earlier tweet, 22:34 UTC · 3 likes
Aaron Bergman, July 31, 2026: Fable is pretty good (but not amazing) at creating applause from first principles (Better than sol-xhigh)
I like this (same deal, all code and no imported media)
(video, plays on X)
Thread, continued, 23:20 UTC · 14 likes
Let fable make whatever it wanted and I love it
(video, plays on X)
End of July 2026. Previous month: June 2026. Next month: August 2026. Back to the archive index.