Andre Antunes retweeted
Michael Phelps went six years without missing a day of training. 365 days a year. Christmas morning, New Year's, Thanksgiving.
People assume I made him do that. It was the other way around. 🧵
Andre Antunes retweeted
A 37-year-old man with a rare blood disease was being sent to hospice.
An AI model searched thousands of existing drugs and ranked a combination no one had tried for his condition.
He responded within a week. He is in remission.
The drugs were already on the shelf. The AI model found the match.
Andre Antunes retweeted
You cannot be creative at a high level unless you are robotic at a low level.
The anti-memorization movement has left millions of students unable to think because every elementary operation consumes working memory.
For instance, solving equations feels smooth when basic arithmetic is automatic. It's like moving puzzle pieces around, and you just need to identify how they fit together.
But without automaticity on basic arithmetic, each puzzle piece is a heavy weight. You struggle to move them at all, much less figure out where they're supposed to go.
It’s the same all the way up the ladder of math: from arithmetic, to algebra, to calculus, to all sorts of university-level math and beyond.
Andre Antunes retweeted
Some cold takes…
- I feel OpenAI is too modest in their defense.
- Work of Córdoba and Martínez-Zoroa are fully public on Arxiv. Any agent swarm worth their salt will obviously look at these public results and would eventually explore those directions given enough compute.
- It seems we still live in the world where we think humans are the center of the universe and only they have monopoly on insights.
- Buckmaster et al obtained their headline results just past month. Given these models take >2 months to train, it seems extremely unlikely that OpenAI model might have seen their recent work.
- Biggest takeaway for me is that compute scaling will keep working for at least 4 more OOMs. We have a perennial question when and if this scaling party will end. Seems we are safe (or unsafe in other ways) for 2 more years at least.
- It’s cope when people try to undermine this achievement by saying they just threw huge compute. When humans work on something for 100 years, they are also just throwing more compute.
- Another cope is people saying proof is slop because humans cannot understand it. Mathematics has no obligation for humans to understand it. Our intelligence is bounded and there will be longer time to digest things. Future prompt: “explain to me like I am human field medalist”.
- Solving millennium prize problem is a massive milestone and we should celebrate this without any reservations. The reach of AI (and therefore humanity) has extended from solving 20 years old problems to almost 100 years. This just happened.
- A sober take is that this achievement likely isn’t going to have parallels in physics/bio. Most of the wins here is verifiability and ton of RL for math. Other fields like physics/bio requires interaction with physical world for experimental verification and we might not see same acceleration there.
Andre Antunes retweeted
A censura começou. Você só está lendo este tweet porque já me segue no X.
O TSE criou por resolução - ou seja, e não por lei aprovada pelo Congresso - um mecanismo que, na prática, reduz absurdamente o alcance orgânico dos candidatos nas redes.
No X, perfis declarados à Justiça Eleitoral estão excluídos a partir de hoje das recomendações da aba “Para Você”. Quem já me segue continua vendo meu conteúdo, mas quem ainda não me conhece tem menos chance de receber minhas propostas, opiniões e ideias, a não ser que alguém as reposte.
Eu tenho muitos seguidores e, em tese, posso ser menos prejudicado por essa regra do que candidatos que ainda têm uma audiência menor. Mas liberdade não se defende apenas quando ela nos beneficia. Muito pelo contrário, como diria Orwell: "se a liberdade de expressão significa uma coisa é o direito de dizer às outras pessoas aquilo que elas não querem ouvir".
Não estão tirando - ainda - o conteúdo do ar. Estão limitando sua circulação justamente durante o período em que o eleitor deveria ter mais acesso ao que pensam os candidatos.
E tudo isso, repito, não veio de uma lei debatida e aprovada pelos representantes eleitos pelo povo, mas de uma resolução criada pelo próprio TSE.
Defendo que todos os candidatos, inclusive aqueles de quem discordo profundamente, tenham o direito de apresentar suas ideias e disputar a atenção dos eleitores.
A democracia verdadeira requer o livre mercado de ideias: cada candidato fala, cada eleitor compara, confronta e decide. O Estado não deveria interferir para reduzir a chance de que as ideias dos candidatos cheguem ao eleitor.
Mais uma vez começamos um período eleitoral sob a sombra da censura. É por isso mesmo que reforço: a mudança precisa acontecer no Senado. É ele quem vai enquadrar os togados que agem fora da lei e da Constituição.
Andre Antunes retweeted
Another hangup I'm seeing: why did Jane Street have long momentum exposure? Not because they're yoloing a factor. Because many stat arb strategies organically accrue momentum exposure and residualizing it out would reduce risk-adjusted returns.
A stylized example with no IP since it's in the literature: your signal is xsection(analyst earnings revision). Long equities with upside revision acceleration, short equities with downside revision acceleration. This is _not_ a bet on price continuation/trend.
However! Analysts empirically cluster around news that has already moved the price. The legitimate fundamental information seeding the cross-section has a latent loading on momentum, via human behavior.
If you try to orthogonalize this signal to momentum, you linearly downweight the equities in the cross-section which report the highest analyst conviction, because that conviction is mechanically reflexive on momentum, even though it is a qualitatively distinct source of information!
Consequently, your residualization will flatten the distribution: post-resid top quintile of the xsection has shifted names with weaker signal from the middle three quintiles, and likewise for the bottom quintile.
Realistic numbers - you squeeze a 1.0 Sharpe out of analyst revisions, 8% annualized on 8% vol. You residualize out momentum, your vol does drop because you took out the momentum risk. Maybe it goes to 6%. But oops - because of the distribution smearing effect, your returns have fallen faster, to around 5%, and now you've got like a 0.85 Sharpe.
Jane Street are not idiots. They understand this intimately well, and far better than I'm even describing here. They're still up $40B net revenue YTD after losing $15B because they rationally accept the higher risk adjusted returns that keep in the factor prone to sharp drawdowns.
Andre Antunes retweeted
I think ambitious people should treat life in “seasons” rather than trying to balance everything every day.
A few years massively overallocated to work. A period getting seriously fit. A year or two nomading. Then family takes priority for a while. Then maybe at 35 you go completely insane again and spend four years building a new company, and so on.
The mistake is thinking every metric needs to stay green all the time. Most great outcomes (building a company, writing sth serious, training for an event) need uninterrupted runs, not daily moderation. Sometimes work should suffer cause you’re travelling, sometimes your social life should suffer cause you’re building, sometimes career progression should slow cause family matters more. That’s fine.
And tbh it probably makes life more fun too. You get actual chapters and completely different versions of yourself instead of spending 40 years maintaining the same perfectly optimised routine. A few intense years in one city, a weird nomad phase, an obsession that takes over your life, then something completely different. Probably much more memorable than one very long well-optimised Tuesday.
It's why I encourage everyone to think of balance in years, not days. A good life can look horribly unbalanced on some random Tuesday and still be very balanced over 10 years.
We should all stop with this modern obsession of having a perfectly balanced, optimised life (8 hours sleep, daily cardio + gym, eat well, deep work, etc.)
Most exceptional lives are heavily overallocated somewhere and any pursuit of greatness (in business, art, sport, research) requires big sacrifices.
Online gurus love selling this perfect recipe of life but reality is you only have a 24h day. No way you can sleep 8h, do 10k steps, gym, morning sunlight, read, eat perfectly, do 9h of deep work, see friends, maintain a relationship and somehow be happy through all of it.
I think a better thing for ambitious people is to think of life in seasons: a year obsessed with work, then a period focused on health, family, travel, etc. You can be massively overallocated to one thing for a while without deciding that’s what your entire life should look like. Balance probably makes more sense measured over years than every single day.
Andre Antunes retweeted
One pattern I find useful for working with LLMs is a nice long ramble session. Sometimes the LLM needs more bits to understand what you're trying to achieve, but you're too lazy to type them. In these cases I like to lean back, switch to /voice and just ramble for like 10 minutes, total mess, anything goes, full stream of consciousness. Sometimes I declare it up top, something like "switching to speech recognition sorry for any typos...". Sometimes I turn it into a small interview of a few turns. But I find that the LLMs are somehow very good at reconstructing long incoherent rambles and often their echo of your own tangle of thoughts comes out quite a bit cleaner than what you started with. The result is that you improve the mind meld and have to correct things less from that point on.
Andre Antunes retweeted
You must go all in. 100% maxed out effort. Your absolute best.
Holding back and “kinda trying” is a very pernicious defense mechanism. If you fail you have this built-in ego shield “well I didn’t give my best so I probably would’ve crushed it if I had.”
It’s cowardice masquerading as intelligence.
Fuck fear. Stop taking the results personally. Go all out. Accept the feedback. Adjust. Iterate.
You’ve got what it takes, but it takes everything you’ve got.
Andre Antunes retweeted
I published a note today that I've been thinking about for months.. About how the US stock market has arguably become too big and too imp to fail.. It's basically America's retirement fund now and poss even the savior of social security which is expected to run out of money in less than 10yrs
-Curr 55% of ppl own stocks, by far most in world. And w/ Trump Accounts bringing in 28 million add'l americans into stock ownership the vast majority of ppl (incl Top 1% (who own HALF of stock mkt), middle class and lower income) will have financial interest in the health of stock mkt and they're all voters = the political pressure to keep stocks out of a prolonged bear market is going to be very powerful.
-As such I think there's good chance the Fed will buy equity ETFs in the next major downturn to support market and it will be common practice going fwd. China and Japan already do this. They may even target certain sectors or Capex cos with the purchases.
-This is a massive variable that I feel like is a blind spot among the experts out there and why the bears get run over time and time again altho I think investors are onto it as evidenced by the persistent flows into ETFs during pullbacks as well as a survey of 1000 ppl showing 3/4 of them are confident the Fed will bail out markets in next crisis.
-This is just one byproduct of the 'Nothing Stops This Train' monetary supply explosion and debt extravaganza sweeping the world but esp in US which at this point feels irreversable.. Thoughts? lol
Andre Antunes retweeted
The best predictor of success for tech companies, at every stage from during the YC batch to public company with billions in revenue, is the rate of shipping new stuff.
Andre Antunes retweeted
Every single one of our guides can answer these questions about each of their students:
What activities make you lose track of time?
What would you do if you had a completely free day with no obligations?
What do you enjoy doing outside of Alpha?
What's something you're really proud of accomplishing?
What would you do even if no one paid you?
Who do you admire, and why?
They can answer because they work with kids in small groups and have time for 1:1 meetings with each one. Knowing what motivates a student is how a guide unlocks them when they get stuck in "It's too hard" or "I don't like this."
A traditional teacher can't do this. Lesson planning, grading, admin meetings, 30 kids in a class. There's no time left for it.
Our guides have the time because academics get handled by adaptive apps in 2 hours a day. That's what AI makes possible.
Andre Antunes retweeted
Saw this post on my TL, paired with snide remarks. But it *does* point to a broad and unintuitive feature of research: small improvements lead somewhat to exceedingly large gains.
Here is a classic old post by Matt Might: “The illustrated guide to a Ph.D.”
Keep pushing.
matt.might.net/articles/phd-…
After decades of work and billions of dollars, this is what passes for a revolution in cancer research.
This drug is being hyped to no end on social media. At the oncology conference, there is a standing ovation from a crowd of 40,000 doctors and industry people celebrating it as a major breakthrough and the defining achievements of cancer research over the last decade. The scientists behind it will win a Nobel Prize.
What exactly is the achievement?
A drug that extends life by six months in less than 10% of cancer patients.
The cumulative R&D costs run into the billions of dollars. The drug is not a cure and pancreatic cancer mortality does not change. Resistance develops and more drugs need to be developed. When those drugs are hailed as breakthroughs, resistance develops again. Everyone in the industry knows this approach does not scale as a durable solution.
It is not a cure for cancer.
This work follows a narrow line of thinking based on oncogene theories that has consumed enormous amounts of federal funding and biotech R&D for four decades. At the same time, environmental factors, prevention, detection, the root causes of cancer, tumor evolution, diet, exercise, and many other areas of cancer research were neglected.
Any criticism of this model is immediately met by a coalition of interests invested in preserving it. Academic researchers, pharmaceutical companies, investors, journals, science media, professional societies, consultants, patient advocacy groups who have been told there is no other way, and online activists all have incentives tied to the existing model.
Critics are accused of attacking patients, opposing progress, or undermining science itself. Cancer patients become shields in a bigger issue that is really about the performance of the cancer research enterprise.
After decades of effort and billions of dollars, this is what the cancer establishment is giving itself a standing ovation for.