“Welcome to 2026, or 1926. Either way, most will find out that even humans hallucinate. And worse than the machines they bet on”
@om is such a wonderful writer.
Clueless nature of @business their utter lack of history of @awscloud and how it came about and how it scaled is shameless, given they have @BradStone who damn wrote a book of it.
Kudos @edzitron you say scary things. I tru and say the same things my way, because I am well, OG and have learned to be cautious vs caustic at times. But I can stop reading you. 👍🏽👍🏽
It’s not about the inevitability about the internet, or commerce or ai. The madness is about the math that does not make sense. Ai is inevitable. So is math madness.
***
Here are my recent pieces about AI
https://t.co/zPz6JAwa14
https://t.co/cCNpjvvNAv
https://t.co/wOH3T93uGF
https://t.co/APTMhwzkkK
https://t.co/Lt6PtErvJi
@tjack@TownAI@firstround@jgreze@tonydevincenzi very cool. I think this sets the bar for how a VC firm should communicate the value prop for a portfolio co. super interested to check it out.
my magnum opus: the tiny part inside every robot and drone that america can't make anymore — and the two startups betting opposite ways on how to fix it
Richard Feynman's Restaurant Riddle
Feynman once wondered how many dishes at restaurants one should try before settling on the tastiest (and maximizing the overall quality of one's meals). New mathematical result offers a solution
https://t.co/wJsPoEG1xJ
Once you internalize the auto-research concept and use /goal with it, you unlock ridiculously fast app improvement on anything that has a numeric rubric.
We’re all still too rooted in the “labor is expensive” world.
@SawyerMerritt If initial interest is weak, Ferrari will put Luce buyers on a VIP list, which guarantees allocations to special cars for some time.
Porsche did this with the 918 and buyers then got allocations for 10 years: 911 R, 911 GT3, Speedster, SC, Dakar, S/T.
Amateur traders armed with AI trading algos are going to be like the technical analysis / chart traders.
A false sense of precision and a great source of alpha.
Cliff knows what he's talking about.
ML is a god at reward hacking. You have to use it with judgement.
Cliff Asness (the co-founder of AQR, who manages billions) on the one belief about that most of his peers would disagree with:
"as we move into a world where machine learning is becoming a bigger part of the quant toolkit, we are big believers that economic intuition as a restraint on overfitting is still pretty important."
"finance is still a high noise to signal ratio place. in that world, overfitting is a problem."
"but one thing ML has done is point out that underfitting is also a problem. and you have to balance the two."
"not using these techniques was probably underfitting and missing some of the true complication."
this is the most balanced take on ML in trading i've heard. it's both problems exist simultaneously and the skill is balancing them.
but for the regular retail trader, I would still say stay far away from machine learning, you will likely just overfit.
@stevemagness Well said! The trickle down effect on youth sports is for me the primary reason we can’t normalize it. Youth sports are already crazy, but to make PEDs the norm is child abuse.
ps this is why Greg Lemond is a hero.
Math and finance nerds that love prediction markets should love perps even more.
Prediction markets : perps :: n! : Γ(z)
Perps are simply the analytic continuation of event markets: from discrete terminal bets to continuous mark-to-market exposure.