@zeroxkyle I’m traveling around China right now and the amount of people watching AI generated tiktoks on scroll blew me away. While driving my Uber (DiDi), at tourist attractions, EVERYWHERE
Also their day-to-day apps are riddled with gambling like pop ups. I hope this is not our future
@liquiditygoblin@idoccor Why are regime switching models a dead end? Is it only for HFT you mean? Surely one can imagine scenarios where informative signals get crowded out because the usual signals no longer carry information in the different regime?
But I dunno
@AgustinLebron3 Do you agree with Leopold’s situational awareness?
And more specifically, do you foresee a mass exodus of European (and perhaps Chinese) AI talent from SF given recent administration decisions, and therefore hope for EU, or has that run its course? is The Window Closed?
@AgustinLebron3 What are you planning to spend your time on post-scarcity, and are you generally optimistic or pessimistic about the post AGI future? :)
@liquiditygoblin@0xDmtri > you’re on a stock trading desk & buy 100 shares of MSFT OTC with a bank, crossing at $420. both you & the bank think you made a good trade.
> why?
> Bc bank thinks it will go down and u think it will go up, no?
> bingo
Lmao
@whitesocksclips The evidence for election fraud in 2020 or devil worship, which both Nick and I believe, is built on similar “something’s off” scattered evidence rather than hard “the Jews did it” evidence, because you never get hard evidence when people are actively trying to cover it up
@turbofish_pk@systematicls@therobotjames I can’t believe this 1 like comment is the one that makes me realize someone else already came up with “my” seemingly ingenious & novel idea (crowdsourced DeFi hedge fund), and implemented it successfully 😭
This is why I follow @systematicls though ahaha
@systematicls 2/2 – which is why I try to relegate it to very specific refactors (which no other LLM can do as well, especially multi-turn)
Gemini 2.5 Pro’s (via aistudio) reasoning has been exceptional though so sometimes I just let that plan a refactor that Sonnet then implements. 3 Pro idk
@systematicls 1/2 In my experience, Sonnet with code2prompt is also the most reliable, but there was a brief period where Grok 3 Thinking wrote really clean, token-efficient, and performant code (also for HFT). If left to itself, Sonnet will tend to keep making logic more and more verbose –
@ryneboi@realspeckzo Vivek's claim to fame is name-dropping The Great Replacement and calling out Nikki Haley in the primaries
I think he figured out that all momentum in online RW politics come from making groypers (or similar) feel heard, which JD Vance also tries to. Too bad they're both indian
@asukagrypr Literally 2 days ago. And now he’s the one preaching to “do nothing”.
How can he simultaneously believe voting doesn’t matter and the system is forever rigged yet sit here with the most live viewers EVER and preach to keep working within the system
What a sick joke. RW = cooked