Stanichor

too wonder-stale to wonder

[Stanichor @stanichor79]

@niplav_site Free will is overrated (and should be eliminated)
https://t.co/DnW7wXzgoh

[Stanichor @stanichor79]

@drethelin there are only two kinds of ethical theories: the ones people complain about and the ones nobody uses

[Stanichor @stanichor79]

@Laneless_ @NathanpmYoung when asked about why they don't check porn, he said pangram doesn't want to host porn bc that comes with a bunch of extra regulations and the like

[Stanichor @stanichor79]

@lu_sichu fit goes hard

[David Manheim @davidmanheim]

Reminder: if you're not donating money to solve problems, it's not effective altruism, it's just sparkling philosophy

[Stanichor @stanichor79]

@lu_sichu true, i guess you have to be chinese and racist

[Stanichor @stanichor79]

@doubtline the 2026 test has to be more g-loaded

[Stanichor @stanichor79]

@zetalyrae https://t.co/aumfIDgsMM

[Stanichor @stanichor79]

@lu_sichu lol

[Stanichor @stanichor79]

@toasterlighting @ApriiSR @1thousandfaces_ when you join anthropic, dario emails you all the alt accounts you've ever used, just to let you know that claude sees all

[Stanichor @stanichor79]

@whybyfire EAs too good at altruism-mogging, how are you supposed to respond to this?

[interstice @an_interstice]

O N E W E E K R E M A I N S

[Stanichor @stanichor79]

@corsaren https://t.co/u0VNDJwDwx

[Stanichor @stanichor79]

@_AashishReddy the act of learning about acausal decision theory decorrelates me from almost all human decision-making, making calculations like these useless

[Stanichor @stanichor79]

@SilverVVulpes very specific, i would’ve just called it tedtalkslop

[Stanichor @stanichor79]

I've also created by own Bayesian ECI (look at Model 4 in the Appendix). I think the biggest difference is the priors on the capabilities and difficulties. If you were trying to match the old ECI (which I was), you'd use Gaussian priors, which pulls the estimates for weak models up (because weak models have less benchmark data, due to being earlier). Epoch's new Bayesian ECI uses Uniform priors, which don't pull up weak models as much.

https://t.co/fo8Vok4U3R

[Stanichor @stanichor79]

But, if you're not happy with your gender, we've got some new options. And I believe one of those options needs to be a public option that will give you a broader range of genders to choose from.

[Stanichor @stanichor79]

(in Obama voice) Let me be clear. If you like your gender, you can keep your gender, period. Nobody is going to force you to leave your gender.

[Stanichor @stanichor79]

@JeffLadish https://t.co/u0VNDJxbm5

[Stanichor @stanichor79]

@_AashishReddy the act of learning about acausal decision theory decorrelates me from almost all human decision-making, making calculations like these useless

[Stanichor @stanichor79]

@an_interstice > [...] the anecdote [...] is a fucking lie
pretty good heuristic tbh