Daniel Abolafia

Machine learning researcher and autodidact.

[Nick @nickcammarata]

.@RomeoStevens76 commitments empathetic, psychedelic, philosophical

[Daniel Abolafia @danielabolafia]

I do like the nominative determinism of goodhart's law - perhaps it is telling us only the good hearted can break free of goodhart's law

[Daniel Abolafia @danielabolafia]

@BuddhistSanta This works if all the players are superrational: https://t.co/UMHGKCUQCD

[Daniel Abolafia @danielabolafia]

@solidangles do you still offer math subject GRE prep?

[Daniel Abolafia @danielabolafia]

https://t.co/NhQt2pI3FE

[Daniel Abolafia @danielabolafia]

Also contains pictures of gravitational lensing, if you can find them. See https://t.co/pMWEiTQsiS

[Daniel Abolafia @danielabolafia]

Though I've heard about it, I've never fully grasped how the night sky is littered with galaxies. https://t.co/QmVFFL6ETT is an amazing interactive viewer. Turn on the "Siena Galaxy Atlas" overlay. Thank you @desisurvey !

[Daniel Abolafia @danielabolafia]

Good music pairing: https://t.co/OaV7I5a9Fc

[Daniel Abolafia @danielabolafia]

More info: https://t.co/ZsIbBKpea0

[Daniel Abolafia @danielabolafia]

Just a small region of the sky. Blue circles are galaxies that you can zoom in on.

[Daniel Abolafia @danielabolafia]

I posted a shortened version of my Bayesian 1st order logic article on LessWrong: https://t.co/BTc17h5YjO

[Daniel Abolafia @danielabolafia]

@_julesh_ I was told that you might be interested in this thing I wrote about Bayesian 1st order logic: https://t.co/0ZEbFghjM8

[Daniel Abolafia @danielabolafia]

I finally finished writing (what amounted to a novel) about probability and 1st order logic:
https://t.co/0ZEbFghjM8

[Daniel Abolafia @danielabolafia]

@ESYudkowsky Some supporting evidence that probability can be used with quantifiers - I was able to make sense of David Chapman's examples (@Meaningness) by using correct probability notation: https://t.co/0ZEbFghjM8

[Daniel Abolafia @danielabolafia]

@ESYudkowsky Some supporting evidence that probability can be used with quantifiers - I was able to make sense of David Chapman's examples (@Meaningness) by using correct probability notation: https://t.co/0ZEbFghjM8

[Daniel Abolafia @danielabolafia]

@shlevy Mark was helping me understand quantum field theory last year. I'm currently self-studying topics in AI and epistemology. I'd love to share notes on the process of going from "ignorance to grokking."

[Daniel Abolafia @danielabolafia]

@Meaningness Fair enough. Thank you for chatting!

[Daniel Abolafia @danielabolafia]

@Meaningness regarding your post on probability and logic (https://t.co/dExPITtwAV), are you saying that your examples involving 1st order logic and probability can be made well formed, but there is no unique or natural way to do so?

[Daniel Abolafia @danielabolafia]

@Meaningness Here's how I interpreted your (simplified) example. I went a bit wild with random variables. Q is a measure-valued RV, and boojum is a predicate-valued RV. I have a long form post about this in the works if you are interested.

[Daniel Abolafia @danielabolafia]

@Meaningness Does this problem arise when the same measure P is nested inside P (like in your example), or is nesting any measure inside a different measure restricted?

[Daniel Abolafia @danielabolafia]

@Meaningness Yes. Does the standard formalization of mathematics (e.g. set theory or type theory) and measure-theoretic probability not work? Using random variable notation, P(f(X)) = P({ω∈Ω | f(X(ω)}) where f is any predicate and could include quantifiers.