niplav @niplav@schelling.pt

QA sessions and office hours are mostly dependant on the quality of questions

Tryonics

"my timelines have shortened a lot" — beginner brag
"my timelines haven't changed since mid 2020" — intermediate brag
Trying to track & reportreality — advamced brag

"Liberating what we can" - 10% Bodhisattva vow

Fondly remembering about the time when I earnestly tried to use ed for all my editing

I was an odd teenager

<expression of strong dislike for university software>

Just as paramilitaries are much cooler than regular militaries, paraacademics are much cooler than normal academics

Markets=Partition-adaptive Distributed Optimization

"we live in a takeoff world"
"there are no friends at foom"

A thing @Aella_Girl's cheating research made me curious about:

Are cheaters matched together? Or is cheating asymmetric: It's usually only one partner cheats (if so, I'd wager the more attractive one, but there might be interesting game-theoretic considerations).

Pet peeve: The words "World GDP"

Technology that got stuck in in adoption: Public key crypto
Used in the background everywhere, but not as end-user product, e.g. we still sign with our hands?
Perhaps because end-users aren't smart enough to understand public key crypto

Substances for meditation enhancement are a 2×2, {low substance, high substance}×{low meditation, high meditation}.

High substance has problems: Insights don't stick/are vacuous, + dangerous
High meditation has opportunity costs: Learning rate and energy parameter are too low

Low substance and low meditation are just less effective modulo relaxation effects

But where is the pareto frontier?

I am *not* convinced by the evidence for multiplicative decomposition in judgmental forecasting[1]

https://www.lesswrong.com/posts/YjZ8sJmkGJQhNcjHj/the-evidence-for-question-decomposition-is-weak

Arbitrary X aren't Y can mean:

* All X aren't Y
* Almost all X arent Y (real numbers, computable)
* Many X aren't Y, some are, and the two groups are hard to disentangle (SAT problems, sovable in polynomial time)
* Few X aren't Y, but we haven't found a way to distinguish the ones who are from the ones who aren't, and can generate arbitrary X that aren't Y (violations of strategy-freeness in voting theory)
* Almost no X are Y, but same situation as above (don't know example)

Related: fully/partially/barely mapped out positive space and fully/partially/barely mapped out negative space