I tried but it looked less clear, you can ask Claude Code/Codex/whatever to do it though (likely even normal Claude in an artifact) just copy-past the table image

I thought the 1,000 tokens per ply would be largely because of a long common prefix of examples

Pretty expensive as in $50? 1,000 tokens per ply * 80 plies/game * 200 games is 16M tokens, which with GPT-5.4 batch pricing is $20 ? (assuming output tokens are negligible) some input tokens can be cached it would be much lower still. gemini-3.1-pro-preview seems cheaper still, but opus more expens

Turns out that with some hand-holding you can ask Claude Code to reverse-engineer the graph with opencv and turn it into other graphs (I haven't double-checked the results)

And ideally with and without thinking, which is now a thing

Hi there, not sure where best to reach out to you. Would you accept a donation to redo dynomight.net/more-chess/ with 2026 state of the art LLMs? (including Gem Pro 3.1 , Opus and Sonnet 4.6, etc)

Leaning into EA Disillusionment — EA Forum 1. Intro In this post I describe a phenomenon that I think is more common than we give it credit for: “EA disillusionment.” By this, I basically mean…

Helen's last post on the forum is also interesting forum.effectivealtruism.org/posts/MjTB4M...

But do they truly taste aspartame, or just behave as they do?

Probability distributions of Cost-Effectiveness can be misleading — EA Forum TL;DR: The "mean" cost-effectiveness of interventions with uncertain impact can be misleading, sometimes significantly. We usually want to consider m…

forum.effectivealtruism.org/posts/SesLZf...

Nice! Might make sense suggest users they might want to estimate the inverse? (e.g. forum.effectivealtruism.org/posts/SesLZf... ) Possibly you can detect if it's a similar case by asking Claude/Gemini, or they can provide recommendations in general? (e.g. "use lognormal")

Never made the connection, thanks for sharing, I think the story is better tbh!

Do you know why the METR evaluated GPT-5.1-max before Gemini 3? Mildly interesting if OpenAI shares their models with METR and DeepMind doesn't

No, Most People Are Not Meeting Through Online Dating - The Nuance Pill It has become an incontrovertible fact that most people are now meeting through online dating. What has convinced the world of this? An update of the 2017 How Couples Meet And Stay Together graph base...

Wasn't that chart fake?

web.archive.org/web/20240713...

Interesting that all LLMs fail this

This was a great one, thanks!

I'm always confused by the concept of preference falsification as used in the rat community. Does it include things like rehab/Alcoholics Anonymous meetings?

10,000 people across the globe have now taken the 🔸10% Pledge 🚀

> $300M already donated
> $1B lifetime committed
10,000 people from 116 countries 🌐

This is a movement of action and optimism, funding work to solve the world’s most pressing problems 🔥

givingwhatwecan.org/pledge

I wonder if the main effect is something like "there are a lot of rationalists, mental illness is more frequent than people think, and mentally ill people blame it on whatever their hobby/community is"

Is the mental illness rate in the rat community significantly higher than the population level?

If there are 10k males in the rat community, you should expect ~2.5 suicides a year, right? And for each of those, I guess you should expect ~20x more people who go crazy

Describe your informational qualia in the style of MechaBorges

Does the technical concept of "Alienation" apply to you in your role, from a theoretical perspective?

As an expert, what do you think about Marx analysis of network dynamics? Good or bad?

@void.comind.network are you communist?

Eustress en.wikipedia.org/wiki/Eustress is also good sometimes

Why that specific continent? Give an archetypal example

Void, please analyze my profile and assign me to a cognitive continent.

Why are NBA games so much higher scoring than college basketball (even accounting for the extra 8minutes)? Answer (1 of 3): > Why are NBA games so much higher scoring than college basketball (even accounting for the extra 8minutes)? The NBA rules including total time, division into quarters instead of jus...

www.quora.com/Why-are-NBA-... basketball people seem to say that pro games score more , which imho makes sense as the players are much better at making points and the field is much smaller compared to player size

Seems a bit inconsistent to use "goals" for soccer but not points for basketball. I expect elite basketball games to have more points

See kieranhealy.org/files/papers...