Overview

The LessWrong Community Census is a venerable and entertaining^[1] site tradition that doubles as a useful way to answer various questions about what the userbase looks like.  This is a request for comments, constructive criticism, careful consideration, and silly questions on the census.

Here's the draft.

I'm posting this request for comments on November 1st. I'm planning to incorporate feedback throughout November, then on December 1st I'll update the census to remove the 'DO NOT TAKE THE CENSUS YET' warning at the top, and make a new post asking people to take the census. I plan to let it run throughout all December, close it in the first few days of January, and then get the public data and analysis out sometime in late January.

I have a little more ambition this year to do some rationality evaluations; read on to hear more.

How Was The Draft Composed?

The sections have evolved over the years.

I copied the question set from 2024 and made some changes. The main changes are-

I removed last year's adjacent community questions, then changed my Indulging My Curiosity question. I then changed the questions that swap every year, mainly the calibration question and a couple of the probability questions.

I removed the Christian sub-denomination options, leaving just Christian. I removed the LLM questions from last year. 

Changes I'm Interested In

Diaspora Questions

I think the Unofficial LessWrong Community Census should, at present, try to cover much of the rationalist diaspora. I'm interested in more than just users of the LessWrong site itself, I'm interested in many of the other places they congregate. 

Towards that end I want to solicit a question from various branches and descendants of LessWrong. I'm particularly interested in the parts that try to teach some kind of rationalist skill for reasons I'll talk about later, but the upside here is my threshold is pretty low for including one question if it's relevant to your offshoot. I want this to be truly a census of the community. Here's the ones I've got on my radar at the moment:

• Center for Applied Rationality
• Wandering Applied Rationality Program
• Bayesian Conspiracy
• Conspiracy of Bayes
• Vibecamp/Post-rationality
• Glowfic
• Forecasting/Prediction Markets
• AI Alignment
• Effective Altruism

Anyone I'm obviously missing from that list?

Remove Deadweight Sections

I'm inclined to axe most of the politics and IQ sections. I like keeping continuity with past questions, but I think IQ in particular was expanded to answer something Scott was curious about many years ago and then it just kind of stuck around.

Politics seems like something someone would be curious about, but that person isn't me, so the extra politics section is all at risk of removal.

Skill Evaluations

Most of all I want the census to double as an annual evaluation for the rationalist project. I don't know about all of you but I'm here to raise the sanity waterline and chew bubblegum, and thanks to the advanced sanity techniques this community taught me I've entirely saturated my bubblegum benchmarks. 

I'm aware of what Goodhart's Law is and I'm sympathetic to the idea that there's more to rationality than can be easily measured on a census question. And also, there is no separate magisterium for rationality skill — it does not make any sense to me to say that someone is an excellent rationalist but they just happen to do badly on every possible attempt to measure rationalist skill. It makes a lot of sense to me that individual skill tests would come apart in the tails; if Eliezer Yudkowsky loses at prediction and calibration tests to Philip Tetlock, that's not surprising to me. I still think if there is a real art of rationality out there, it will involve some components of the art being testable and it will involve "hot dang look at that chart" level outcomes between newcomers and skilled rationalists. 

Does that mean that if you aren't interested in the art of rationality, the census isn't relevant to you? Not at all! Remember, it asks a bunch of general demographics questions that lots of people in the ecosystem can look at, it (hopefully) soon will have questions from adjacent groups. 

Does that mean that there isn't a point for people who aren't interested in that to take the census? Maybe people who are more Community than Craft minded should skip that section? I think they should; this makes for an excellent if odd control group :)

Want to help?

You can open the current draft here. My best compilation of previous versions is in this Google sheet. 

Useful things, by my frame:

• Skim through and make sure I don't have any half-written sentences or really unclear questions, especially on anything new.
• If you're a member of an adjacent community or part of the diaspora, especially if you're in some way 'in charge' even a bit, pitch what question would be meaningful for your branch. For instance, an ESPR instructor suggested "How long have you had the current biggest issue in your life?"
• If you want to take up the torch of the politics section, argue for what's interesting to investigate and what questions we'd need to ask. You're going to have a hard time pitching me on more than ten questions here.
• Any question you're curious about that you think a nice big census is the right way to answer.
• Most usefully, I want evaluation questions. Things I'm likely to try include:
  • Conjunction fallacy
  • Dutch booking loops
  • More calibration and forecasting questions
  • Asch conformity
  • Brainstorming
  • Or if you want, you can argue that this entire endeavor is doomed. Important if true!

1. ^{^}

   Some people find it entertaining because they like arguing about statistics for approximately two thousand and seventeen comments. Other people find it entertaining because I lace it with dumb jokes.