From 2011-2017, there was an annual LessWrong census/survey. Much like a national census, this provided a valuable lens into the demographics and beliefs of LessWrongers. Unfortunately, this tradition appears to have stopped in recent years, with the exception of a mini-revival in 2020. (Scott Alexander appears to have moved the census to SlateStarCodex.)

From what I've read, this is mainly due to of a lack of will/time among those in the community to run this project, and not a general judgement against the census. 

If this is the case, I'd like to start a new version of the census this year, with a greater emphasis on alignment research/beliefs about AI and timelines. 

Is this a good idea?

New Comment
13 comments, sorted by Click to highlight new comments since: Today at 5:41 PM

I also think this is a good idea, and recently I worked a little on one myself! It has about 200 questions so far on lots of topics. I was thinking I'd share a final draft of it on my shortform for feedback before shipping it, but now that you've written this, I've PM'd you a link to a google doc where I'm putting it together.

I do a lot of surveys and have been thinking about doing a rationalist census too, including sending a message to Ruby about it a while ago. Can I be involved in this project too?

just sent you a link to my current doc!

This sounds great! Hopefully there'll be a version of the census that doesn't take too much time to answer? I've seen so many surveys that purportedly require 30+ min (and often that's already a ridiculous underestimate), and that's just an unreasonable ask.

I don't know what the common solution here is - maybe most questions could be optional, or maybe there would be a staggered version where you have to answer the first X questions, and could then either stop and submit or answer an additional Y questions, rinse and repeat. In any case, it's psychologically much easier to answer a long survey if you don't have to commit in advance to spending 30 minutes, and can instead decide every ~5 min whether to continue.

In addition, long survey forms really need a way to save progress so you don't have to answer all questions in one go, and so a browser crash or refresh doesn't mess up all your work.

I agree. I am currently intending to go with the "Here is the MVP section up-front, the remaining 80% is all bonus" solution.

One approach could be splitting the census into two (or more) parts. The "lite" section would include high-value 2017 census questions, to see how the LessWrong community has evolved over time, and would be reasonably short. 

The "extended" section (possibly split into "demographics", "values/morality", and "AI") could contain more subject-specific and detailed questions and would be for people who are willing to put in the time and effort.

One downside of this approach would be that the sample size for the extended section could be too low, however.

That is a fantastic idea! Please use one of the old surveys as a template. If you have enough time to work on it, you may want to source additional questions here.

I agree, it's useful to be able to compare how the views on central questions evolve and that's not possible if the way the questions get asked changes too much.

Yeah, sure. I think in the later years it got a little too long, but I'd take a census. Might want to plan it with the site mods.

If you would like to create a survey with increased emphasis on AGI stuff, consider that the site is getting swamped with AI content, and that not everyone cares about it. So at the very least, make that section almost or entirely optional.

I want to start an annual Cryonics Census asking a few questions about what people think of cryonics. If we could coordinate to release it at the same time, and you wouldn't mind linking to the Cryonics Census from the LessWrong Census, that'd be great. I could also include a question asking whether the survey-filler consents to us associating their answers from both surveys.

It might also be interesting if someone were to set up a prediction market for the results of the census. I'm not really sure how to do that, otherwise I'd do it myself.  You probably need some idea of what the census will be about?

Once the census is up you could create manifold markets.

New to LessWrong?