My suggestion, as usual, is a Basilisk question.

And in light of recent events, it might be interesting to factor the cryonics survey question into two: whether cryonics works in principle, and whether cryonics as currently practiced by ALCOR/CI will work.

The high rate of depression among LW readers is significant and, well, depressing. We should dig deeper and ask questions such as "Were you frequently depressed as a child?", "Has either of your parents suffered from chronic depression?", "Is your depression probably caused by loneliness?" and ask the open ended questions "What if any depression treatments have you tried that have helped you?" and "What if any depression treatments have you tried that have not ended up helping you?"

How about asking "What is the single most important change that would make you want to participate more frequently on Less Wrong?"

This question would probably not be useful for the census itself, but it seems like a great opportunity to brainstorm..

I think the questions of the next survey should be a superset of those on the last survey. Maybe not strictly, but it's too interesting to track year-on-year changes to remove questions unless it's really unquestionably obvious that they're superfluous.

I have some slightly worried suspicions about a question for several reasons. That question is:
"How many close friends do you have?"

Another question I'd be interested in seeing:
"How often in the last year have you exercised the Virtue of Silence?"

In addition to the standard census questions I think it would be good to have an optional page two that asks for agreement with various statements about rationality. I think it would be very interesting to see how many different dimensions there are in our disaggrements.

One of the statements might be: It's useful to identify oneself as an aspiring rationalist.

Some people in our community like labels like that while other find them annoying.

If there are many questions, put the most important ones on the first page, and make everything else optional.

Any question that requires more than 2 seconds to answer should be optional. (That obviously includes any question that requires clicking on a link to an external web page with a long explanation of some definition.)

I would propose to get rid of the question about the birth month. It doesn't provide useful data and it reduces the annonymity of the data relatively strongly because months of birth are often relatively easy to look up.

1.60% in America, 1.06% in the LessWrong survey population. There are at least 2 ''out'' LW contributors with schizotypal personality disorder. I don't know how many LWers there are in order to contrast this with the 3.9percent US prevalence.

There is frequent comment on the high prevalence of autism in our community. But there is less report of the opposite trend: It appears that LWers have quite a bit lower schizotypy than the general populations.

I would be interested in teasing this out more. I'd encourage deeper questioning on schizotypy. I'm not a psychiatric researcher or clinician so I would encourage someone else with greater experience to chime in on the next steps.

My intuitions are that questions about personality disorders such as schizotypal, avoidant and paranoid personality disorders, as well as various classes of anxiety disorder such as generalised anxiety disorder would be a good idea to look at claims that we're a rather worried, AI-paranoid, ill thinking bunch in context!

Also in light of recent events (big breakthroughs in accuracy of CRISPR editing, stem cell<->gametes<->embryo creation, & complex trait polygenic scores), I'd be interested in questions about embryo selection, iterated embryo selection (IES), and embryo editing. This would be similar to the other belief questions, I guess - maybe median date of first baby born with any CRISPR edits, median date of first IES procedure? (I don't see any point in asking for probabilities for CRISPR, at least; everyone knows it's a matter of when, not if.)

Also thanks for initiating this ELO; upvote for proactivity :)

I think there should be a "do not identify with a race" option under race like the "do not identify with a gender." Seems strictly parallel, and if anything, the "real" answer is less clear than the having or not of parts. I bet that while we can all successfully tic a box on a DMV form, I'm not the only person who really doesn't identify with a race, and I think it's different enough from "other." Just my $.02.

I'd suggest that the questions on sex, gender, and orientation be re-factored. One schema I've seen that may be worth starting with runs:

• Gender role: masculine, feminine, non-binary
• Gender identity: man, woman, non-binary, agender
• Romantic orientation: heteroromantic, homoromantic, biromantic, aromantic...
• Sexual orientation: heterosexual, homosexual, bisexual, asexual...
• Sexual behaviour: heterosexual, homosexual, bisexual, celibate...
• Sex: male, female, intersex