I have a severe psychometrics addiction, so I've regularly been tempted to create some sort of "rationalist type test" or similar, which rationalists could take to classify themselves compared to other rationalists. I particularly like to create such tests using data-driven methods, such as factor analysis. But factor analysis requires a bunch of primitive variables to summarize. And I thought, the most logical way to get such variables is to ask people what they've already come up with.

So, do you have any thoughts on ways that rationalists differ from each other, or ways that rationalists differ from nonrationalists, or perhaps even ways that nonrationalists differ from each other in ways that might be interesting to put into a test for rationalists?

Also, if you don't have a LessWrong account, or want to share this question with rationalists who don't hang out on LessWrong, I've created a Google Form that you can fill out instead. If you've got something spicier that you would prefer doesn't become public, then I have also created a private version of the form here.

I don't know how useful my analysis could become, but at least in theory it seems like it could become more generally useful. For instance, it might be useful for providing targeted advice to certain subgroups of rationalists.

Edited to add: A lot of the comments seem to be bringing up abstract, general traits that have been studied in differential psychology. Certainly those could also be interesting to investigate rationalist on, but I have numerous philosophical objections to many of the traits, and I would instead like to design a new test with traits specific to the rationalist community, in terms of more concrete things that rationalists do and experience.

New Answer
New Comment

8 Answers sorted by



I would not try to reinvent the wheel and choose a subset of the Big Five sun facets. https://en.m.wikipedia.org/wiki/Facet_(psychology)

One dimension that might be missing in Big Five but relevant for rationality and is present in HEXACO is honesty/humility with it's sub facets; https://en.m.wikipedia.org/wiki/Honesty-humility_factor_of_the_HEXACO_model_of_personality

Independently, it would be interesting to see how these correlate with

  • choice in prisoners dilemma
  • onebox twobox
  • dualism
  • ...

I'm not sure why not to reinvent the wheel. Personality psychologists do so all the time; the facet-level structure of personality is not particularly well-settled (and existing measures seem unlikely to be very relevant to rationalist culture).

I'm fine to (re)invent parts that make the test better or add valuable parts. But why not reuse what is already good?
Well so if we take a closer look at the facets, we see that often they are measured by questions that often ask about slight variants of the same behaviors. Furthermore, the facets in most existing personality models were theoretically posited rather than empirically derived (surely 6 facets per factor is a biiit to convenient? and indeed more direct work finds a totally different facet-level structure). Furthermore, I have a more philosophical issue with the way personality scales are usually constructed, in that they focused excessively on very abstract behaviors, which can make them functionally opaque, and also makes them "not really tests" (the test is not performing much cognitive work; most of the evaluation comes from e.g. yourself). These are very general problems which mostly apply to personality tests in general. And in fact I've been tempted to do a bottom-up replacement of existing personality models using my own methods that fix all these problems (that's part of the "psychometrics addiction" I mentioned in the OP). But overall since I have so many problems with existing personality models, this also makes me have those problems with using them for this. Less general/more specific to rationalism is that I think many of the nuances among rationalists would get lost. For instance, I'm not sure which classical personality facets are the most related to worrying about wild animal suffering, but I strongly suspect that none of them are very strongly related to worries about wild animal suffering.
Thanks for linking to the SPI. I wasn't aware that there is now better psychometry than Big Five. I'm delighted! I also didn't know that   and more worrying The topics in IPIP seem to span a lot of topics. You write: I found 10 items related to animals in the IPIP. Some are about suffering animals, though not explicitly "wild" animals. No animal question is included in the 135 selected items for SPI. I am a bit confused about this as there is some claim that the selected items cover all topics. I retract my suggestion to use Big Five items and replace it with the suggestion to use suitable items from SPI plus missing ones from IPIP, e.g., an animal-related question.
Right, so I'm thinking I would write novel items specifically for rationalists. SPI was basically constructed by taking 696 items from common preexisting personality inventories, and submitting them to factor analysis. This means that it is comprehensive with respect to those inventories, and also that it improves upon them, as there are many places where SPI's factor analysis gave different results than the structure that had been imposed on the prior personality inventories. However, it is not comprehensive with respect to IPIP, as IPIP has 3320 items and not just 696 items, nor is it comprehensive with respect to personality more generally, as there are places where IPIP is missing items.



Iconoclastic, distrusting of mainstream institutions, seeing oneself as an outsider (vs. identifying more with the people inside institutions who are trying to make institutions work decently well)

Scrupulosity, especially about honesty/integrity/commitment/authenticity (e.g. when you say you'll do something that is an ironclad promise) (e.g. feeling uncomfortable w the job interview process where you know what you're supposed to say to improve your chances)

Demandingness of rigor vs. willingness to seek value from a broad range of lower quality sources (e.g. the tension in this post)

Being quick to dismiss of standard ways of doing things if other ways look better to you vs. treating the standard ways of doing things as a pretty good default (e.g., styles of social interaction) (e.g., lifestyle things such as eating soylent/huel)

Being in touch with your own emotions, living in your body

Avoidance of "woo" / things related to spirituality

Trying to form your own independent impressions, always trying to build your own gears-level models vs. trying to form all-things-considered beliefs, foxy aggregation of multiple viewpoints, inclined to favor variants of outside viewish reasoning

How much you identify with heroes vs. sidekicks vs. villains/antiheroes

How much you identify with various fictional characters (e.g. HJPEV, HPMOR!Hermione)

How easily/thoroughly you get nerd-sniped & delve deeply into random topics or puzzles

Head-in-the-clouds absentminded professorness vs. tracking the concrete practicalities around you

Thank you, this is exactly the sorts of things I was looking for! Especially the concrete examples like soylent or job interviews or the links to posts with examples (because the factor analysis itself is capable of abstracting things given data on lots of concrete variables, so I'm especially interested in the concrete variables as this also e.g. tests that we are anecdotally abstracting the concrete variables "correctly").

Mateusz Bagiński


I remember having written some thoughts exactly on that but can't find it now, so some mostly obvious stuff from the top of my head:

  • High IQ
  • High Openness (Big5/HEXACO trait)
  • High need for cognition
  • Very little smalltalk
  • Being uncomfortable with compartmentalization, both when it comes to beliefs (different parts of map should match each other) and beliefs-actions (if I think X is good for me, I should do more of X)
  • Ability to discuss controversial topics without getting annoyed with people whose views differ from ours

I guess one thing I should say is that traits like IQ or Openness are very high-level abstract general traits. They're the sort of thing that you tend to get out of a factor analysis, whereas I'm more looking for the sorts of narrower traits that you put into a factor analysis.



Here are some attributes I've noticed among people who self-identify as rationalists. They are:

  • Overwhelmingly white and male. In the in-person or videoconference meetups I've attended, I don't think I've met more than a couple non-white people, and perhaps 10% were non-male.
  • Skew extremely young. I would estimate the median age is somewhere in the early to mid 20s. I don't think I've ever met a rat over the age of 50. I'm not saying that they don't exist, but they seem extremely underrepresented relative to the general population. 
  • Overweight the impact / power of rationalism, despite having life outcomes that are basically average for people with similar socioeconomic backgrounds and demographics
  • Tend to be more willing than average to admit that they're wrong if pressed on a factual issue, but have extreme confidence in subjective beliefs (e.g., values, philosophy, etc). This might just be a side effect of the age issue, since I think this describes most people in this age group. Or perhaps the overconfidence in subjective beliefs is just normal, but seems high relative to the willingness to switch beliefs on more factual matters.
  • Have a very high "writing and talking / doing" ratio. I think this is a selection bias kind of issue: people who are actually out doing stuff in the world probably don't have a lot of time to engage in a community that strongly values multi-page essays with a half-dozen subheadings. Although perhaps this is also just another side effect of the age skew.
  • Undervalue domain knowledge relative to first-principles thinking. As just one example, many rats will gladly outline what they believe are likely Ukraine / Russia outcomes despite not having any particular expertise in politics, international relations, or military strategy. Again, perhaps this is normal relative to the general population and it just seems unusual given rat values. 

Overwhelmingly white and male. In the in-person or videoconference meetups I've attended, I don't think I've met more than a couple non-white people, and perhaps 10% were non-male.

Demographic variables aren't so suited for a factor analysis, in a sense because they are causally upstream of the factors of interest. It might be interesting to take some of the outcomes from those demographic variables, though; for instance probably much of what makes rationalists so male is that rationalism selects for abilities/interests related to programming, which is i... (read more)

This is just pushing the question one step back though, I don't know of any good theories for why software engineering is heavily biased towards males either.
The point is just that factor analysis assumes that the items/variables end up correlating due to the factors. If you put variables that are upstream of the factors, such as sex, into the factor analysis, then those upstream variables would have no reason to correlate with each other in ways that match the factor structure (and in fact due to collider bias, would in this case have reasons to end up correlated in ways that precisely oppose the factor structure), so therefore it would be nicer to avoid demographic variables as much as possible.



If you want to look into this in more detail the data from the censuses we have is good. It tells you how the intensity of LessWrong and other rationalist engagement correlates with all sorts of different factors. 



I see the rationalist community as a center of a large onion. There are multiple filters one must pass in order to get there.

On the outside, there is the filter of intelligence and abstract thinking in general. It eliminates the people incapable or unwilling to think about complex matters.

Next is a filter of scientific/realistic orientation. It eliminates the people who decided to use their intelligence to study horoscopes or conspiracy theories.

Near the center is the filter of humility/nonattachment. It eliminates people who are unwilling to change their mind, or unable to stop fighting for their political tribe.

(This is a simplified model, and I forgot a few things, such as caring about the future and other people. If we add those missing things, the sequence of the filters will probably not remain linear.)

What I wanted to say here is that the answer to "who are rationalists" depends on the specific filter. Who are rationalists, as opposed to the general public? Who are rationalists, as opposed to university-educated people? Who are rationalists as opposed to skeptics or atheists in general? Depending on which question you choose, the answers will be quite different, and will sound wrong to a person focused on a different question.

Well so my goal is more to characterize differences between different kinds of rationalists than to characterize differences between rationalists and non-rationalists/answer who the rationalists are. 😅

Ah, ok. Anyway. I think your sample will be contaminated by people who are not rationalists in spirit, but for some reason they like to hang out with rationalists, and may identify as such. So, those would be one subgroup. It seems like half of people at LW meetups are STEM oriented, and another half is not. That might be interesting. Generally, if any trait is stereotypically overrepresented among rationalists, it may be interesting to study those who do not have this trait. I feel like there might be an important difference between people who are trying to change themselves, and people who are just studying the outside world without trying to change themselves. Self-help vs science, kind of.



I would love to see proper data on this.  In particular, including the facets and not just broad buckets.  Or if possible, even including findings for specific items.

The ones I've met at a meetup seemed (compared to the broader population):

-Very high in Interest in ideas, which was by far the most noticeable trend.
-Somewhat neurotic

Agreeableness was mixed.  Some were unfailingly agreeable, and some were starkly low in agreeableness.  Maybe data would show a clear trend on facets or items.  For the more strongly utilitarian ones, as a group, I'd speculate they are lower in Honesty-Humility from HEXACO.  Yet none ever seemed to make me "worry" in that way, as if they couldn't even manage to have Dark Triad traits without being helpful.

3 comments, sorted by Click to highlight new comments since:

One thing that annoys me with "normal" people is their inability to easily talk about the meta level of a particular topic. I feel like if I start talking about something meta some people get internally confused a bit, and instead of asking for clarification they will interpret some parts of what I said at the object level, discard the rest, and continue the conversation as if nothing happened.

Sure, you can talk about meta topics with most people with enough effort, you can try carefully prompting them (like "so what I am going to say may sound strange, I am not talking about X things by themselves, but I am saying something about all Xs in general, disregarding this particular example of X we have just been talking about, does that make sense?"), or if they have previously been exposed to meta-level discussion on this topic that also makes things much easier.

I feel like most rationalists can jump between object and meta level with ease, and I particularly enjoy conversing with people who can do this.

Using the meta level or being comfortable with it sounds like a good idea. Somewhat related: Ability and willingness to reflect and introspect.

🤔 I think it's a good one. I do however wonder how much it is just the g factor of general cognitive ability, as well as how well people can self-evaluate it.

I think it would be nice to have informant-reports for these sorts of things, where one gets evaluated by some other rationalist one has had discussions with. However I don't know if I can convince random people on LessWrong to collect such informant-reports.