Of Gender and Rationality

Eliezer Yudkowsky16 Apr 2009 0:56 UTC
60 points
360 comments5 min readLW linkArchive

Among all self-identified “rationalist” communities that I know of, and Less Wrong in particular, there is an obvious gender imbalance—a male/​female ratio tilted strongly toward males.

Yet surely epistemic and instrumental rationality have no gender signature. There is no such thing as masculine probability theory or feminine decision theory.

There could be some entirely innocuous explanation for this imbalance. Perhaps, by sheer historical contingency, aspiring rationalists are recruited primarily from the atheist/​libertarian/​technophile cluster, which has a gender imbalance for its own reasons—having nothing to do with rationality or rationalists; and this is the entire explanation.

Uh huh. Sure.

And then there are the less innocuous explanations—those that point an accusing finger at the rationalist community, or at womankind.

If possible, let’s try not to make things worse in the course of having this discussion. Remember that to name two parts of a community is to split that community—see the Robbers Cave experiment: Two labels → two groups. Let us try not to make some of our fellow rationalists feel singled-out as objects of scrutiny, here. But in the long run especially, it is not a good thing if half the potential audience is being actively filtered out; whatever the cause, the effect is noticeable, and we can’t afford to ignore the question.

These are the major possibilities that I see:

(1) While the pure math of the right Way has no gender signatures on it, we can imagine that men and women are annoyed to different degrees by different mistakes. Suppose that Less Wrong is too disagreeable—that relative to the ideal, just-right, perfectly-rational amount of disagreement, we have a little more disagreement than that. You can imagine that to the men, this seems normal, forgivable, takeable in-stride—wrong, perhaps, but not really all that annoying. And you can imagine that conversely, the female-dominated mirror-image of Less Wrong would involve too much agreement relative to the ideal—lots of comments agreeing with each other—and that while this would seem normal, forgivable, takeable-in-stride to the female majority, it would drive the men up the wall, and some of them would leave, and the rest would be gritting their teeth. (This example plays to gender stereotypes, but that’s because I’m speculating blindly; my brain only knows half the story and has to guess at the other half. Less obvious hypotheses are also welcome.) In a case like this, you begin by checking with trusted female rationalists to see if they think you’re doing anything characteristically male, irrational, and annoying.

(2) The above points a finger at the rationalist community, and in particular its men, as making a mistake that drives away rational women. The complementary explanation would say: “No, we have exactly the rational amount of argument as it stands, or even too little. Male newcomers are fine with this, but female newcomers feel that there’s too much conflict and disagreement and they leave.” The true Way has no gender signature, but you can have a mistake that is characteristic of one sex but not the other, or a mistake that has been culturally inculcated in one gender but not the other. In this case we try to survey female newcomers to see what aspects seem like turn-offs (whether normatively rational or not), and then fix it (if not normatively rational) or try to soften the impact somehow (if normatively rational). (Ultimately, though, rationality is tough for everyone—there are parts that are hard for anyone to swallow, and you just have to make it as easy as you can.)

(3) It could be some indefinable difference of style—”indefinable” meaning that we can’t pin it down tightly enough to duplicate—whereby male writers tend to attract male recruits and female writers attract female recruits. On this hypothesis, male writers end up with mostly male readers for much the same reason that Japanese writers end up with mostly Japanese readers. In this case I would suggest to potential female authors that they should write more, including new introductions and similar recruiting material. We could try for a mix of authorial genders in the material first encountered on-site. (By the same logic that if we wanted more Japanese rationalists we might encourage potential writers who happened to be Japanese.)

(4) We could be looking at a direct gender difference—where I parenthetically note that (by convention in such discussions) “gender” refers to a culture’s concept of what it means to be a man or woman, while “sex” refers to actual distinctions of XX versus XY chromosomes. For example, consider this inspirational poster from a 1970s childrens’ book. “Boys are pilots… girls are stewardesses… boys are doctors… girls are nurses.” “Modern” cultures may still have a strong dose of “boys are rational, girls are un-self-controlled creatures of pure feeling who find logic and indeed all verbal argument to be vaguely unfeminine”. I suppose the main remedy would be (a) to try and correct this the same way you would correct any other sort of childhood damage to sanity and (b) present strong female rationalist role models.

(5) The complementary hypothesis is a direct sex difference—i.e., the average female human actually is less interested in and compelled by deliberative reasoning compared to the average male human. If you were motivated to correct the sex balance regardless, you would consider e.g. where to find a prefiltered audience of people compellable by deliberative reasoning, a group that already happened to have good gender balance, and go recruiting there.

(6) We could be looking an indirect gender difference. Say, boys are raised to find a concept like “tsuyoku naritai” (“I want to become stronger”) appealing, while girls are told to shut up and keep their heads down. If the masculine gender concept has a stronger endorsement of aspiring to self-improvement, it will, as a side effect, make a stronger endorsement of improving one’s rationality. Again, the solutions would be female authors to tailor introductions to feminine audiences, and strong female role models. (If you’re a woman and you’re a talented writer and speaker, consider reading up on antitheism and trying to become a Fifth Horsewoman alongside Dawkins, Dennett, Harris and Hitchens...?)

(7) We could be looking at an indirect sex difference. The obvious evolutionary psychology hypothesis behind the imbalanced gender ratio in the iconoclastic community—the atheist/​libertarian/​technophile cluster—is the idea that males are inherently more attracted to gambles that seem high-risk and high-reward; they are more driven to try out strange ideas that come with big promises, because the genetic payoff for an unusually successful male has a much higher upper bound than the genetic payoff for an unusually successful female. It seems to me that male teenagers especially have something like a higher cognitive temperature, an ability to wander into strange places both good and bad. To some extent, this can be viewed as a problem of authorial style as well as innate dispositions—there’s no law that says you have to emphasize the strangeness. You could start right out with pictures of a happy gender-balanced rationalist unchurch somewhere, and banner the page “A Return To Sanity”. But a difference as basic as “more male teenagers have a high cognitive temperature” could prove very hard to address completely.

(8) Then there’s the hypothesis made infamous by Larry Summers: Male variance in IQ (not the mean) is higher, so the right tail is dominated by males as you get further out. I know that just mentioning this sort of thing can cause a webpage to burst into flames, and so I would like to once again point out that individual IQ differences, whether derived from genes or eating lead-based paint as a kid, are already as awful as it gets—nothing is made any worse by talking about groups, since groups are just made out of individuals. The universe is already dreadful along this dimension, so we shouldn’t care more whether groups are involved—though of course, thanks to our political instincts, we do care. The remedies in this not-actually-any-more-awful case are (a) continue the quest to systematize rationality training so that it is less exclusively the preserve of high-g individuals, and (b) recruit among prefiltered audiences that have good gender balance.

(9) Perhaps women are less underrepresented on Less Wrong than may at first appear, and men are more likely to comment for some reason. Or perhaps women are less likely to choose visibly feminine usernames. The gender ratio at physical meetups, while still unbalanced, seems noticeably better than the visible gender ratio among active commenters on the Internet. Not very plausible as a complete explanation; but we should consider hypotheses that involve unbalanced participation/​visibility rather than unbalanced attraction/​retention.

Part of the sequence The Craft and the Community

Next post: “My Way”

Previous post: “Bayesians vs. Barbarians”

Eliezer Yudkowsky16 Apr 2009 0:56 UTC
60 points
360 comments5 min readLW linkArchive