Huh, that minority-squared effect is interesting, but I’m not sure it need apply here. It’d be individuals coming here, right? It doesn’t take a group to, well, come to LW.
Network-effect makes a big difference too. After all—you have to arrive here some way—usually by being told about it by friends. Sure, some people arrive by accident—just happen to be browsing through HP fanfic or something… but a lot will arrive through their friends… and a lot will stay because they find friends.
...which leads us back to people’s friends tending to be same-sex. If there are few people of your own sex in the group then it’s got less… er… ambient friend-potential…
you have to work harder to be with a bunch of people that have a different culture than yourself. Genders have different cultures, so add that on top of the new culture of LW itself and unless you’re a particularly socially-capable person (and LWs tend not to be), then it’s less likely that you’ll find friends.
Obviously this is a generalisation and likely only a very small part of the pressures involved in a very complicated process… but it’s there.
While it’s ultimately true that individuals come to LW, not groups, I’m far more likely to follow and especially to comment on blogs that my friends also read. For me, one primary way I get really interested in subjects and motivated to understand them well is by talking about them to my friends in real life. And most of my friends are girls.
hrm… actually, I’m reminded of something. Several years back, someone designed these simulations that basically ran an algorithm like “assume people don’t mind being around people that are different, so long as at least some small fraction of their nearest neighbors are also like themselves.”, and basically simulated people moving around to fulfil those criteria.
I think the point is that there are multiple factors which all reduce the chances. In startup founding, you need multiple similar people; in LW browsing, you need multiple personal characteristics. Maybe 90% of women can handle the disagreeableness; maybe an independent 90% can handle the male-style-writing; maybe another 90% is unswayed by cultural gender differences; maybe another 90% are unaffected by a female genetic predisposition against reasoning (I’m just running down Eliezer’s list), and so on.
A LW commentor who is female would be in the subset of women who is in all these groups. (Just with these few factors, we’re down to something like only 60% of women are ‘eligible’ for LW membership to begin with!)
Sure, that all makes sense, but an LW commentor who is male would also have to fall into multiple subsets.
No, remember that our various sets are already biased towards males (obviously males don’t mind ‘male-style-writing’). The point of my comment is that a few small biases can quickly multiply up. If on all these factors, the males are at 95% where the females are at 90%, then we only need like 10 factors before we would expect twice as many males than females based just on those factors alone and ignoring any feedback or network effects.
Why we mostly have male-style-writing, or why there might be a female genetic predisposition against reasoning, are all different issues one would expect different answers to.
(That there are such gender differences isn’t too terribly surprising to me, personally—finding that males and females are exactly the same on all these factors would be like finding that all of humanity is 100% genetically homogeneous, and that there’s no truth to, say, the Japanese having a low tolerance for alcohol or some groups being lactose-intolerant or Africans being disposed to sickle-cell anemia.)
Huh, that minority-squared effect is interesting, but I’m not sure it need apply here. It’d be individuals coming here, right? It doesn’t take a group to, well, come to LW.
Or am I misunderstanding your point in some way?
Network-effect makes a big difference too. After all—you have to arrive here some way—usually by being told about it by friends. Sure, some people arrive by accident—just happen to be browsing through HP fanfic or something… but a lot will arrive through their friends… and a lot will stay because they find friends.
...which leads us back to people’s friends tending to be same-sex. If there are few people of your own sex in the group then it’s got less… er… ambient friend-potential…
you have to work harder to be with a bunch of people that have a different culture than yourself. Genders have different cultures, so add that on top of the new culture of LW itself and unless you’re a particularly socially-capable person (and LWs tend not to be), then it’s less likely that you’ll find friends.
Obviously this is a generalisation and likely only a very small part of the pressures involved in a very complicated process… but it’s there.
While it’s ultimately true that individuals come to LW, not groups, I’m far more likely to follow and especially to comment on blogs that my friends also read. For me, one primary way I get really interested in subjects and motivated to understand them well is by talking about them to my friends in real life. And most of my friends are girls.
hrm… actually, I’m reminded of something. Several years back, someone designed these simulations that basically ran an algorithm like “assume people don’t mind being around people that are different, so long as at least some small fraction of their nearest neighbors are also like themselves.”, and basically simulated people moving around to fulfil those criteria.
The simulation would consistently produce highly segregated results. Aha! here’s a site with applets that run such simulations: http://www.dartmouth.edu/~segregation/segregation-simulator.html
Just an short addendum: Thomas Schelling is the one who originally thought up this model.
I think the point is that there are multiple factors which all reduce the chances. In startup founding, you need multiple similar people; in LW browsing, you need multiple personal characteristics. Maybe 90% of women can handle the disagreeableness; maybe an independent 90% can handle the male-style-writing; maybe another 90% is unswayed by cultural gender differences; maybe another 90% are unaffected by a female genetic predisposition against reasoning (I’m just running down Eliezer’s list), and so on.
A LW commentor who is female would be in the subset of women who is in all these groups. (Just with these few factors, we’re down to something like only 60% of women are ‘eligible’ for LW membership to begin with!)
Sure, that all makes sense, but an LW commentor who is male would also have to fall into multiple subsets.
The question isn’t “why are so few members of the total human population on LW?” but “what’s with the different proportion of males and females?”
No, remember that our various sets are already biased towards males (obviously males don’t mind ‘male-style-writing’). The point of my comment is that a few small biases can quickly multiply up. If on all these factors, the males are at 95% where the females are at 90%, then we only need like 10 factors before we would expect twice as many males than females based just on those factors alone and ignoring any feedback or network effects.
Why we mostly have male-style-writing, or why there might be a female genetic predisposition against reasoning, are all different issues one would expect different answers to.
(That there are such gender differences isn’t too terribly surprising to me, personally—finding that males and females are exactly the same on all these factors would be like finding that all of humanity is 100% genetically homogeneous, and that there’s no truth to, say, the Japanese having a low tolerance for alcohol or some groups being lactose-intolerant or Africans being disposed to sickle-cell anemia.)