The comparison to law/medicine is particularly interesting, and I haven't seen a good explanation of the differences with that (one might exist, but I haven't found it). My feeling is that 20-30 years ago, computing, law, and medicine were all seen as being fairly similar gender-wise, both in terms of their ratios and perceptions of how friendly/attractive to women the fields were. But in the years since, they've gone in opposite directions: women now make up more than half of med-school students, and an ever-increasing proportion of law-school students, but they make up a smaller portion of CS students today than they did in the 1980s.
Another interesting comparison, IMO, is to banking and finance, where the atmosphere is even more distinctly "big swinging dick", yet the gender ratio is still not as imbalanced as in IT (IIRC, the ratios in finance are more like 65:35, whereas in IT it's somewhere around 80:20 or maybe even worse, depending what sector of IT).
This is one of the main reasons I don't buy the argument that women stay away because men are assholes in tech: the men I've interacted with that are in finance put the worst of the IT douchebags to shame when it comes to misogyny (the guys in the money fields often really hate women, or at least have no respect for them, whereas in tech, they just don't really interact with than many women, so most things tend to be male-centric, I rarely see actual hostility), yet women work in that field far more readily than they do in tech.
I'm sure the remunerative aspects play some role here; it's pretty easy to get rich if you can make it through hell-year at a hedge fund, whereas being code-monkey #8 at some random startup is likely to do little more than pay the bills for another year or two.
But still, you'd think any financial disincentives would apply equally to men and women.
I definitely think there's something else going on here, and it seems to have a lot to do with interest, not incentives or disincentives. To me, the main question is whether the interest discrepancy is cause by an innate difference between men and women (for whatever reason, men like building stuff, writing code, solving problems or whatever, and women don't) or a social one (men are encouraged to do it, women aren't).
Given that I ended up in this field with pretty much zero influence from anyone (I started messing around with APL when I was about 8 and found the executable while poking around for games on my dad's computer), and to this day I've met a grand total of maybe 4 females that were interested enough to learn to program at all, as compared to hundreds of males, I'm skeptical of claims that it's purely social, the numbers are just too extreme for that if the same natural urges exist in all of us.
