(Personally, I suspect that Lawrence is underestimating the effect of overt sex discrimination in science because he's a biologist. In physics, at least in the US, it's much, much more obvious. But that's beside my point.)
Anyway, in DeLong's comments, somebody identifying himself as "George W. Bush" explains what he thinks is the real problem with women in technical fields (specifically in engineering), and it's kind of strange:
In every engineer's first 2 or 3 jobs, over the first 10 years of his career, there comes a time when he faces a _difficult_ problem. One which MUST be solved. And the only way to solve that problem is to sit down at the computer / pull up a stool in front of the machine / go out and walk the plant floor / search and search in the library * without stopping * until the problem is fixed.Leaving aside for a moment the question of whether this is actually true of female engineers, I find his initial premise about critical problems (particularly all those musts and onlys) hard to believe. I've never actually done any of these 25-hour stints; I always get some sleep after about 14 or 15 hours at the absolute most, maybe as much as 18 or 19 when I was in school.
And it is the nature of these problems that it usually takes 20 or 25 or 30 hours of * continuous * effort by the same person to arrive at a solution.[...]
Clearly you can only do this when you are in your 20s or early 30s. But you must do it at least once to be a successful engineer or engineering manager, and to gain the respect of your peers.
And women just won't do it. [...] The women give up at the 12 hours mark. Or at best 18 hours. They want to call for help, form a team, bring in consultants, get the manufacturer's field engineer in. Anything but stare at the problem for 25 hours until a solution appears.
Granted, I didn't hack it as an academic physicist, so maybe you really need the crucial 25 hours in your hallucinatory vision quest to succeed in physics. But I think I've done pretty well and garnered a certain amount of respect in a software field since then (in particular, wherever I go I tend to accumulate a rep as a tenacious and prolific bug-stomper), and somehow this requirement to go for 25 to 30 never came up. The most heroic hackathons tend, on the outside, to be more like repeated 12-to-14-hour stints over a few weeks. There are people who will stay longer, but they're not necessarily the best programmers (certainly not the best at working with teams, which over the long haul is more important); they're typically just the youngest ones.
In my experience, when I am in one of these obsessive solitary jam sessions, getting several hours of sleep in the middle helps immeasurably. I may lose a logical thread or two and have to pick them up again, but, on the other hand, that can be good. The logical threads that I go down when I'm several hours past bedtime tend to be crazy bad ones, and letting my subconscious stew on the problem and coming back to it in the light of morning can help add in some lateral thinking.
If, on the other hand, he's right, it's fortunate that the really mission-critical engineering problems in this universe all happen to require between 25 and 30 hours of continuous work. If they required 48 or 72 hours, few people would be able to solve them at all and civilization would collapse.
Maybe I'm just reading him unsympathetically, and he's allowing breaks to sleep, decompress and have dinner, and is just adding up total hours until you "call for help, form a team," etc. It's true that, as a male-type guy with a few of the borderline autistic characteristics that Lawrence mentions, I tend to be loath to do that. But if that's what he means, I disagree that no women will do this, because I know and work with female software engineers who easily match or exceed me in solitary bulldog persistence.
The "gain the respect of your peers" part may well be true in some organizations, but, if that's true, that's a cultural problem: they're using obsessive hackathons that don't necessarily produce the best technical solutions as a kind of macho initiation ritual. I have a certain amount of mistrust for the idea that good coding and problem-solving comes from being tough enough.