My attempt

My attempt

They are patently false, since IQ tests are designed so that both genders average at 100.

That's only the bottom line. These tests don't try hard to make women score good on 3D recognition, and they don't try hard to make men score good on wording stuff. They actually use a mix of all these different skills to drive the overall score to 100 for an average person (gender independent). So while the test makers acknowledge that there are different "performance profiles" between males and females, they also produce a gender-neutral result. If you like to, you could probably also go on spec.org and create a CPU-neutral benchmark (e.g. neutral between Athlon and Core 2), by deliberately changing the weight of the particular benchmarks (there are enough of those that you can probably solve the equation for quite a number of CPUs to be "equal"). It now no longer tells you which CPU is "faster", but it still will tell you which is faster on 189.lucas (or whatever particular benchmark you choose). Still note that a different "performance profile" of the average male/female doesn't mean that a particular woman is bad at math or a particular man has problems with words. It's just statistics, it just tells you "how many".

And remember, it doesn't have to be an actual "hard" skill. FOSS development is not just about being able to do it, but also about being motivated to do it. You don't get money, it doesn't boost your career, at least not instantly. A lot of capable male software developers will refuse to do FOSS just for this reason. And since FOSS development is not "cubicle based", it also means that you won't meet you peers in person very often. You discuss with them on mailing lists, with a lot of heat on them. If you can't stand the heat, get out of the kitchen. That's the attitude. It can drive away quite a number of people, not just women. A number of high-profile people are famous for being difficult to work with (RMS, Theo de Raadt, recently I've the impression, Linus Torvalds wants to join the list). Maybe being a jerk is a particular required skill for some kind of work in FOSS. Especially when you want to pave a road that's full of thorn bushes (like RMS did). If it's so, we can't get rid of it and we know why so few females participate: There are just so few female jerks (just kidding. Being a jerk is unfortunately not limited to one particular gender ;-).

All this together can make a huge difference in numbers. Some of them can be changed, some of them can't. E.g. if we found out that capable women rarely seek endeavors for themselves (that's how most of the men ended up in FOSS), we might need to build up a mentoring system to get women to participate. If women can't stand the heat of a mailing list discussion or IRC, we need some other communication mean. Maybe women developers prefer talking to each other on the phone? I don't know, many male developers prefer telephone conferences as well, but few of them end up in FOSS development. So some of these changes could expand the community in other directions as well.

What is probably a wrong approach is an accusing tone. If some outsider comes in and without having any other merits, tells me "you scare me off because of <insult>," would I follow that advice? Probably not. There might be some truth behind the insult, but being open about other's faults doesn't work out that well. You don't report outright bugs. You report problems. The problem might be you yourself; the problem in software is often located 50cm in front of the screen. In social interactions, it's even more often. This procedure is well known here, we can deal with it. Follow it, we can find solutions.