There's your problem. By using the word "shouldn't" you seem to be conflating "our best guess at objective reality" with "what we think would be morally correct". It is unfortunate that thinking like this is still allowed to infest some higher educational institutions but it doesn't cut the mustard when nobody's funding is on the line.
In terms of objective reality this assumption is unwarranted. There are significant documented differences between the distributions of intellectual capabilities of genders including the higher variance in IQ for men and men's aptitude skew towards maths and away from language.
> In the past, simple, personality/preference/constitution-based explanations for gender discrepancies have proven false many times - women didn't have the constitutions to be doctors, women didn't have the temperament to be lawyers, etc etc - so I'm inclined to distrust this sort of explanation, at least initially.
In the 70s, 80s and even 90s this would have made more sense. But as the years of higher university attendance of women stretch out, as women continue to succeed in previously male dominated areas such as medicine and law you have to ask 'why not in tech/CS/startups'. It isn't as if it is a area that has ever been reputed as having a history of institution chauvinism. To me, the weight of evidence points more and more towards the differences in distributions of capabilities and preferences between genders playing a significant part in participation rates in tech. However much of an 'Inconvenient Likelihood' this may be for both the tech industry and many women, it shouldn't blind us to evaluating the evidence as objectively as possible.