On the contrary, if you look at lists of the representation gap between professions, computer programmers fit right into their surroundings. It requires special pleading to explain why programming is systematically discriminating against women, and not civil engineers, sound technicians, chemical engineers, industrial engineers, mechanical engineers, aerospace engineers, professionally active mathematicians, chemical engineers etc...
Computer science, physics, and engineering are the three male bastions of STEM. Nothing else in STEM compares. Moreover, the rest of STEM is significantly and increasingly dependent on computer science, so a very large number of practicing scientists in non-CS, non-engineering fields are spending much of their working time programming computers. Somehow, we've led ourselves to believe that plugging form fields into database columns is serious men's only business, but working with computational models of molecular biology isn't.
It would not challenge my intuitions to find that women in STEM are doing more significant CS work than the overwhelming majority of professional computer programmers.
I would tend to agree. I think that women are not interested in staring at a computer screen and making CRUD apps all day, they do something more interesting or rewarding that require technical skills. That is, the lower aptitude men stick it out, while the lower aptitude women quit and become nurses or whatever. It's just a theory, but plausible right? You seem stuck on a false choice between it there being some simple super-cause or else sexism. It's probably a bunch of things. I mean why else would the most gener-equal societies show the most segregation between professions?
1) Men & women have differently structured brains.
2) Different brains result in different behaviour.
One suspects that if there was some subfield of programming, like distributed systems, that had 45/65 female/male, there'd be an elaborate justification for how that programming discipline was so different from the rest of programming that any comparisons to it were invalid (and, in fact, I've noticed at academic crypto conferences that women are much more common than in the rest of CS, so maybe that's an example).
Again, this is what it means for an argument to be a special pleading: we're required to accept an obstacle course of criterion that ignores the simplest, most plausible comparisons.