Personally the reason I'm hesitant to entertain this argument is that code is so awesome - you can reach a billion (1,000,000,000) people practically overnight if your idea is good enough. While women traditionally are more interested in the design/social impact of what they're doing, if the engineering environment is easy enough there's no reason they can't code it themselves. And they are
highly incentivized to do so.
I personally look at the outcome (you can reach a billion people just by having a cool idea and sitting down and doing it) then look at the hurdles to that outcome.
One of the hurdles is that engineers love wasting literally hundreds of each others' hours. If I were building something as a newbie, people have no trouble sending me off on a wild goose chase to spend a hundred hours with my head in documentation. When an alternative ecosystem lets me literally type a single line and it's done. (In this case the first ecosystem hates my time and the second one respects it.)
This isn't really related to gender, but it's another hurdle. So when you look at the incentive (reaching people, fame, fortune, money), if there is a highly unequal distribution you have to look at some of the hurdles.
By the way no one knows your gender online when you're learning stuff, you can use a non-gendered username and get treated exactly the same asking stuff online and so forth.
In this case the hurdle is artificial: being cooped up with a computer while the tools you're trying to use waste thousands of your hours. But it's still a hurdle that impacts genders in a different way, and that matters.