Lets say you took a population of candidates with some distribution of individual programming skills, and assigned them randomly into two categories, with a 90-10 split between A and B. E(skill of a candidate in A) = E(skill of a candidate in B)
Now you rank all the candidates in A and B respectively from best to worst, and you pick the top N candidates from A, and top N candidates from B. Taking that set, E(skill of picked candidates from A) > E(skill of picked candidates from B)
There is no prejudice here! A and B are randomly assigned, if you pick any one individual from A, they are exactly the same on average as an individual from B. But conditioning on the constraint that you need to pick the same number from A and B, where A and B have different population size, then you will get the result that A is on average better than B.
In a hypothetical world where sexism does not exist, given the current ratio of men to women in tech we would expect to see a skill disparity in the average if we enforced hiring equal numbers of both men and women.
Edited to add some further points:
1) Note that the above does not hold true if the individual programming skills are constant after some threshold, so the top N from A and top N from B all have the same level of skill. I don't think this is true in my personal experience, but I have seen people make this argument.
2) The above also does not hold true if the ranking mechanism (interviewing process) does not actually rank the programming skills of A and B successfully. I've also seen people make this argument, but note that as long as some correlation exists between [interview result] and [programming skill] then the above still holds true. It seems highly unlikely that there is zero correlation.
3) I don't necessarily agree with the author that this is a bad thing - I think in the long run to get equal numbers of women and men in candidate pools, having equal numbers of female and male programmers at places like Google can only help. While theoretically there may be an average skill difference, I'm not sure that you can really notice it on an individual level because of the sheer number of qualified candidates that apply to a place like google.