Let me ask this: what difference does the statistical differences found between genders in the general population matter for any small subset of that population?
If I had test scores from 1-1000 with 1 being the worst and 1000 being the best for a million people that match a normal distribution and I told you that I have a set of 10 that I picked non-randomly but I don't tell you how, what can you tell me about that sub-set of 10?
Nothing. Absolutely nothing.
What does it say about how that set of ten will do on the next test given only that you know the test will be an iteration on the last with some differences?
Nothing. Absolutely nothing.
So, why would it matter if, statistically, females exhibit certain behaviors more or less than males in the general population when applied to the subset that apply and work for Google?
Same. Answer.
What disturbs me (having Google stock, as many do) is how a person that doesn't understand this basic fact got past the Google hiring process and ended up an employee.