Take population with a massive bias in it (the gender gap in CS/CE/C*) and sample for something that's fairly distributed among genders (the nebulous "talent"), and you'll get a biased result. That's basic statistics.

The goal of the blind review was to decrease the effect of that bias, and (probably) to encourage more women to apply knowing that they wouldn't be rejected on a name-only basis. That failed (probably for the reason above), but the goal still hasn't changed.

I personally think that's a laudable goal, but there's plenty of room for discussion there. I just wanted to point out that it's not hypocritical or contradictory for GitHub to behave the way they have, as the goal hasn't changed.