How? Assume for the sake of argument that discrimination is the only factor influencing the gender distribution among CS professionals. From kindergarten onwards the numbers are slowly winnowed down to 80% men and 20% women.
But now this 80/20 population enters a highly selective but objective contest and comes out with 349/1 (assuming there have always been 25 finalists in the 14 years of the competition and there was only this one woman I can't see in the photo proof). Surely this means that men are just better at programming?
But in fact men are just better at being outliers. The effect is not very strong at small deviations (i.e. the difference between an average person and an average programmer), but can absolutely become significant when you only look at the extreme ends of the distribution. (I have written a comment about that here https://news.ycombinator.com/item?id=14997524) When you only look at the 25 top performers, that is bound to influence the result.
Again, I don't think things are nearly so black-and-white, but pointing at one extreme case of men performing better wouldn't be enough to completely validate Damore's memo, if anyone were so stupid to make that argument.
I agree on your point that the discrimination begins very early on.
The aforementioned article, although, bolsters the point that females may aspire to adopt different career options. Therefore, the organizations can't be held responsible for creating the gap/divide. And I seem to agree with Damores's memo because he mentions solutions to the problem of gender-ratio-gap by suggesting programs to attract women to tech.
In conclusion, complaining about the gender ratio alone won't fetch us equality but collectively attacking the root of the problem may.
And so the further assumption was that if we do what other fields have done, make evaluations "blind" (the way it was done in Orchestras) those biases would not be able to express themselves and we would get equal outcomes. Or at least more equal outcomes.
Easy peasy!
The problem is it doesn't seem to be working out that way in tech. In fact, when we "blind" the processes, the needle appears to move the other way. Suggesting that, yes, there was bias, but not the bias we thought.
But that's not in the set of politically permissible answers/outcomes.
So companies are scrambling and trying their best to somehow bend reality to fit the accepted parameters. Which, well, leads to a lot of bending. Of rules, laws, reality.
In fact, With the exception of 2011, in 14 years of Code Jam, no Female has ever made it to the finals. Google's own non-discriminatory machine tool that evaluates codes, which doesn't take into account one's age, gender, nationality or skin color into account, has shed some light on some embarrassing outcomes.