[cm2187]

The population of Google technical employees is I think the equivalent to the population of Olympic athletes in my example. If the general population has a different distribution (again whether you agree or not with that assumption), then there should be a natural imbalance among the cream of the top that made it to Google. And as you said, those who made it are all "Olympic athletes" (in theory, they should have been recruited based on individual merit) so these distributions do not mean that a female google employee would be more or less capable than a male, but it will result in less female google employees than male employees.

[pg314]

> If the general population has a different distribution (again whether you agree or not with that assumption), then there should be a natural imbalance among the cream of the top that made it to Google.

Your conclusion does not follow from the premise. It's not hard to come up with a distribution that is identical for the top x% but differs for the general population. Or one where the difference in top x% is in the opposite direction of the general population (e.g. two normal distributions where mu1 < mu2, and sigma1 > sigma2).

[KirinDave]

> but it will result in less female google employees than male employees.

Well, again, the Google memo started by arguing Google's technical standard was being hurt so James doesn't seem to believe his colleagues all got the same interviews he did.

I'd be more inclicned to this sort of argument for racial and gender diveristy if Google & others didn't have such abysmly skewed distributions towards white men. Go look at Google's diveristy report and then do some Bayes Rule hacking with it.

[cm2187]

Well, is Google's diversity profile really different from the diversity profile of the computer science departments it recruits from?

[KirinDave]

Incorrect question: you want the graduate rates. Not the staff rates.

Google's diversity rates are worse than the graduate rates for most schools I've checked. I have not found a composite data source to compare a larger average to.

[lawnchair_larry]

It actually beats them.