If there is any difference in preferences, and thus career choice and thus depth and strength of candidate pool, between men and women - and there is strong evidence that there is along one relevant axis [1] - then a policy that tries to create a 50:50 balance by fiat will create sub-optimal outcomes if we're trying to optimize for hiring strong candidates. It may be optimal for equalizing gender ratios, but that may not be as important to company performance as hiring strong candidates.

Of course variance within groups is larger than variance between groups, so making discriminatory policy decisions based on group averages is deeply unfair and offensive. That doesn't mean that setting 50:50 gender representation targets in job roles wouldn't be sub-optimal for what appear to be biological reasons.

A lot of stuff in that memo I thought was very offensive - reasoning about specifics based on group attributes is a logical error in the same class as racism and antisemitism - but it's also not reasonable to say that everyone is equal, and thus the distribution of outcomes should reflect the distribution of the population. Everyone is not equal.

[1] https://www.researchgate.net/publication/38061313_Men_and_Th...