At my company (tech) I've noticed that for the past 2-3 years grads are almost 50% women.

Now, I have absolutely nothing against this outcome. But I do wonder - instead of optimizing for a specific distribution of employee features shouldn't we be optimizing for hiring the best?

And you could say "they are the best, 50% of the best are women".

That's a possible explanation! However.... 5+ years ago when grad were roughly 100% men, weren't we hiring the best then? Surely back then they also thought they were hiring the best. Surely 5 years ago if you'd told the hiring manager "hey from those 20 people you hired, 10 aren't the best, 10 the best are these other people and they happen to be women", the hiring manager would've said "no way, we don't look at gender when hiring, we just hire the best".

My point is that we didn't understand why back then we were ending up with 100% men despite the fact that 50% of the population are women. We just mandated that 50% should be women. This is like you believe you have a bug and so you tweak something at random. Now it's different and you think it's fixed.

Anyway, they don't pay me enough, so I don't care :-)

I could see it if stakeholders feel they are engaged in a quiotic endeavor...

As a father of 3 daughters, I still see a push for women in stem, but anecdotally my youngest is often the only girl in such activities.