[federererer]

Are you saying that among those 30 qualified candidates, they're all at exactly the same level of skill/talent/experience, or are otherwise completely interchangeable?

I find that hard to believe, at least for any realistic scenario.

Even if all of those candidates exceed the minimum level of merit required for the job, it's very likely that some will still have more or better skill/talent/experience than the others. There will still be an ordering based purely on merit.

Ignoring this ordering when choosing the 10 successful candidates would be a case of ignoring merit, which would indicate a lowering of standards.

[geofft]

In practice, there just isn't a total ordering among candidates like that. You'll find some who are better at some things than others, but you need a mix of strengths for the team anyway.

And I'm not aware of any company that actually manages to rank all candidates by their exact position relative to other candidates. Interviewers tend to get to say just "hire" / "no-hire", not all candidates talk to the same interviewers, etc.; that loss of information isn't, in practice considered an unacceptable lowering of standards.

So, given that standards have already been lowered in the real world from this ideal, affirmative action is certainly not lowering them any more.

[wpietri]

I definitely agree on there not being a total ordering. Part of it for me is the mix of strengths thing. But I noticed two other factors:

One is just the small amount of information gleaned. Last time I hired, I ended up spending ~5 hours with each candidate who made it all the way through. But that's 0.1% of how much time I hope to spend with them, and it's them attempting to put their best foot forward. Any score I might give them would have big error bars.

The other is the extent to which ranking candidates is personal and nearly arbitrary. This time we did blinded reviews of code and it amazed me to see how often each reviewer valued different things. We would eventually converge with discussion, but I don't have a lot of faith that the parallel-universe versions of ourselves would be particularly consistent. And that's without reviewers even knowing the identities of the code writers.

The whole "lowered standards" thing strikes me as built upon a fantasy of clarity that is nothing like my actual experience of hiring.

[sulam]

You're making an assumption here, which is that diversity does not provide value in and of itself. Yet there is ample research that shows that more diverse teams are more creative in problem solving and have better outcomes with tasks that require such creativity. Therefore, picking someone who can't whiteboard quite as well as another guy, but clearly has the skills necessary to do the job and adds diversity, is potentially a net win for the company.

[Pyxl101]

> Yet there is ample research that shows that more diverse teams are more creative in problem solving and have better outcomes with tasks that require such creativity.

Could you point to some of this research? I'd be interested to learn more about research that has been done.