I think the thing is that in a huge amount of hiring, we don't know what attribute K is, nor if it is good to be considered favorably during hiring. It is amazing how often we look back and see what we used to evaluate people by 50 years ago was so wrong, but believe that we have it right THIS time.
So, as a people we default back to hiring people who look like us, think like us, and act like us, even if that isn't representative of the larger population, and even if it restricts the population of people who are likely to be perfectly qualified, given the chance. Keep the playing field tilted towards "us", whoever that may be in the given situation.
And it just so happens that with our enlightenment we exclude the people who have always been excluded. A funny coincidence how we keep coming up with that same solution, rather than challenging our own assumptions as lazy mammals.
Abstractly, if attribute K which is uncorrelated with merit is considered favorably during hiring, then merit will decrease as an outcome as compared to the baseline where attribute K is not considered, unless attribute K is used strictly as a tie breaker.
Do you think that DEI considerations are in fact strictly used as a tie breaker, and specifically in a government setting? Are there enough ties in the real world for the representation needle to be moved based on altering only tie decisions? I am skeptical of both of these based on my experience with DEI practices, and as far as I know they have moved the representation needle so something must be false.
I’m curious how you are thinking about it though, you sound fairly confident.