To be clear, Meta's "Diverse Slate Approach" was explicitly discriminatory with regards to protected class. It prohibited hiring managers from proceeding with an offer until at least one of the on-site interviews contained an "underrepresented group", meaning women and URM males. In practice this meant that white and Asian men had to wait a long time to get an offer even after they passed the interview rounds. But women and URM never had to wait for the DSA to be fulfilled because their presence caused it to be fulfilled.
To see how this is discrimination, consider a company that decides to wait 90 days before proceeding with any offer to a Catholic candidate, but no such delay if the candidate is not Catholic. Surely we see that this is discrimination. Even if we argue that the Catholic candidates are still getting their offer eventually it's still plainly discrimination. What if instead we rolled a dice and added 0 to 90 days of delay? Some Catholics have zero delay, just like non-Catholics. Is that still discrimination? Of course.
Now considering how the DSA works. Most candidates are not "diverse". If you look at the sequence of candidates with non-diverse (N) or diverse (D) candidates the sequence looks something like this:
NNNNNNNNNNNNDNNNNNNNNNNND
So if we introduce the requirement that at least one interviewed candidate must be diverse, a non-diverse candidate may get lucky and the next candidate is diverse. Or maybe they'll be waiting a long time until a diverse candidate is interviewed and fulfills the DSA. But a diverse candidate never has to wait for th DSA to be fulfilled. Sounds like a familiar situation where an offer is randomly delayed on the basis of protected class, right?
It's true that the DSA never lowered standards for diverse candidates. But the way it "improved" the demographics of Meta was through systematically delaying offers on the basis of protected class.