I think implicit in your calculation here is that someone let in on AA grounds is inferior to someone who wasn't. But the premise of AA is the exact reverse, in fact. Saw two candidates A and B are born with equal amount of future potential: 100fp. Candidate B, for reasons of systematic circumstance, did not perform well before college. Without accounting for systematic circumstance, then, candidate C, who was born with only 90fp, will get admitted before B, even though his fp is higher.

It's not a matter of lowering standards, but normalizing them across racial lines.

A common data-based rebuttal to this is that minority/AA students tend to perform worse even in school than their non-AA counterparts. But that's a separate issue, and is pretty obviously going to be the case considering a lifetime of circumstance. The AA response to this is to promote programs that help AA students to overcome their past circumstances and perform to their full potential.

Your example of Caltech is great! That's a perfectly valid way to normalize demographics, I think, but it doesn't seem to exclude AA or even address the same issue. They seem perfectly complementary to me.