It is wrong, if not dangerous, to evaluate universities by abstract metrics such as research or academic performance.
I fail to guess in what sense research and academic performance are somehow "abstract metrics" while non-academic diversity isn't. You may dispute their effectiveness but SAT/GPA scores, number of peer-reviewed publications, million dollars in grants, etc. are certainly less abstract than the color of your skin or the word on the street about the school you attended.
They are abstract in the sense that it is not immediately clear what makes them go up or down.
Does the presence of a sports team make students feel more school pride, increasing student performance? How much does commute times factor in? What about whether there's political activist students that successfully negotiate better benefits for TAs?
If you look at metrics alone, you lose the woods for the trees. It looks like CalTech have their head on straight, at least according to some comments here, about potential and passion, not about numbers. This is what the blog post advocates, and that I think is a very short-sighted position to take.
I fail to guess in what sense research and academic performance are somehow "abstract metrics" while non-academic diversity isn't. You may dispute their effectiveness but SAT/GPA scores, number of peer-reviewed publications, million dollars in grants, etc. are certainly less abstract than the color of your skin or the word on the street about the school you attended.