I think that's the trick. These university admissions committees are essentially choosing the ruling class for the next generation. What makes a good ruling class depends on more than just test scores and grades, so admissions committees look at other things the applicant has done, and at least they used to also do an interview with an alumnus. All of this is fairly gameable though, and the kind of person who would excessively game these metrics might not be person who they want to choose. Knowing that someone is the child of someone who already was admitted and indoctrinated into the values of the university is a pretty good signal that this person is more likely to be the kind of person they want to admit.

Now all of this runs into the same fundamental issue that any decision like this does, namely, that ideally you want everyone to have an equal chance, but also, you want them to do a good job in their role. Unfortunately, people, through no fault of their own, are born into different circumstances, and some are prepared, in many different ways, better or worse than others, and this strongly affects how well they will perform.

Well, my post sure was a doozy. I wasn't thinking along racial lines, and I'm sorry if it seemed to imply that. Downvotes humbly accepted.

My kids went to a competitive high school, and I saw how the top students funneled themselves into an extremely narrow range of interests. Those kids were nice, but putting 1000 of them in one place would be a freak show.

The choosing of rulers is an interesting and complex problem. An idea with some popularity in HN is "sortition" which is the selection of rulers at random. This could be applied to college admissions in the following way. The college doesn't want to admit the top 1000 applicants according to any short list of KPI's that can be gamed. So they admit from the top 10000 students by manual curation, which is guaranteed to be controversial.

Instead, why not identify the top 10000 applicants, and then send out acceptance letters to 1000 students chosen at random by a neutral third party. (Making up parameters here, just for definiteness). The schools would get the variety they want, with an opaque selection function that can't be inverted, and the same potential benefits that sortition offers for choosing the ruling class.