I suspect that proving code is safe using static analysis is probably NP-complete.

That /. is written in Perl.

What you need to examine for something like this is how it compares between people that went to these schools and people that could have, but didn't. Those who had the grades and test scores, maybe even applied, but elected to go to a state school instead. My bet? Not much difference.

In 1999, economists from Princeton and the Andrew W. Mellon Foundation looked at some of the same data Eide and his colleagues had used, but crunched them in a different way: They compared students at more selective colleges to others of "seemingly comparable ability," based on their SAT scores and class rank, who had attended less selective schools, either by choice or because a top college rejected them. The earnings of graduates in the two groups were about the same — perhaps shifting the ledger in favor of the less expensive, less prestigious route.

but the output is algorithmic with a bit of randomness thrown in.