[jrells]

There's a lot to take issue with in this article. Graduate institutional affiliation is huge because more and more research is being done by undergraduates, so top institutions can more easily pick out the best talent early. It is still true in academia that research>institution, but the latter is becoming more and more correlated with the former. Top PhD graduates come from top schools, with few exceptions (and those exceptions still get jobs). With undergraduate this correlation is weak.

Yes financial privilege is huge for getting that far, but everyone applying for an academic PhD program has spent decent time in academia and has chosen that instead of money. I've never seen or heard of someone turning down an elite academic PhD program because of a low stipend (students do use stipends to choose between similar schools). Any academic adviser would tell you you'd be crazy to do so.

Edit: This might only be true for math. In undergrad I was constantly surprised to find students from elite vs sub-elite institutions differed hugely in privilege and less in ability. In my PhD program, there was a huge difference in ability between our school (arguably #3 in the area) and nearby Harvard/MIT. While this is all anecdotal, elite academia is a small community and I've seen a large chunk of math. Ability here in math is easy to measure (if they can solve my problems and I can't understand theirs). Mathematicians keep mental rankings of other mathematicians. When you apply for graduate school and later jobs, lots of other mathematicians have a really good idea how you compare.

[nuncanada]

About being correlated, I guess it is simply not true, just check statistical models. What happens is that increasingly what is considered good research are what top schools are researching... Like String theory in Physics...