3 ms·
It's a nice piece of Bayesian reasoning, but it increasingly fails to hold up. The problem is that while there's a very high probability that a randomly-chosen
by eli_gottlieb 14y ago
It's a nice piece of Bayesian reasoning, but it increasingly fails to hold up.
The problem is that while there's a very high probability that a randomly-chosen graduate of MIT is very smart and qualified, there's an extremely low probably (and going further down) that any randomly chosen very-smart-and-qualified graduate went to MIT. So by biasing your hiring towards MIT, you're almost eliminating false positives at the cost of massively increasing false negatives during the hot part of the business cycle when demand for new workers is highest.
I've seen a similar effect in action personally. I went to UMass Amherst for undergrad, and when I finished I did interviews. Some companies wanted to talk, many didn't, I ended up with two offers from BigCos and lots from SmallCos and start-ups (this was 2011, so yeah, during a hot-hiring cycle), but still a noticeable "failure rate".
I get a little bit of real work experience and then come to Technion (which, at least in this country, is not known to be called "Israel Institute of Technology" for nothing) for graduate school. Internship-application season arrives. Now everyone wants to talk, and the only things to which I can rationally attribute the change are the name-brand of Technion and the fact of having had some work experience (which does set me strongly above "pure academics", but many people work between undergrad and grad school).
I didn't change (much). I just acquired a pre-made Bayesian evaluation of my potential from someone whose word is taken more seriously than that of UMass Amherst. And the really neat trick here is that Technion simply is not as selective as MIT, by sheer numbers: I'm told that maybe the top 10% of applicants can get into Technion for undergrad, and another 20% of those (2% total) can do well enough to go on for an advanced degree. These are far more than the percentages of applicants who can get into the CS programs at Stanford, CMU, Berkeley and MIT combined. Technion just doesn't have equally tough competition for entry due to its smaller applicant pool.