According to an anonymous friend of mine, the University of California at Riverside reported having received 1700 applications for one faculty position. I hope they have some kind of smart text mining application, because sorting out 1700 applications ought to be a hard task.
I submit to you that finding the best candidate, in such a setting, is an ill-posed problem. You cannot possibly tell me that under some sensible ordering of the candidates, one comes up on top.
Here’s a proposed randomized algorithm: find the best 10% candidates according to some sensible measure, then choose one candidate at random. I bet my algorithm is hard to beat and would be able to predict accurately how a selection committee works.