Spend time looking for a good candidate in a random group rather than
from a group that is otherwise prequalified would be likely be
neither prudent nor profitable.

That is not to say the occasional street urchin can't be a real gem,
but that's not the way to place your bet. And by the way, hiring
someone is always a bet, and at the professional levels, a big one. In
my business an underperforming post doc carries a significant out of
pocket cost, and the lost opportunity costs are even higher. In a
moderately high tech company like the one my husband owns, he figures
hiring a new engineer will cost him a three to five man-months of
labor, including support and supervisory time, before his technology
manager will be sure the new hire can really make significant
contributions.

Again, these are real world considerations, not theory.