The more selection criteria depends on cut-off (which it is in the university setting), the more likely selection from normal distribution is going to be skewed. Then Fields medalists are selected from long tail of already skewed population, all bets are off.