If competence is, as the article suggests, normally distributed, then there's no natural level at which to draw the cutoff line. But anyway, iirc, studies show no correlation between outcomes and class rank (nor selectiveness of medical school), at least for surgeons.