Only if there are significant talent differences. If everyone who went to Harvard already has the same base level of talent necessary to be a success, then you might as well pick random grads (which is basically what the surveyed peoples' methodologies boil down to) because the remaining variation isn't worth measuring.