This gets even murkier when you consider that grant award and citation statistics are correlated only like 0.3 or something.

It's the Matthew effect multiplied by Goodhart's law.