Why not accept it (and try to improve it)? It seems futile given the basic fact that it is impossible to have perfect information about the decisions you make. Therefore the options are to make decisions completely randomly, or use some Bayesian probability to try to make better bets.
The assertion here is that the effect of these signals on the chances of success are not very large and in fact the systemic prejudices they create against otherwise qualified pool of candidates, and the pressure they create for people to obtain these (arguably meaningless) credentials has a worse overall effect on the labor pool.
So these signals wouldn’t really help you make better bets, and they have bad side effects, which is why we need to bin them.
