This sounds like risk aversion to me. People place twice as much emphasis on avoiding loss than they place on gaining something equivalent (even though the end result is the same). If you look at it in this light, then of course the good ones weren't as good as the bad ones were bad. The end result is that you end up with employees who definitely don't suck, but aren't really great either because you spent all of your time avoiding bad programmers and not enough finding good ones.