It stands to reason that most papers will not be impressive, so even if the filter is biased it likely has some sort of value.

It's always a balancing act. Become too restrictive and good ideas could get snuffed out, but too open and it's just a giant cluster of bad ideas and too much noise.