I have the impression that a proposal has some basic merit, and fussing over the details often doesn't change the appeal of the proposed work has to the panel. If one is going to discover a new planet, it's easy to get every one excited, even if there are typos and more substantial flaws. If one proposes repeating an experiment already conducted 100 times for the 101st time in another place, polishing the proposal may not improve any one's opinion. But maybe I'm an optimist in thinking that the system tends to be fair.

And polling the reviewers' score is not a reliable measure. The panels look for incisive comments to discuss, rather than take the scores at face value. Probably half the reviews don't have much influence on either the discussion or the scoring due to lack of compelling reasoning behind their opinion.

I'm not denying the random component to the funding of proposals, but people often read too much into reviewer scores and erratic reviewer comments.