Perhaps that should also be compensated but you're talking about another cost associated with publication (and even non-publication). Since there's no guarantee that reviewers even read the crap they are assigned to read, I don't think paying them is the best approach. It would perhaps be better to publish the names of the reviewers. And if you reject a paper, you have to go on the record with your gripes. I'm sure these policies would slow reviews down an awful lot though. Reviews are only supposed to uncover blatant errors I suppose, and not offer definitive endorsement of results.