My statement boils down to "trusted relays as a concept, depending on execution, is either good or terrible" and I lean towards community crowdsourcing of opinion as the best way to execute on that concept.

Although transparent policy listings aren't a bad idea either.