Bayesian filter rules like SpamAssassin, but somehow crowdsourced with weights affixed by community voting, would be more effective than leaving it to individuals. Anointing moderators that get to arbitrarily determine what is “abuse” has always regressed to rampant censorship. It took all of 5 weeks after Trump was elected for Twitter to turn from “free speech wing of the free speech party” to Milo Yiannapolous and I being the first political bans.

There’s no way that intelligence services and governments won’t quickly lobby whoever moderators are the minute they are responsible for a significant number of users, *even if they are anonymous*. Because pubkeys can prove identity trivially. And at least one will approach governments known to pay good money for censorship.