Yeah, we should, at the protocol, provide a mechanism for spam moderation such that we don't need relays to do moderation just to keep from being DDoS.

Users should be empowered and spam prevention is really fundamental, but we don't want to divide communities or trust other agents to decide what is spam. That invisibly permits content moderation right beside spam moderation.

I still don't know that we'd ever be able to prevent content moderation, but enabling smaller niche hobbyist relays to exist and function well, without being DDoS or costing so much in resources, is a big step towards ensuring that it really doesn't matter if some relays moderate content too much, if you have client support for work. I would assume that the easier it is to run a relay the more relays are likely to exist that will accept your speech.