I argue that, in the specific problem space of Internet discussion communities, the absence of central guidance has been shown again and again to result in a race to the bottom.

That's why computer networks have struggled with the problem for literally decades, since before http was a glimmer in the mind of Tim Berners-Lee. I well remember early USENET node providers claiming "completely uncensored" access to all newsgroups, only to find within 6 months or a year that they had to dramatically scale back on that promise by restricting the newsgroup list, or cancel certain customers, due to lawbreaking behavior. The problems of discussion forum moderation gave us Section 230, which grants immunity to site moderators for good-faith actions to restrict distribution of information which is "obscene, lewd, lascivious, filthy, excessively violent, harassing, or otherwise objectionable".

Section 230 is pretty much an acknowledgment that without moderation, forums will almost inevitably descend into threats and harassment. And if you think that surely even a non-controversial forum could survive without moderation, look at what happened to Ravelry.