It’s simply this: In order for local standards of civility (whether those be vanilla-custard bland or three-chili-pepper strong, it matters not) to be maintained, what’s required is vigorous, hands-on, and visible moderation. Simply providing users with a “flag this post if you find the content objectionable” button isn’t enough. All that does, so far as the user can see, is push the problem off onto a faceless, and possibly automated, minion someplace, for said minion to deal with, or not, according to whatever invisible algorithm may or may not be in place. This does nothing to chide the offending user, or provide immediate feedback to the offended user, or steer the discourse into a less offensive channel.
To do all that, you need a person — a name, a face (even if that face is a cartoon avatar), a consistent presence — to be on the spot and monitoring the venue for discourse that’s about to go toxic. A good moderator can defuse or shut down toxic discourse as needed; a great moderator can spot the warning signs far enough in advance to change the conversation before the toxicity gets a chance to arrive.
Done well — and it has to be done well, if it’s going to be done at all — this is a full time job, and not one to be undertaken by volunteers. If you want somebody to stare into the abyss 24/7 — or better yet, two somebodies, so that they can take enough time off to stay sane — you need to pay them for it.
And sooner than pay good moderators a regular wage, most commercial online fora will either close down comments altogether, or go to one of the now-standard automated systems that end up pleasing nobody.
Not because these sites are run for profit. But because they are run for profit by cheapskates.