> I always thought that a better model for moderation would be to distribute the effort. There will be people in my social bubble with similar reactions to the same content and once an algorithm understands these connections, it can distribute the effort of moderation amongst these like-minded people.

This can work in some scenarios, or when the junk is basically randomly distributed — e.g. flagging people spamming random business links — but it's a lot harder when you have coordinated attempts with resources. That's been heavily used to game things like Facebook or Twitter and it's not just bots — someone saying “go here and retweet it” can generate a lot of activity from humans who are decidedly not neutral. That can be done for profit, political reasons, or griefing and the latter two can be quite hard to detect.

I'm not sure we have a great solution for this other than destroying the model those companies currently have of surfacing popular/trending content based on the activity of people you don't follow.