I'm sure lots of stuff gets flagged on Facebook, and in theory they could let people debate it sometimes, but then they still have to make a decision, right? On what basis do they do it?

We have lots of experience with online communities now and it's been pretty well proven that "the people" don't spontaneously organize into well-moderated communities. Software architecture can help make that more likely, but even places like Hacker News don't run themselves - the moderators have to keep tweaking things.

Having some good choices between communities that are moderated differently is about the best we can do.