This is an interesting one. There has been a lot of discussions around this on certain creator platforms, where moderation is applied (predictably) arbitrarily. In several cases the platform itself doesn’t disclose its decisions. But assuming one wanted a transparent, fair, consistent moderation effort, how would this system be designed at scale? Appeal systems? Internal auditing? Pattern recognition to catch rogue moderators?
The rest of the points, apart from moderation gaps, sounds like the same exact problems many organizations experience internally as they grow. 