Yeah I think you have to stem the tide to avoiding the "unpaid" shills. I've even seen people who shill in one direction just to be accepted by a community so t

Yeah I think you have to stem the tide to avoiding the "unpaid" shills. I've even seen people who shill in one direction just to be accepted by a community so they can troll them. I hypothesize you could detect this type of stuff using semantic analysis of text if you check word order and frequency (or go ML with fastText or w/e) you could get maybe 90% accuracy at determining whether someone isn't a shill. There needs to be an investigative process which humans undergo after this. Then some form of identification would be required before any sort of ban (ideally you tell them, this only stops the false positives from rejoining) can be lifted, just incase any local laws are broken. The shills all have some source material, so they should be closer to it language-wise than those that they teach/brainwash. But hey, I wish I was paid to shill against other shills and detect them, it would be an interesting job.