I got an idea. Current machine learning techniques should be advanced enough to detect trolls. What if we have a community that is completely moderated by an unsupervised AI?
The good thing is that the AI can be completely open: how is it trained? what are the parameters? This AI can still have bias, but that bias will be obvious to anyone joining this community.
> Current machine learning techniques should be advanced enough to detect trolls.
So your idea to counteract people playing psychological games on others is to put something without the common sense of a three year old in charge of moderation. That's just glorious.
The good thing is that the AI can be completely open: how is it trained? what are the parameters? This AI can still have bias, but that bias will be obvious to anyone joining this community.