So? The system doesn't have to be perfect, it only needs to make trolling time consuming enough to reduce it to tolerable levels.

> hate speech is posted or linked to by a notable public figure

Here's how I would approach it. Accounts that are tied to real identities have a much higher threshold for bans -- public figure or not. Sans a credible threat of harm to specific individuals they can largely say whatever they would like, hateful or not. When pseudoanonymous accounts get flagged their choice is to attach a publicly visible identity to the account or be banned.

Even a single mistake will end up in the news as evidence that Facebook is "right wing" or "hard left," and perhaps both at the same time.

If you think this is easy, you don't understand the problem.

It has to be perfect. People will be enraged if they can find any example of it not working. The human brain only understands examples, not rates. Facebook is now an all-knowing god that cannot make mistakes or not know something.