Earlier quoted context omitted.
If you don't believe in censoring communications between a willing-speaker & willing-listener, anything they do has a 100% false-positive rate.
If only Facebook, Twitter, and YouTube were completely full of willing listeners for hate speech, then that might be applicable. Hate groups aren't sending private forum messages to each other, they're broadcasting Even John Stuart Mill, a founding philosopher of free speech, postulated that speech needs to be curtailed when the expression harms others.
And I wonder if his reasoning applies to the kinds of speech policing Facebook is doing, or only to the same sorts of things modern US free-speech law makes illegal: direct threats & incitements to imminent violent acts. (Notably not illegal: 'hate speech', blasphemy, insults, abstract advocacy of most crimes, etc.)