Moderation mistakes are inevitable, and the consequences of a bad moderation decision can be as bad, or worse, than the effect of the message itself.
IMO, people should learn to defend themselves against trolls rather than helplessly relying on a nanny from above to do it. Unfortunately, using those platforms doesn't give us much of a choice in the matter.
That's a problem in some generic sense perhaps, but not with the application of this policy, which does not rely on a determination of “good faith criticism”, but instead of dehumanization.
Note that the examples of what is covered all are dehumanizing in a very narrow sense (stating directly that the target groups are vermin, viruses, etc.): https://blog.twitter.com/content/dam/blog-twitter/official/e...