> "personal media" should not be arbitrarily controlled
This threw an exception, because "hate speech" is an arbitrary type. It has no objective definition. It is defined by whoever happens to be defining it.
Eg death threats and all sorts of illegal wording are inaccurate in the literal, programming sense. Humans (courts/etc) interpret and ultimately rule if a set of vague concepts is in violation.
If it wasn't for this capability it would be hilariously easy to avoid almost all repercussion from written/spoken word by just adding a dash of vagueness to the subject. Throw in a pinch of speaking in code and you'll never be charged for any crime, yay!
I loosely agree with you, but lets not pretend that english is even remotely as well defined as a programming language in all cases.
We do not allow all english that isn't strictly defined as bad, i thought. I can threaten you in ways that will get me in trouble but are not, by themselves, a threat. English is not that simple.
So my point is merely that it seems to be the precedent is already set for humans to moderate human english. It is not a programming language to be parsed and verified.