It legally can not use that power for non-government entities, including those mirroring that content.
And you're suggesting it should be able to?
> Except that the toxic people drive the others away
And who gets to decide who is toxic? I've seen a user have a meltdown because someone compared democratic socialism to communism, and claimed everyone who disagreed with them was toxic. Should they get to be the judge?
> threats can made sufficiently arguable to be legal, but still clear enough to threaten.
Let's be clear - a threat in and of itself is _not_ illegal. A threat must be a direct, actionable call to violence to be illegal. Someone telling you they're going to make fun of you, post embaressing things or otherwise "threaten" you isn't illegal.
> Putting the burden on each individual to filter everyone
Is a recognition that humans are free agents and get to make their own choices in life. No man or woman is fit to play the censor.
> they will have inflicted their harm upon a lot of people.
What harm? Mean words on the internet? That pales in comparison for a regulatory agency, say the NSC, to suppress true information, say about a major pandemic, by abusing it's powers of censorship.