sort of agreed, but “we won’t ban anything ever” has the failure state that eventually the government will send people to shoot you and society will cheer them for doing so, which seems like it’s also not really a stable equilibrium to me
The glib thing to say here is “I’m prepared to wait and prepare for the day when the government sends men with guns to go deal with the wicked wicked Tumblr posts.”
But in fact that’s too glib, because there is in fact a category of Tumblr posts where this is something akin to a genuine danger: Actual For-Reals Child Porn.
And that example does a pretty OK job of illustrating the principle I have in mind. You can have a viable ban on Actual For-Reals Child Porn, because a large majority of the userbase is willing to say “yes, not only is that terrible, it is distinctively worse than all the other kinds of posts that I hate.” A policy that bans Actual For-Reals Child Porn, and nothing else, can have legs; it exists at a level of censure whose existence has widespread intuitive support.
This does not work with (for example) bans on “Nazi propaganda.” Even though, in fact, pretty much everyone loathes Nazi propaganda. There’s no real agreement on what content falls into that category, and there’s no bright line distinguishing that category from a host of adjacent categories, all of which are beloved and despised by different combinations of constituencies.
[To be absolutely clear: this is pure pragmatics. There is also the overarching moral rule by which banning speech is bad, and by which banning speech by dint of its discursive content is very bad. But I am not particularly trying to have that moral fight right now.]