Presumably most people don't want their life threatened so almost nobody will pick an instance that allows that to happen. The few people who are interested in posting death threats can go ahead and make a rogue Gab instance or something, and every other instance will block them. They'll be their own silo where super sketchy things go on until the feds decide it's time to step in (since death treats are legally actionable). Such silo websites are already possible without the fediverse (and by some accounts already exist), so my model doesn't change anything here (assuming blocking is effective). (BTW even Gab dot com would block that; they ban people for death threats and porn and maybe even Doxing).

Disinformation campaigns? Well again, if you don't want to be misinformed, pick your gatekeeper. Probably all the better. If somebody's going to protect me from being misinformed, I want to know what their biases are. I don't trust Twitter or Facebook; it's not a simple and objective task.

I think your point that this stuff is complicated is exactly my point. It is complicated, and people have different preferences. No one company can be the gatekeeper that satisfies the entire planet's preferences (at least, not under the current models).