Most users don't want to wade through toxicity to get to signal. If they're discussing a topic of interest, say baking, and someone comes in and starts ranting on how a vast global conspiracy made up of surprisingly-homogeneous ethnicity given its global scale is pushing up the price of yeast to weaken the market for white bread, either the moderators squelch that noise or people who want to talk about baking go somewhere else to do it.
Given their own freedom, when given a choice, users tend to select moderated channels over unmoderated ones. We've been doing the Internet long enough to know this to be true.
The end result is the same, but it’s important to understand exactly where the mechanism is failing if you want to fix it.
This is true, but unfortunately the same mistakes keep being made because people don't pay attention to the history of the internet or didn't grow up during that era. We've known that completely unfettered discussion leads to self destruction since the Usenet era. But the lessons aren't heeded or ignored, so we get people that either stay ignorant or learn the hard way.
Maybe toxic people just congregate in places where their speech is accepted, therefore making the rest of the site toxic as well.
Maybe it’s not “hate mongers and trolls” that overrun sites and that the concept of free speech and being able to say anything just naturally brings out the worst people and the worst in people.
It makes it hard to start a new platform. People start free-speech platforms with good intentions of having open debate about controversial topics. But they quickly get overrun by hate mongers and trolls, and become too noxious for most people to read. Intentional or not, it's a good strategy by the existing platforms to kick out the nasty people, ensuring that they're first to sign up for every new social network.