Functionally, my experience within the Fediverse is it's literally just Twitter but with moderation fully broken up between instances and not agreed on. An instance's administrator(s) can decide what instances users are allowed to interact with. This is done either via whitelisting or blacklisting, though almost all instances work via blacklisting. Furthermore, administrators have total control over moderation.
Users pick the instances that closely align with their own ideals. If they don't like the moderation, they can move instances.
The software that backs the Fediverse and the software that consumes it is quite varied. Users are unusually adept at navigating all this.
We had the same sort of frontend diversity in Twitter for many years. This isn't much more than that, honestly. Now you just can match any client with any set of moderation rules you like best. It's very much a free market approach to the question of moderation of user generated content by Big Tech.
When it's so easy to move between instances and you have a large selection of moderation rules to select from for your account's "residence", you no longer really have a legal leg to stand on for "censorship" because you can just move next door where explicit adult gifs are totes cool and your followers will be updated with your new home. There's literally a network of instances that can be as moderated as your most loathsome ban happy video game board to as unmoderated as your favorite anonymous image board. And you get to pick. Not Musk. Not wall street. And certainly not Congress.