to me, the effects of censorship and moderation may feel about the same to a sender (speaker, writer), namely stopping the sender's expression from reaching (some for moderation, most for censorship) others, but the most valuable and most relevant differences in this context seem to be for receivers (listeners, readers). it's moderation when receivers get to tune into the signals they wish to receive, while what they deem noise is filtered out; it's censorship when someone else imposes on the receiver values and restrictions that the receiver would rather not be subjected to, and can't reasonably bypass.

centralized social media with "moderation" ends up being censorship, because it is the platform's values that get imposed over all users, there's no choice. on decentralized platforms, each community (instance) defines its own standards, and it's possibly and relatively easy to move from one instance to another, or even to start one's own, so it's harder for me to regard this sort of moderation as censorship.

I'd still like platforms, both centralized and decentralized, to give users control about how they wish their feeds to be filtered, dynamically selecting others' filtering lists to apply to their own feeds, instead of being subjected to inflexible platforms' values. I picture it more like joining multiple moderation providers, that one can add, check, remove and override when desired, than like joining a network (or an instance thereof) and getting subjected and tied to a provider's unaccountable choices (= too much power)