It's already pretty common to choose federation status based on moderation policies. Many instances will only federate with instances that require content warnings for NSFW content, and some won't federate with instances that allow NSFW content at all. This information is machine-readable (see https://instances.social for an example of how it gets used), and lying about it will get you on lists for defederating pretty quickly.
This I understand from a [Popular Servr X has Y rules] - vs other server possibly being different.. and essentially a server admin or moderator group being choosy about broadly banning Z type of servers, and particular AA and BB ones as issues arise.
What I am having a hard time figuring out, is what would tumblr do? If people from twitter servers post stuff via activity pub - (comments?) - and it's against the prude TOS they adopted some years ago - will they be blocking twitter servers outright?
Find a way to block individual twitter users from cross posting?
Will people on tumblr side be notified that a comment reply was attempted but X was blocked and so whatever.. will there be a fail notification send to the activitypub person from server V ?
There is often confusion when comments are written directly to a blog - did the blog owner delete it? Did akismet (wp/automattic's spam filtering thing for comments and more) kill it and no one ever saw it? Was it punted to the spam comments section based on a server it's from or a word in the comment field?
I just don't see how tumblr or similar could possibly scale this without a bunch of moderators - and I also think the ux for the tumblr users and those who would communicate with them via activity is going to go well as people find big holes in communication (no warnings and no info as to what is blocked, how it is blocked, etc)
