every instance is accoutable for the actions of its users. if an instance is not vetting/managing its users wholesale then it gets defederated by other instances. basicially the network self-sorts. abuse instances get lumped to gather to bother each other.

we are always going to have individual users that try to game and abuse the system. email (another federated system) has evolved quite a few defences against just this class of attack - its not perfect, but it is manageable. you will end up with instances that are tightly content moderated, instances that are lightly content moderated and instances that are essentially 4chan or worse.

I am pretty sure we can build in mechanisms to make this tolerable. just my ramblings on the issue.