The problem is the federated nature gives a limited view of user behavior to everyone except the home instance, which means identifying spammers, bots and influence ops and effectively banning them is much harder, assuming the mostly volunteer admins even have the time and desire to do so. Federation also introduces the possibility of malicious instance owners.
The problem is it takes time to identify them and kick them out. At a large scale, there will be a constant churn of poorly managed and malicious instances getting access to the network and then not getting booted until weeks or months later.
The problem is the federated nature gives a limited view of user behavior to everyone except the home instance, which means identifying spammers, bots and influence ops and effectively banning them is much harder, assuming the mostly volunteer admins even have the time and desire to do so. Federation also introduces the possibility of malicious instance owners.
I’d argue that it actually makes it much easier not harder. If an instance refuses to moderate itself, it gets kicked out of the network.
The problem is it takes time to identify them and kick them out. At a large scale, there will be a constant churn of poorly managed and malicious instances getting access to the network and then not getting booted until weeks or months later.