One day there is going to be some sort of reckoning, where the question of federating and de-federating gets solved for good.
The way it's looking right now, each instance will have to formally join the federation whose rules most closely match its admin's own values (and thus presumably the values of that instance's users), and each federation will have its own name, and published policies, and processes for applying to federate or appealing de-federation.
I'm not sure how a new instance with no users would get approved, but perhaps it would work like a state declaring independence, or splitting a bee hive. This would make it difficult for individuals to run their own instances, but the big federations might think that's a price worth paying to not have to deal with a never-ending deluge of newly created temporary instances that only exist to send spam.
Ironically this is quite similar to the situation today with email, which is also a federated network, except the big players all federate with each other, and it is in principle possible to send a message to a Gmail address from an SMTP server you run yourself.
Another solution might be to introduce another layer of abstraction at the protocol level, allowing people to run multiple accounts across different federations, and synch up a consistent view of the world. This might, however, require people to tag each post with metadata about which groups of people it is likely to offend, so that the software knows which subsets of your friends should receive it.
Perhaps an AI could help suggest the tags, which would make this process more seamless, but I worry that brigades would form demanding that certain users not be permitted to even have read-only access to certain instances, because of messages they sent on other instances. That seems to be the same social dynamics as the article is discussing.
The way it's looking right now, each instance will have to formally join the federation whose rules most closely match its admin's own values (and thus presumably the values of that instance's users), and each federation will have its own name, and published policies, and processes for applying to federate or appealing de-federation.
I'm not sure how a new instance with no users would get approved, but perhaps it would work like a state declaring independence, or splitting a bee hive. This would make it difficult for individuals to run their own instances, but the big federations might think that's a price worth paying to not have to deal with a never-ending deluge of newly created temporary instances that only exist to send spam.
Ironically this is quite similar to the situation today with email, which is also a federated network, except the big players all federate with each other, and it is in principle possible to send a message to a Gmail address from an SMTP server you run yourself.
Another solution might be to introduce another layer of abstraction at the protocol level, allowing people to run multiple accounts across different federations, and synch up a consistent view of the world. This might, however, require people to tag each post with metadata about which groups of people it is likely to offend, so that the software knows which subsets of your friends should receive it.
Perhaps an AI could help suggest the tags, which would make this process more seamless, but I worry that brigades would form demanding that certain users not be permitted to even have read-only access to certain instances, because of messages they sent on other instances. That seems to be the same social dynamics as the article is discussing.