There’s much less control about the software.
In a federated system you have no control about wheter remote instances are running up-to-date software or even the same type of software (think Lemmy vs Kbin), which makes breaking changes really hard to impossible, since you never know what ancient version another instance might run.
This is part of the reason why e-Mail works the same now as it did in the 80s. If e-Mail was a centralized service, it would be a full communications- and office-suite now, but since it’s federated it’s still separate messages in folders and stuff like grouping messages by thread are considered innovative.
This is part of the reason why e-Mail works the same now as it did in the 80s.
I still want to see a proof that there isn’t a technical solution for this.
There are things like versioned APIs, backwards compatibility… You can make your network protocol modular and extensible… Think of XMPP and some other examples.
E-Mail is somewhat alright and has a few good design choices. That’s why it’s still around today. With the additional lessons learned since then, todays knowledge and tools, I bet we can design some technical solutions to the upgradeablility-problem.
[This comment has been deleted by an automated system]
Lack of centralized control.
Until there’s some kind of organizing central committee of servers that could mutually defederate problematic instances, every server is forced to play whack-a-mole to deal with fascists and pedophiles and the like. Every server can not be an island onto themselves, they should be in communication with each other and then collectively decide on the rules of the federation.
I really really don’t like the idea of a central committee of liberals that will defed any instances that are more radical that “vote blue no matter who!”