Why should a fallible, biased human “admin” be able to block an individual from communicating with another individual or group of individuals? This just creates weird social silos and echo chambers. It’s made worse because the people who choose to be jannies are often deeply broken people.

The Fediverse has the whole model backwards as a result of its design. Either individuals need to run their own instances, or identity needs to be completely ephemeral and flow through many servers as it does in Nostr. A Nostr admin can choose not to host data from a given identity, they could even use an identity whitelist, but users normally subscribe to multiple relays. In the end, the data gets through, while the relay admin can keep their own hands clean if they have a problem with the data.

Note that having this freedom does not preclude you from voluntarily signing up for blocklists. But this is an active choice, and it can’t be suddenly imposed on a set of users.

> Where did we get this idea that federated instances should never block anything?

I'm not sure, but the OP was postulating that Forgejo instances could federate to solve the problem that the Dutch government runs its own instance to develop its software (including using AI extensively) but wants to engage with the wider community so is currently hosting things on Codeberg's centralised instance of Forgejo, and Codeberg's administrators hate AI.

Adding federation doesn't solve that problem, it just moves it one level up. To collaborate with the Dutch government, you either need to directly register on their site, or set up an entire copy of Forgejo, publicly hosted, so you can federate your one-user instance with them, or you need to find a different instance that someone else hosts, and be at the mercy of that instance's admins, who may be even more cruel and capricious than Codeberg's.

What I predict is a future codeberg.org/about saying

    code.overheid.nl  Suspended (slopfondlers)