The federation/migration problem in mastodon seems very important. My broader thought: SMTP/NNTP/etc. thrived because of a set of trusted entities (govt/uni) that could reliably run federated servers. My choice of server is going to be based on which one is most likely to be up and running years in the future.

Who should those entities be today?

I bet someone has already proposed using blockchain to enforce global username uniqueness, so that accountnames aren't tied to an instance. Am I right?

I worked on instant messaging federation for a time back in the day. Check out the history of xmpp (and Google/Pidgin) if you want to see things NOT working out for attempted federation.

Unfortunately, all the easy solutions I can think of begin, "Assuming the triumph of global anarcho-syndicalism..."

The blithe "Pick a server that you trust" on the mastodon sign-up page sums it up. I'd feel foolish saying that I trust any server right now, and I don't even know what criteria to assess servers by.

The most obvious criterion is number of users, which will dissuade federation, as @bcrypt points out.

The second most obvious criterion is, I guess, "Run by some big/huge entity likely to stick around for a while," which has its own problems too.

@auerbach @bcrypt How about trust as in "run by you or someone you know" -- maximizes federation, but limits accessibility to people who can figure out how to set up a server.

@enkiv2 @bcrypt I don't trust that any server I or my friends run will be up in a year, much less 5.

And I'm not sure I understand why you want to limit accessibility?

@auerbach @bcrypt Re: trust, I think people drawn to federated social networks have an interest in having them controlled by people they trust not to fudge/sell their data or keep/distribute it outside what they allow, even if it means having many temporary identities or doing a lot of migration. The two kinds of trust are sort of in conflict, because of the economics of scaling even a pretty tiny federated hierarchical system.