Xor, <whoosh>, as the joke flies over my head.
You can't do this without having a central authority (which people don't want to be beholden to), making names cost money (or be otherwise difficult), or accepting squatters. People aren't going to pay for names and you've gained nothing if you centralize the name store. People aren't going to accept the immobility of federation like they do with email (and it still gravitates towards centralized authorities).
I think we're just going to have to accept large hard-to-gen and hard-to-read identities (e.g. base32'd pub keys like onion service names) and we each "tag" friendly names to them locally as we choose.
Until an indexing service becomes a defacto standard, once again making it centralized. This is what I meant about federation (i.e. sharing your name lists) gravitating towards centralization.
> Squatting on previously-unused-in-the-service IRL identities doesn't matter at all.
It does when I can register thousands and thousands of names. This is why it must be too difficult for squatters and not too difficult for users.
Personally I see no reason to make global addressing mandatory -- could be optional for those who want to be globally findable. Otherwise, use ssb pubkey as identifier and only connect with immediate FoF network.
On top of this I propose building a protocol for store and forward messaging. Something where it is possible for anyone to develop a server that speaks this protocol. Instances of these servers can be registered in the namespace above and this is how these servers can find each other to push data around. Anyone should be able to run these servers, providing service just for themselves, small communities or even large organizations, businesses, institutions or resold commercial services.
Lastly I propose an application layer on top of this. This application layer has two components:
1) A suite of standard protocols for user agents to contact these servers and download user data 2) A suite of standard protocols for rich content, encryption or group distribution
From these pieces, it would be totally possible to build a giant decentralized social network that could present the same sort of data we see on social networks. While things could start with simple messaging, similar features to what we find on social networks could be built.
Of course there's the problem of bad actors and abuse, but it seems this could be solved by merely blocking all communication except for that from a list of known senders... We could call this the friends list.
It would be a lot of work to build and get right, it's a shame nothing like this exists today.