I don't think we necessarily need a single protocol to "win", but I will say the Bluesky/ATProto folks seem the most pragmatic when compared to the ActivityPub and nostr communities.
Mastodon is a bit like email. Not great but it works right now and is reasonably easy to get into and integrate with. The feddiverse is bigger than Mastodon. But it's mainly Mastodon. And Threads soon apparently. Millions of people daily active on it. Not quite critical mass yet but closer than most other things have gotten since Twitter started imploding. Good enough?
IMHO mastodon with signed content and per user key pairs would make a lot of sense and it should not be that hard to add it. Why is this not a thing? Just like using pgp makes a lot of sense for email. Except that never happened either. Already when I started using email in the mid nineties, some people were insisting on using pgp. That's thirty years ago and it was only some people. It remains very uncommon for people to use that. Key management is an obstacle. Techies get it. Everybody else doesn't.
I mean, that's basically the story of the internet as a whole; a teetering tower of "eh, not great, but it'll do for now", since the 80s.
There are active efforts to bring other groups to nostr- check out nos.social for example.
For what?
Just to be clear, nostr doesn't directly require the blockchain for anything. It has a nice integration with lightning payments, but that's optional.
> Bluesky is mainly a thing that people talk about rather than join. And it has the Twitter founder backing it
Pretty sure that's not really true anymore. Jack Dorsey has apparently moved on to nostr, interestingly enough.
> IMHO mastodon with signed content and per user key pairs would make a lot of sense and it should not be that hard to add it. Why is this not a thing?
There's some work in this area[0], but I agree it's surprising this wasn't baked into the protocol from the beginning. Same thing with signed requests. Things have de facto evolved in that directly, but we seem stuck at a local optima using a 5yo version of the HTTP signatures spec (see discussion here[1]), which is really confusing for new implementors as well.
Ultimately I think a lot has been learned from implementing ActivityPub in the wild, but I feel it badly needs a version 2.0, and even then I'm not convinced all the most important lessons have been learned yet. For example, it's very concerning to me that Mastodon still doesn't provide an easy way to bring your own domain to any instance.
[0]: https://codeberg.org/fediverse/fep/src/branch/main/fep/8b32/...
[1]: https://socialhub.activitypub.rocks/t/state-of-http-signatur...
Two of my main grievances with the communities both have to do with identity:
ActivityPub as implemented in the wild doesn't make it easy to migrate your identity between instances. This has caused endless problems and drama with instances shutting down. You also have to use multiple accounts if you want to interact with different formats, ie Mastodon vs Lemmy vs PeerTube. As a user that seems pretty ridiculous. You're losing a huge portion of the advantages of federation. IMO a better model would be for you to create a single account on a universal ActivityPub server that supports all data types. Apps like Mastodon would act as clients, not servers. My reading of the specs makes me think this was the original intention (there is a client to server "C2S" protocol), but then Mastodon got big and no one ended up implementing the C2S protocol, treating Mastodon as the de facto C2S API.
Note there is some work to create these universal servers[0], but I wouldn't hold my breath on Mastodon/Lemmy/etc implementing the C2S protocol.
As for nostr, private key management is a complete non-starter for identity, barring some major new breakthrough. DNS is the pragmatic solution to identity.