[defanor]

That's roughly my first thought too: the project seemed like pubsub with message authentication and without federation. Was about to bring up "Atom and RSS, mailing lists, NNTP, XMPP, ActivityPub, Matrix" as examples of existing protocols capable of that (to a different extent, and usually with more functionality), thought to add Mastodon to the list of examples too, but the README already mentions that Mastodon is undesirable because of the dependence on domain names/administrators (and elsewhere mentions undesirability of running servers by users, of having many servers, while considers simplicity from the point of view of in-browser JS, apparently), that the servers aren't run well, and so on; seems pretty opinionated and enthusiastic about this protocol, so I guess those existing protocols won't satisfy all those requirements either.

Initially restrained from posting it, since it's possibly rather grumpy, and the authors seem to have fun with the protocol, but here it is.

[bawolff]

The story around relays seems to me like the most suspect part of this when it comes to scaling.

The protocol doesnt want users to run their own servers, but there also aren't really any incentives to run relays, so what happens when it gets big enough that running a relay is expensive or non-trivial? I feel like it would just fall back to users running their own servers like in mastadon.

[Kinrany]

That's not a problem if the server owners don't have any power over users.

[bawolff]

They still have to exist. The usual reason for existing is to exploit their power over users. When its cheap to host people will do it out of the goodness of their heart, once it hits scale the goodness tends to dissappear.