Besides the main implementation synapse being slow, the new implementation dendrite is unfinished but progressing.

But technical standards aside, the hassle of managing encryption keys is too buggy and confusing IMO. That'll deter most people I feel.

I want it to be great but it just isn't there yet.

I personally really like Matrix, but there are a few outstanding complaints about it. The biggest one is that the reference implementation everyone uses by default is known to be bloated and slow, and poor at scaling. Server admins have had a huge challenge of supporting a large amount of data for things like room history, which in the past required propagation to every server hosting every participant. The protocol itself has been described by some developers as overtly complex.

Some of this seems to be improving, particularly with development of a Go-based backend implementation, Dendrite.

For some weird reason many people just plain ignore the existance of XMPP when they're discussing decentralized protocols. Is it because Matrix outshadowed it?

Feels like Jabber has been around for a long time. If I'm not mistaken I remember my buddy pimping it as an AIM alternative back in the Napster/mp3 days.

I think what @deadsuperhero@lemmy.ml is trying to explain is that Mastodon and Lemmy for example may be able to implement Sup as a part of their softwares too. Currently if you try to DM someone on Lemmy, the options are a Matrix redirect, or Private Messages designated as not secured.

Though I do wonder about the viability of this. People would still need to set up with Sup to receive messages, at least as I understand it. As a result, this would run into a similar problem as matrix, just without needing to use a different federated server and login I suppose.

This would also require that project leads for AP softwares would want to accept the service as a part of their distribution, assuming it even is implmented in such a manner.

TL;DR it's hard to say anything for certain when details seem to be slim.

Mobile-first, requires phone number, proprietary server-side, google play market.

Signal is great, but the core developer is inherently against federation with his own product. It's just one giant centralized service.