Aside from "XMPP is old" I don't see any reason to favour Matrix. I also don't see XMPP being around long enough to have a strong ecosystem that's constantly built on to be a bad thing.
From a #selfhosting perspective XMPP is far easier to run as well.
For E2EE they both use OMEMO.
And ofc it's in the name, XMPP is extensible.
Technically speaking I should say XMPP can and often does use OMEMO but can be used to implement literally any type of E2EE you want. You can outright port the full Signal Protocol to it and encrypt metadata if you modify the code a bit. You can even use PGP (although I view this as a legacy feature when OMEMO is more features, uses only ECC, and provides forward security).
@aspie4K Matrix is from one company that provides clients for all major platforms and manages community work. There is no such thing for XMPP. It‘s only thanks to the effort of some single persons that there now is some common sense about which protocol features to implement. Which is not for a long time. And only few clients do implement these features in an acceptable manner.
@patrick @aspie4K Hm, well. That's kinda debate of principles. Many projects out there only work because very few people garantuee their existence. And only few projects are able to do otherwise. XMPP would not have reached any reasonable attention without the work of Daniel Gultsch f.e. You could argue the same about Linux, Linus Torvalds and Red Hat…
I guess my concern is with celebrating this kind of choke point in a project (as opposed to acknowledging it and doing something about it).
Daniel Gultsch doesn't guarantee XMPP's existence. Yes, he brought it back to the public's attention by driving things and when he leaves, things may go on more slowly until somebody else picks up the slack. Daniel wasn't the first to move XMPP along and even today he maintains just one (if popular) part of the ecosystem, with the standard work a side effect.
Will Matrix survive such a drought when NewVector takes a turn for the worse? I'm not so sure about that.
I suppose another question is if anything in Matrix is worth preserving when NewVector goes under. People can simply move to the next big thing, and maybe that's just how it should work. But in that case, why bother with federation? Centralized designs have the same single-vendor issue (see Signal) but without the overhead.
@patrick @aspie4K I mean, did you read "This week in Matrix"? https://matrix.org/blog/category/this-week-in-matrix
That's (at least) at the level of what Red Hat/GNOME does: https://thisweek.gnome.org/
@letterus if as you yourself just said Matrix is one company making everything, wouldn't that be a point against Matrix being some kind of community lead ecosystem? Furthermore your statement about XMPP make it seem like it has much more claim to being an open ecosystem as opposed to one company just making everything.