@mlabowicz apparently XMPP is still going strong, and is the "best" solution for secure e2ee federate-able chats.
Matrix is showing some growing pains, especially with the implicit federation models. Running a small user instance still requires a lot of resources to cache all the federated media that spammers are bringing to the party.
The impression that I've been getting is as if Matrix was started by someone naively jumping into protocol design without much background in it. It's as if you asked a younger 'full-stack developer' to make a chat application, and they were trying to check all the boxes for -isms of the time (just transport JSON over HTTP/HTTPS instead of a purpose-built protocol, Comet-style communication, RESTful APIs, etc), as well as ignoring the warning of creating an overly state-dependent protocol.
There's still the major infancy of server implementations with the Matrix world, meanwhile in the XMPP world there's a selection of very mature and performant codebases which are actually quietly running very large high-availability workloads, just like with ejabberd running the notifications system for the Nintendo Switch user network, chat in EVE Online and Fortnite (https://xmpp.org/uses/gaming/) as well as in WhatsApp and Zoom (https://xmpp.org/uses/instant-messaging/), just unfortunately that these platforms choose to not federate likely over control and moderation concerns.
I remember the shitshow of the earlier years of Google Talk and Facebook Chat trying to steer XMPP usage solely within their interests, as well as the different mindset of chat protocols back then that understandably soured people's outlook on the protocol. I've been following and using Matrix from back when the webclient used to be hosted under the domain vector.im (and had an account under that domain), as well as the later rebranding of the client to Riot, and then Element. Matrix was trying to pitch the idea of account portability, true decentralization (not just federation), and other lofty goals, and I was hopeful to see something come of it. Instead we just have another glorified webchat with a RESTful API that federates and implements Double Ratchet, and it doesn't have that big of a sell (in terms of protocol architecture) over XMPP.
Meanwhile there's a lot of modernization that's happened within XMPP in just the past 5 years, but because of people's soured experiences from Google Talk from long ago, or Cisco's abomination in the corporate world, everyone still writes it off based on decade old experiences, or keeps repeating several years old complaints about XMPP on mobile, when that's completely changed since. Either way, everyone keeps showering Matrix with attention, as if it invented the idea of federated messaging, just like how people do the same with Mastodon instead of acknowledging any other fediverse platforms.
Except this is the exact scenario as marketing Linux: XMPP is a building block to federated messaging, just as Linux is a building block to an operating system. You don't very often see a heavyweight company advertising the concept of Linux-based operating systems to the public. There are marketing efforts such as one group doing the branding project of "Snikket", and there's also plenty of material being pushed out there, but nearly everyone instantly writes it off based upon their past experiences. Nearly everyone just chases after hype, new projects, new VC-funded platforms, and seldom bother with mature and non-controversial projects. Once hype faded, it's always attention diversion to "the next big thing" even though much of what's come about in recent years is just reinventing the same thing over and over again.
@arcanicanis@mlabowicz@bobdobberson You raise some very good points. Yet Marketing is also a key aspect to make a successful standard, even in the FOSS world.