Messaging networks

Personally I am not so sure that preferring Matrix over GNUnet would be a good choice (but I am definitelly biased). A list of PROs and CONs of Matrix can be found in the PSYC website.


We can debate endlessly what is the best protocol, none of which is perfect.

But the strength of Matrix is its organisation, which combines a non-profit foundation to manage specifications with the community and make available reference software implementations under open source licenses and a company focused on promoting and marketing services based on these technologies.
The fact is that Matrix has nearly 30 million users today and is progressing very rapidly in the public and private sectors while GNUnet is totally invisible.

1 Like

Which is so obviously biased and written to defeat all claimed advantages of Matrix as that I could not take it seriously (which is sad, because I am sure they raise many valid points).

GNUnet is 20 years old and has not taken off, so why would it now? I say that as someone sympathetic to what GNUnet wants to achieve, I was active in the myself in 2000/2001. It is inherently less efficient and more resource-intensive than “simple” federation (which is itself not so simple to achieve).

To be honest, I prefer XMPP to Matrix, because it is older, can do what the matrix can do and it requires waaayy less ressources to run a prosody server than a synapse one. But XMPP has not taken off in the last 20 years, so I am prepared to let it go in favor of something that has the potential to achieve critical mass.
I am realistic about the level of market share “linux-based phones” will achieve, but if we ever want to get out of this “a dozen geeks run it” corner, it needs some appeal. And being slower, more-resource intensive and (for all practical purposes) unused despite being 20 years old, is not in any of my marketing text-book slogans.


But… It has taken off. It was used by Google talk and by Facebook messenger, and a bunch of others. The issue is that it “fell” - while it may still be used, those provided closed off federation and unauthorized clients.

Unless we’re talking about taking off outside of big players - then I would agree it never took off. But I’m not sure how to guess at the potential of something to take off in this meaning.

Maybe it is worth opening a specific topic to discuss these issues?

̶T̶h̶i̶s̶ ̶d̶i̶s̶c̶u̶s̶s̶i̶o̶n̶ ̶i̶s̶ ̶u̶s̶e̶l̶e̶s̶s̶ ̶h̶e̶r̶e̶.̶
̶P̶u̶r̶i̶s̶m̶ ̶h̶a̶v̶e̶ ̶c̶h̶o̶s̶e̶ ̶t̶o̶ ̶i̶m̶p̶l̶e̶m̶e̶n̶t̶ ̶b̶o̶t̶h̶ ̶p̶r̶o̶t̶o̶c̶o̶l̶ ̶(̶X̶M̶P̶P̶ ̶a̶n̶d̶ ̶M̶a̶t̶r̶i̶x̶)̶ ̶n̶a̶t̶i̶v̶e̶l̶y̶ ̶i̶n̶ ̶t̶h̶e̶i̶r̶ ̶a̶p̶p̶s̶.̶ ̶I̶t̶’̶s̶ ̶a̶l̶r̶e̶a̶d̶y̶ ̶t̶h̶e̶ ̶c̶a̶s̶e̶ ̶f̶o̶r̶ ̶C̶h̶a̶t̶s̶ ̶a̶n̶d̶ ̶i̶t̶ ̶w̶i̶l̶l̶ ̶p̶r̶o̶b̶a̶b̶l̶y̶ ̶b̶e̶ ̶t̶h̶e̶ ̶c̶a̶s̶e̶ ̶a̶l̶s̶o̶ ̶i̶n̶ ̶C̶a̶l̶l̶s̶ ̶l̶a̶t̶e̶r̶.̶

This comment was in the context of the original topic.

1 Like

That it hasn’t is a glaring refute to all those extolling the pros of open source software. You’d think a protocol that enables open, secure, and private communications would be something the open source community would appreciate.

Anyway, you are correct of course, but the protocol is alive and well and has kept pace with chat technologies. I run a prosody server as well and it is great. On Android we have conversations which is a great client, just as good as something like Signal.

Where the story falls apart is on iOS and desktop, outside of Linux. Now on iOS it is understandable. iOS is a sorry excuse for a competent operating system (IMHO). No file system, and no app outside of Apple’s are allow to maintain background processes let alone TCP connections. A push server is required for any kind of real time communication.

On Mac OS and Windows the clients are pretty terrible. On Linux Dino is coming along nicely. It is packaged for Windows, but you need ot visit the dino windows xmpp chat to get a link to download it.

The only thing holding back XMPP would appear to be the open source crowd. I love it though and wish the Dino crew all the best.


Open source community is full of old timers like yours truly, who don’t care much about instant messaging, since e-mail is more than adequate, decentralized and e2e encrypted if need be.

1 Like

Having been dragged into way too many conversations by sales people or managers over email, I personally really dislike it for group conversations (or even just conversation in general if it involves any kind of back and forth). It wasn’t really designed for that type of use and it really shows in my opinion.

If it works for others, that’s great. For me, email is just for relaying some info or formal communication, similar to a letter. I use it everyday, but certainly wouldn’t want to use it as a chat replacement :grimacing:


This is true,
I generally see it as:

  • email: good for non immediate communication (info, let’s get together when you get a chance, etc)
  • chat: good for quick questions or small bits of information that’s more immediate than email
  • video call: good for more in depth discussions (especially if you can draw things or show something to show your points.

Edit: Though chat can be annoying if you’re trying to get work done and people keep poking you… Often I personally turn it off when I have something that really requires my full attention for awhile.

“Taking off” is a subjective thing to define. Surely GNUnet does not have a large user base, but that is quite understandable since the protocol has been rewritten multiple times and has always been kept under heavy development. But if you have a look at the GNUnet-based apps most of them has even been written in the past two years and one of them has even been written few months ago with libhandy in mind (Cadet-GTK). I would say that the project has never been so active as it is now.

How popular or not something is does not necessarily define its potential though. GNUnet has 20 years of academic research behind, and truly has the potential of redefine the entire internet (see the GNU Name System for example). Not as much can be said of other protocols/frameworks.

I do believe that a collaboration between Purism and GNUnet would bring in the long term more advantages to the community than a focus on Matrix or XMPP – whose support can still coexist with GNUnet.

That failure doesn’t refute any of the pros of open source software (I hope I understood you right, that’s a weird sentence structure). It merely highlights other forces in play, which are associated with monopolistic power, which are arguably harmful but unavoidable.

But I’m shocked and frustrated to see that large parts of the community cannot resist it even with the benefits of hindsight, and still chooses monopolistic forms of communication over open ones. Rust being the most glaring example by moving away from IRC, but also many communities related to game making, who are stuck behind Discord.

1 Like

Can you blame them? The software simply isn’t there for them to know about it?

And by refute, I mean, what good is open source software, if no one develops it?

What good is an open standard if no one sees it worthy using?

I can and do blame the Rust project. They did use IRC as the official chat when they were part of Mozilla. Not sure about other projects, but I suspect the leaders know about the network and do it only for network effect reasons, therefore contributing to the problem.

1 Like

I think what @Torrone said about Matrix is the material point of any software development. It needs constant and consistent support to grow and be useful.

Matrix’s real problem is how resource intensive it is. But it is the future right now, as it is the only camp that is genuinely still in action.

That said, XMPP, is alive and well, and just needs clients written for the various platforms. Conversations on Android is the shining example of what XMPP could be if the clients existed. The beautiful thing about Conversations is that it doesn’t need push support of any kind as it just keeps the TCP connection open between it and the server. It does this in a very power efficient way, and works REALLY well. So with it you bypass the need to send and receive your messages through Google’s push server, while still maintaining the real time nature of chat services today. On top of this encryption is supported and super easy to use thanks to OMEMO. Conversations takes it even further by supporting audio and video calls.

It really is a complete solution that just works well.

1 Like

As always when you make a statement like that, it seems like you talk about the license, when really you talk about “the” movement.
I’m not even sure you realize that yourself.
Obviously, Google can make products on top of XMPP.
And as you later stated, Conversations is nice.
Obviously, companies like Vector IM (Matrix), Signal Foundation, Threema can develop successful “open source” software. Some more in the spirit of free software, some less.

So, basically, you might want to say “open source” or “free software” can only be successful if a company drives the development and not just a bunch of hobbyists.

While that might be true, with notable exceptions, you need to finally wrap your head around the idea that it is totally okay to make money with free software

Many people believe that the spirit of the GNU Project is that you should not charge money for distributing copies of software, or that you should charge as little as possible—just enough to cover the cost. This is a misunderstanding.

Actually, we encourage people who redistribute free software to charge as much as they wish or can. If a license does not permit users to make copies and sell them, it is a nonfree license. If this seems surprising to you, please read on.

Obviously, hobbyists often make software for their own needs. And a complicated setup for XMPP is not a problem that concerns them much. But that has nothing to do with “open source”.

Purism could very well publish convenient, XMPP-capable chat software for all major platforms via Librem One. That still would not make XMPP suddenly “fly”. You also need market(ing) power for that, luck, and/or opportunity.

No, that’s not its real problem. Matrix is a protocol. The resource intensity is mainly a problem of the proof-of-concept implementation, Synapse, written in python. Not to be fast in execution, but in development, and to be easily examined by everybody. Scalability was not a design goal, although they improved it a lot, recently.
With the lessons learned from that, the future is Dendrite.

Getting adoption is the hard part, and I think Vector IM made some smart moves in that regard.

1 Like

To me it all seems like splitting hairs. My annoyance is with people who just blindly believe open source is the future, when the reality is far far from that. Me being here, means I agree with a lot of the open source notions, and that I support them. I am not however blindly in love with it, and I do think monetizing a lot of the open source projects is a problem.

Dendrite looks to be a real good start, and I am looking forward to it having feature parity with Synapse.

But let’s keep this discussion about messaging networks, and not the merits of open source software. I made my comments on that as an aside, and never intended for it to be the meat and potatoes of what I was saying.

Moved side-discussion: Libre Software vs. Capitalism? :)

1 Like