To be fair, IRC hasn't had an accurate RFC in decades. https://modern.ircdocs.horse/ is the closest to accurate client<->server protocol documentation, but is fully outside the IETF process.

Matrix is an open protocol[0]. It's not managed by the IETF, but it has an open process for submitting changes. See the Spec Change Proposal instructions[1].

[0]: https://spec.matrix.org/unstable/

[1]: https://spec.matrix.org/unstable/proposals/

That process seems to be "our company will decide what we want to do with our proposal"

yup, much as the W3C and IETF "companies" decide what they want to do with proposals to their standards bodies. The Matrix.org Foundation is a non-profit foundation too.

I don't know why but comparing Matrix.org Foundation with standardization organizations such as IETF seems just not right. Maybe it would be more correct to compare Matrix.org with XMPP Software Foundation?

I'd genuinely be interested to know what the difference is between something like IETF / IEEE / ITU / W3C and a non-profit which was created as a standards body for a specific standard (e.g. Matrix.org Foundation or XSF). Is it just that you're recognised as a peer by the other long-established standards bodies? Or is there a standards-body-for-standards-bodies somewhere?

I mean, yes? The IETF has additional cachet as having created the internet. ITU and IEEE are international orgs relied upon not only by companies, but by governments. The W3C isn't as important as it once was, because people stopped listening to them (WHATWG is the new org). But I would trust the IEEE and IETF like I would the ISO, and Matrix.org as far as I would trust Microsoft.

> But I would trust [...] Matrix.org as far as I would trust Microsoft.

Ouch. Did you read https://matrix.org/foundation or https://github.com/matrix-org/matrix-doc/blob/matthew/msc177...?

I'd agree that skepticism was warranted if we hadn't split out the Foundation and the protocol was de facto controlled by Element. But instead we made damn sure to create the Foundation independently and frankly protect it from being sabotaged by Element or any other commercial entity building on Matrix. To suggest otherwise is pretty insulting to the other Guardians/Directors whose only role is literally to oversee and ensure that the protocol isn't sabotaged by commercial entities.

This is very different from Microsoft's model.

I for one do not trust the ISO at all. They are a profit-seeking organization with an opaque standardizing process. That the ISO9660 standard (you might know it as the .iso file format) from 1988 is still locked behind a 140chf payment is a disgrace. And that won't even give you the full standard, because ISO loves doing this thing where a standard will reference 5 others, which themselves reference 5 others, etc...

IETF is one of the best standardizing organizations out there, I'll certainly give you that. They have fairly transparent process, and a really good track record when it comes to creating robust protocols.

Thing is, I don't see why Matrix.org would have any more or less "cachet" than WHATWG, or Khronos Group. In the end, the identity of the standardizing org doesn't really matter too much. What matters is that the incentives of the standardizing org are aligned with those of the community.

XMPP is actually managed by the IETF. The XSF just develops extensions to the protocol (but it's not the official steward of XMPP, confusing as the name is)

While this is true, basically everything useful to an end-user in XMPP is in one of those extensions!

Indeed; that's why I think this model has worked pretty well. You get a nice core protocol and then if you want fancy features that keep up to date with proprietary offerings that can be developed in a lighter weight way. Certainly not perfect, but I'm glad the IETF is in charge of the core spec and not the newer, less experienced, foundation (though at this point the XSF is well established too, but in the beginning it was the close relationship with the IETF that let it build that institutional knowledge).

Not all the reviewer are from the Element company.

Isn't that pretty much always the case though? A standard is only as good as it's most popular implementation.

I read a bit about that and I think it’s positive, but a single company controlling a standard and considering changes is not the same as a protocol.

https://matrix.org/foundation/ at the bottom of this page, you'll find the people who are actually in charge with accepting protocol changes (the Spec Core Team). It's true that most are part of New Vector, the for-profit company behind Element and Matrix, but Alexey Rusakov is not as far as I could tell. So there is at least one non-New Vector voice.

It honestly feels pretty likely that this is just a maturity thing - as more products are built around Matrix, the Spec Core team will likely become more diverse.

As I understand it, there's actually an explicit desire for more non-NV people to become involved with the SCT. There just aren't very many other organizations to fund it yet - SCT members need to eat too.

The situation here is that when we set up the SCT we deliberately picked a 50/50 mix of core Matrix team and community members. What we didn't anticipate is that the community members then were sufficiently sucked into Matrix that they were prepared to work on it fulltime, and a bunch joined Element as the only viable way to do so. Given the team is functioning pretty well and we're improving Matrix, it feels nuts to penalise people based on who they work for, hence the current blend.

This is one of the big issues with Matrix that doesn't get talked about enough: it's not managed by a real standards body. It was previously run by a company who kept trying to monetize the IP, then it was run by a few people who split off that company (and still kept trying to figure out how to monetize it and pay themselves a wage with it). I'm not against the devs being able to pay themselves, that's great, but the specs themselves shouldn't be run that way. Existing standards bodies have more experience, more legal protections for the users, and are just generally better at developing standards.

> It was previously run by a company who kept trying to monetize the IP, then it was run by a few people who split off that company (and still kept trying to figure out how to monetize it and pay themselves a wage with it).

This is just false. Speaking as co-founder and project lead for Matrix, it's been the same team all along since we began in 2013. We were incubated until 2017 in a company which never tried to monetize the protocol, and then we span out to set up Element (formerly New Vector) where we keep the lights on by selling Matrix hosting and support/consulting.

At the same time we set up The Matrix.org Foundation as a non-profit neutral standards body, with an independent board where the original founders are deliberately in the minority - and when we set it up, half of the spec core team were independent of Element too. (This changed as folks on the team opted to join Element so they could work on Matrix fulltime).

Rather than spreading FUD about Matrix, why not collaborate and work together? Or at least spend the energy on improving XMPP rather than negging us...

This topic seems to come up a lot, and I didn't know that, for example, the Matrix founders are in the minority on the Matrix Foundation board.

Is there a place with this history or the governance structure that people can link to the next time this comes up?

https://matrix.org/foundation is intended to be the single source of truth for this, and has bios of the foundation board members (or Guardians, as we call ourselves as 'board members' or 'directors' sounds boring :)

I am not advocating for XMPP, nor am I nagging you. You showed up to advocate Matrix on a thread about IRC, stop accusing me of things and collaborate yourself instead of reinventing the wheel in terms of specs and in terms of standards bodies. Everything I said about the company and foundation is accurate as far as I can tell. I am very glad there is a foundation, but it's still not okay: submit to an existing standards body and stop advertising on threads about IRC.

Negging != Nagging

https://en.m.wikipedia.org/wiki/Negging

Oops; either way, I was doing neither. They asked, I answered, then they had a hissy fit because it turns out they're apparently the CEO trying to score some users. Let's just end this line of discussion and keep it related to IRC and stop talking about Matrix.

I think you're being completely unreasonable. They didn't have a hissy fit.

You're trying to moderate discussion that you don't like, because it's not about IRC. Matrix is relevant in a discussion about IRC because it's a valid alternative. You can't police natural and useful discussion you don't like.

Fair enough, I shouldn't use such strong language. That being said, the CEO of a company making money off Matrix showed up to advertise, then complained about how I hadn't disclosed that I do some volunteer work at the XSF while simultaneously not disclosing that they have an entire company based on the thing they're asking about). That seems rather egregious and made me quite mad. My original answer I still think was perfectly reasonable too.

All that being said, you're right, I shouldn't get drawn in every time this person (whom I've just realized is the same person who goes by another name elsewhere) jumps in on every chat across the internet trying to advertise their product. My own fault for getting drawn in and using language like "hissy fit" that is, as you said, unreasonable. I'm sorry about that.

It's okay dude, don't worry about it.

We're all only human.

Thank you for working on XMPP

> Existing standards bodies [...] are just generally better at developing standards.

Speaking in a strictly personal capacity, I don't think that point can be taken for granted. The W3C's missteps with HTML5, and WHATWG's success, is a particularly notable failing of a dedicated standards body. The Rust programming language is also developed and codified outside of a traditional standards organization.

That's fair; I don't mean to suggest that all standards bodies are perfect all the time, or that they've never made mistakes :) just that it's better than making up a foundation spinoff from a company that doesn't have any experience and will re-invent the wheel yet again instead of submitting the standard to one of the existing standards bodies.

Matrix as a protocol is still in active development. Unlike "low-level" protocols such as HTTP, the Matrix protocol is much closer to end-user experience and so it must be able to move relatively quickly to remain competitive with proprietary systems. This generally does not fit into the process of standards bodies like the IETF very well.

There's a reason why eg. the WHATWG exists, basically.

Sure, it doesn't have to be the IETF, that was just an example. But even they tend to do this well (by eg. spinning off a smaller more agile standards body to keep up with building extensions more rapidly).

Also, rapid development has its own set of problems as we've seen with XMPP (where no two clients support the same set of features because new ones are being developed to keep up with various proprietary things all the time).

Anyways, point is, don't reinvent the wheel, I'm sure one of the standard bodies could have been a good fit if we needed this at all, but Matrix definitely isn't a good fit for this Freenode replacement and this is one of the reasons why (the other is that Freenode works just fine and the point is that this is a drop in replacement).

> real standards body

Who decides when a standards body is real?

I don’t think there’s a bright line or litmus test, but I judge for myself.

Single standard organizations created by the dev group are just naturally biased certain ways.

I don’t think there’s a single perfect body, but it’s definitely curious to me why someone would start a stand-alone foundation rather than use w3c, ietf, or others that seem to work pretty well.

I also judge the number of different clients and servers implementing the spec for whether it’s real or just a controlled intermediary. Sun/Oracle seemed pretty bad about this back in the day with JSR stuff although I haven’t paid attention in 10 years.

Projects can make their own choices and set up foundations if they are willing to fund and monetize, etc. But that stuff isn’t as robust as IRC or protocols that have independent standards bodies.

Seems like there are lots of different clients [0] and a couple of different servers [1].

The idea of a single dev team being VC funded means that there’s a drive to monetize. That’s bad pressure on a protocol and one that doesn’t exist for IRC and other protocols. There’s multiple members in the matrix foundation and that helps but there’s just not much diversity of interests.

Again, not the end of the world, and the project seems to me to be going to great lengths to collaborate. But I feel the same way as if Microsoft started up an independent foundation to drive development of the Skype protocol. Good on them, but it’s not as robust as an independent protocol.

[0] https://matrix.org/clients/ [1] https://en.wikipedia.org/wiki/Matrix_(protocol)

It just happens over many years and many successfully published and adopted standards. I don't have a good definition for you (although that's a really interesting thing to think about, maybe it's worth writing about) but I suspect most people know them when they see them.

Well, if this comment is not a textbook example of FUD, I don't know what is.

So only "standards bodies" should create standards, and you have to create standards to become one? Then standards bodies cannot come into being and therefore don't exist.

You're making giant leaps from what I said. I didn't say it's an absolute truth forever and always throughout the universe that you can't create new standards bodies. I said this was a bad place to do it.

Good standards bodies, I think, allow lots of bee SME members who can propose new standards.

I think IETF is a good example of how anyone can propose standards on a plethora of topics. When they were very young it was hard to trust but now they are decades old with hundreds of standards demonstrating how easy it is to propose and work on new standards. Maybe even the “gold standard” of standards bodies.

What are you even talking about lol

What part confused you? If I can clarify I will.

You just seem to be making many negative claims with little substance behind them.

I don't understand what you want; it's all true as far as I can tell, do you want a detailed technical breakdown of the protocol in an HN thread? This doesn't seem to be the place but the graph protocol mechanisms are pretty easily verified from their spec

These claims seem pretty basic and are easily read about in their Wikipedia article, https://en.wikipedia.org/wiki/Matrix_(protocol)