Matrix-Libera IRC Bridge Temporary Shutdown, a Retrospective (2023)
libera.chat
libera.chat
matrix has other driving forces and incentives. there’s only so much time they can spend on things like individual bridges.
meanwhile, they devote developers to writing two different homeserver implementations in parallel. or writing an experimental p2p homeserver - or the three guys working on thirdroom. ugh
i just hope they realize what’s important. people just want to chat on a distributed platform that isn’t irc - make it as simple and fun as possible. that is your entire mission.
no metaverse, no experimental backend shit, no securitygasm cryptography - nobody is buying drugs on matrix. this isn’t signal or whatsapp, this is discord for tech dorks.
just focus on making GROUP chatting good, simple, and fun. the ux just utterly sucks right now.
it’s the difference between scrolling through a menu on an ipod versus a self checkout kiosk - the kiosk just FEELS bad, simple human revulsion. the element interface offers the same experience.
and we _still_ don’t have custom emojis or selfserve moderation, even though matth himself promised them to us 2 YEARS ago: https://news.ycombinator.com/item?id=33014245
element is _the product_, and it hasn’t moved a single inch in years from my point of view.
However, regarding your last point there seems to be an active focus on improving element and their recent post about elementX [1].
I agree with your other points though, there are a lot of experiments that often lead to nowhere (e.g. P2P is basically abandoned AFAIK).
I'm not using it right now because I switched my server to authenticated media and Cinny doesn't seem to have that implemented (yet), but that's mostly because I'm ahead of the curve on that change.
It's a bit unfortunate to see matrix still not there yet, but on the other hand I'm still hopeful for the future. I think ElementX is the result of realizing that an excellent chat experience is what is needed. I also think the matrix foundation governing board elections [0] show that the community agrees that a rock solid chat experience is what is needed. At least for the individual members category, none of the candidates who were elected had a platform statement advocating for more work on the experimental matrix features. Sumner (who I think had the highest support by a good margin, although I can't find the exact numbers right now) specifically stated that experimental matrix features should not receive further investment until the baseline features are acceptable.
I realize that getting funding is a beast of its own and that is probably why some of the experimental things took focus at times so I can't fault that, but I really do want matrix to be something I can eventually recommend to all my family and friends instead of whatsapp or discord. I think it is slowly heading that way..
[0] https://matrix.org/blog/2024/06/election-results/
[1] https://matrix.org/foundation/governing-board-elections/#nom...
This Matrix case in my opinion is either lack of talent or intentional. One should never discard the second option when considering how governments tend to interfer in such tools and Matrix is quite sus on the funding side.
In 5 years we've seen little to no remarkable progress. I'm not recommending this platform any further to anyone, other options have delivered what this one had been promissing for years. Time to move on.
thanks for understanding…
As I've said before: won't be using your platform nor recommend it.
Are you telling me that with the multi-millions of dollars you get from law enforcement, you cannot delegate a manager to, as a minor part time duty, oversee the free work given to you? Especially when the various Element apps have been in an effective maintenance mode for at minimum the last year? This sounds like either Element is run very poorly and despite the massive funds rolling in can barely maintain themselves, or there is a distinct distain for external work that does not generate immediate revenue for the company.
Trust me, I understand full well what you’re saying. You’re maximizing profits with no care for the “freeloaders”. That’s fine, you can do that, but don’t act like I have to care.
Correct. We are not profitable, and frantically trying to get sustainable by focusing on things which generate money so we can keep paying our salaries, rather than taking on maintenance of new “free” contributed features when we’re still focused on the core. The reason we don’t have more $ is because we gave away everything as Apache licensed FOSS, and turned out most people were happy to use that gratis rather than paying for anything. The most likely manager to oversee custom emoji was laid off last year due to lack of $.
Fwiw, we make almost no money from law enforcement.
> there is a distinct distain for external work that does not generate immediate revenue for the company.
No. We really appreciate the various custom emoji PRs, and I personally feel awful they are stuck. At this point the most likely way to get them merged is if i do so myself.
> You’re maximizing profits with no care for the “freeloaders”.
No. We are not profitable. We are trying to stop losing money, so we might actually be sustainable.
I know custom emojis sound dumb, but they are basically the differentiation factor for how Discord gets their billions. People almost literally pay hundreds of dollars for Nitro just to get their emojis/reactions. Other than bigger uploads, that's all the lower Nitro tier does.
Some people just laugh in my digital face when I suggest Matrix, because they will never, ever, countenance losing their emojis. I don't know why they care that much, but they absolutely do.
It feels a bit like trying to sell a car without Bluetooth. You might think it's a trivial frill and a distraction from the best engine ever, but the interest will be lukewarm at the very best. I bought a car without Bluetooth but then I fitted my own adapter. Even that isn't possible here.
However, right now we're not focusing on building an encrypted Discord alternative but instead a self-hosted WhatsApp (or Teams) replacement for governments: https://element.io/sectors, on the basis that they are the ones paying. Sure, we could have gone after Discord ~5 years ago and launched our own Nitro equivalent, and perhaps we should have. But instead we observed that governments REALLY want to run their own encrypted interoperable comms systems, and we thought that having Matrix used as the backbone for public sector communication would be a good way to prove and fund it sustainably. Ideally we could then use the public sector as a launchpad into other areas - a bit like how email & the web and even the internet spread from DARPA / NSFNet / academic etc to the rest of the world. After all, what better endorsement for Matrix than someone like NATO using it for comms?
As a result, custom emoji are stuck in the middle of the todo list still.
Some of the other features that we've worked on instead have been:
* Making encryption stable. As hilarious as https://www.reddit.com/r/elementchat/comments/1evz3kk/unable... is.
* Make encryption secure - i.e. migrating from the old C/C++ libolm implementation to the rust vodozemac implementation.
* Instant login/sync/launch - i.e. an entirely new sync mechanism.
* Rather than developing 4 different client stacks (js-sdk, ios-sdk, android-sdk and rust-sdk), converge on a single one: matrix-rust-sdk, and make it excellent.
* Rework the core UX to make encryption invisible (rather than full of confusing unactionable warnings and verification nags etc).
* Native Matrix-encrypted scalable voip/video calling.
* Migrating to OpenID Connect for auth, so getting 2FA/MFA etc
* Public sector enterprise features: antivirus, regulatory compliance, secure border gateways, cross-domain gateways, active directory sync, SCIM sync, kubernetes operators, etc. etc.
Now, the plan is to use the govtech business to get back to funding mainstream Matrix uptake. But first we need to be able to fund ourselves to work on it.
Honestly though, I would contest the logic here:
> bit like how email & the web and even the internet spread from DARPA / NSFNet / academic etc to the rest of the world.
It spread because it was useful and revolutionary, not because DARPA specifically used it. Yes there were network effects, but the network effect overlap today between "NATO" and "a friend group hangout" is pretty minimal.
> what better endorsement for Matrix than someone like NATO using it for comms?
Honestly, if all I knew was NATO funded and used it, I'd assume it was enterprise-grade cost-plus shitware 15 years behind the curve, and with zero organisational interest, if not outright hostility, to it being used by anyone without a governmental chequebook and service contract in hand, and steer well clear.
It's OK to write software for NATO and governments for the foreseeable future, but the website still says "communicate with friends, family, communities and co-workers", so you can see why people might think it's still targeted at them and then wonder why it doesn't even provide the baseline of human-interest features people have been getting from other platforms for about a decade.
Not that it helps pays the bills to court Brian Everyman and his Warhammer group chat, I do realise that.
However, we're at where we are now, and we're committed as to the current path as a way to get sustainable and then go back to supporting mainstream users (including Brian and his Warhammer clan). Perhaps a profitable govtech business can subsidise mainstream FOSS Matrix without having to do Nitro style value gates (which are tough to enforce anyway in a decentralised world).
> but the website still says "communicate with friends, family, communities and co-workers"
That was the messaging prior to early 2023, when we shifted gear to "get sustainable by focusing on govtech". Which is why the frontpage of the website currently doesn't say anything about friends/family/communities, sadly.
I just had to fish around to find out where you saw the old wording - i'm guessing it's in the footer for element.io/blog? ...which is a snafu due to the blog being Ghost and the website being Webflow.
As for the wording, it's on the front page of https://matrix.org, just below the fold. Scroll down once and it's centre of the page.
The fold below that is "A cozy, safe, supercharged place for your community" and the image says "Kayak club".
Which, yes, is Matrix not Element, but it recommends Element specifically in the click through.
In terms of why consumer FOSS software often sucks:
* Good UX requires top down product discipline: cathedral model, not bazaar.
* Product folks are less likely to contribute to FOSS, especially in their spare time
* Therefore they need to do it as their dayjob.
* Running a sustainable FOSS company is hard.
Aren't you the matrix.org project lead though? (if not, maybe your bio needs an update).
Those governments might be paying your bills for now, but you've lost the support from the open source community on the way to get there and soon be forgotten by everyone. I just hope one day we won't have the sad news of discovering that Matrix injected backdoors from the Israeli government to receive funding.
Nevertheless, I wish you all the best.
In terms of Matrix having lost the support from the open source community because Element hasn't implemented custom emoji yet... time will tell. In the end I suspect the open source community would be even more upset if Element had run out of money and exploded (although you never know; haters gonna hate :)
If you are willing to try out new things other than XMPP, I'd recommend looking at the NOSTR protocol and more specifically at the 0xchat client. They provide private message with E2EE but the biggest advantage IMHO is that you get connected to a huge variety of other services and people using the same protocol.
I find that better than building the connections on what are essentially closed gardens or, at best, isolated islands. Over there exist hundreds of servers (relays) run by volunteers where new features are proposed by users and actually implemented (success or failure depends truly on themselves).
Other good features being delivered on NOSTR are for example the anonymous IPFS (blossom files) where you can host complete static sites across the relays. There is truly plenty to discover there and to build, feels like the 90s and has been fun.
Not a great first impression.
"Recommending" anything is pretty much worthless. Stand them in front of a done thing that is already set up, that's worth a hundred times more.
I would like to do the same, but with the possibility of scaling it up a bit.
https://github.com/spantaleev/matrix-docker-ansible-deploy/b...
Later, though, I stood up my own Matrix server, this time federated, and that only took maybe an hour or two. :)
I did not use the community Ansible playbook that someone else in this thread suggested, because I enjoyed instead figuring out how the individual pieces work together, but I tried it much later, and it seems to work very nicely, so I too can recommend it.
You'd think they could dig into their hundreds of millions of donated dollars they don't even need and help Matrix and themselves but apparently not.
I still think they'd get more value out of slinging you guys a small-to-them wodge of cash to run a community homeserver than most other things they spend money on.
(Login with your matrix.org account)
The web app has been evolving too, but less dramatically - we need to enable instant sync there.
In terms of custom emoji and selfserve moderation: both are the direct casualties of having to focus exclusively on servicing the government customers who actually pay for Element development. Since the end of 2022 we have had to completely change focus to first generating $ in order to be sustainable, and that meant parking everything but Element X, Web, Call and Synapse. The Libera mess is also effectively a symptom of that.
P2P, Dendrite, Thirdroom etc have all been shelved for over a year. (Weird to see so many year-stale comments on a year-stale blog post).
P2P and Thirdrooom aren't really news but Dendrite itself? :/
That's significant and not at all the expectations set by current state of docs.
https://github.com/matrix-org/dendrite
I thought it was just a slump and have been holding out on some pending changes. Sad.
I wish. I think this hasn't been the case for at least a year or two.
I very much disagree. E2EE with FOSS and decentralization is THE reason to use it over discord or other "more simple and fun" messanger.
>there should be one key that protects everything
That is more or less the case with the recovery key. https://element.io/help#encryption16
Yes they are, and they have since the beginning.
That's proper level of security, when people don't even suspect someone using messenger to buy drugs
Requiring a persistent TCP connection is simply a complete dealbreaker on the mobile phone platforms which make up, like, 90% of computers in existence. Fixing that would go a very long way, but it requires a lot more server-side storage since messages must be stored on the server until a client reconnects.
Data privacy requirements go along with server-side storage. If messages are stored for a few minutes that's probably okay, but that's too short to implement the feature. If they're stored for weeks, and especially if images are stored, then for legal reasons there must be a way to moderate them. On another front, at least the server only needs to store one copy of the message for as long as the slowest client might need it; it might as well just store the entire channel history for a fixed time. These considerations lead to a design like Slack/Discord/Mattermost/etc, and trying to make it decentralized leads to a design like Matrix.
You may think of IRC bouncers which don't have these issues, but notice the responsibility structure of an IRC bouncer is completely different. With a bouncer, the responsibility for storing the message lies with the reader, not with the network, so completely different considerations apply.
Sharing images and videos is just useful sometimes, even though 90% of the time it only serves to increase spam. And I don't get why users are extremely horny for emoji reactions, but they are. A platform can exist just fine without either of them - every social platform develops a unique personality in response to the constraints and nudges the platform imposes on users - HN doesn't have images, videos or emojis and that's a good thing for HN. It's the persistent connection thing that's the real killer, because it means the platform is simply unusable 90% of the time.
A version of IRC tailored to slower connections might look identical to (the non-binary segment of) Usenet. However, even Usenet has light moderation, that differs per server, to avoid being clogged with spam and binaries.
though even social channels work better with images. my family channel would be very limited without images. most of what we talk about is things we see, and can share images about.
emoji reactions are another matter. but given that it is possible to just use unicode emojis, i think that reactions reduce clutter.
the point is that usecases where media support is needed do exist.
Since then I just run 1-to-1 heisenbridge connections with my homeserver. It's not as fancy but it works reliably.
The idea of transparency being a "burden" is ridiculous. It implies that sharing information is a problem, which seems like an excuse to avoid accountability.
Free speech absolutism has its place, but so has diplomacy.
Umm libera itself allows "easy anonymous registration". You don't even need to have an account. You just join with a made-up name if you want to.
This is in fact one of the things I love about IRC. Nothing wrong with it but it feels wrong calling matrix out on that.
The archiving makes kinda sense for matrix' features of scrollback on demand. I believe there's an IRCv3 feature for it too. The bridge should have a provision to prevent users to see any backlog from before the moment they joined.
Ps I really hate those new "single puppet" third party bridges they recommend. Because they break nick colouring and also the actual nick of the user speaking is in a different place. Having each matrix user have their own IRC puppet is much nicer on the IRC side. Especially when there's more matrix than irc users.
The difference is that Libera relies on blocking anonymous connections abusive IP ranges (proxies, gratis VPN providers). When abusers connect through Matrix, Libera only had their username, which delayed spam mitigation.
> The bridge should have a provision to prevent users to see any backlog from before the moment they joined.
That's already how it's supposed to work, though it had some issues.
> Ps I really hate those new "single puppet" third party bridges they recommend. Because they break nick colouring and also the actual nick of the user speaking is in a different place.
I think everyone hates them (and I say this as someone who maintains a few of them). Some IRC clients have scripts to substitute nicks from relayed messages though.
Yeah but that's not Matrix' fault. They just offer the same ability as Libera: To create an account with no validation.
Putting pressure on Matrix server providers to stop anonymous connections when they themselves provide anonymous connections is a bit too hypocritical to me. If they have a problem with the practice then just block matrix connections (as they have done). But don't complain about the practice.
Personally I love the idea of anonymous connections like on IRC and also some Matrix servers. I'm a bit sick of every single service needing me to make an 'account' these days and needing all kinds of personal info from me.
I once dared ask for Javascript advice on solving a front-end problem without Node.js. That particular detail was taken as a personal insult. I’ve never experienced anything like it.