Matrix is a decentralised encrypted chat protocol on which you could build something like Zulip, except decentralised and end-to-end encrypted.
Element is the actual app being trialled here, which feels more like Slack and/or Signal than Zulip. The point is that you get something you can selfhost while also interoperating with other deployments… while also encrypting the data end-to-end with Signal protocol.
I'm sure you could do some of Zulip's features on top of Matrix.
But for what it's worth, as Zulip's lead developer, every time I'm looked at whether we could have built Zulip on top of Matrix, it just feels impossible to me. And a big part of it is the architectural decisions Matrix made to support a decentralized E2EE social network, which are not required for a federated, self-contained chat system like Zulip or Slack. Permissions enforcement, performance, and lots of other details really benefit from the more focused goal, where we've explicitly decided we're not building a generic distributed network architecture and are not competing with WhatsApp.
That said, I think it's great that we have multiple OSS chat systems with different strategies that are targeting different collections of niches!
I will never understand why so many organizations entrusted the communications fabric of their organization to Microsoft and SalesForce Cloud services over the last decade. If an organization can succeed in escaping Teams or Slack to Element/Matrix, that's great, even if it's a use case where Zulip would be a better end-user experience for their requirements.
Federation can feel like "just a feature" but the E2E encryption (also in group chats) is a reason for Matrix to exist and a big reason why it's so slow.
"Slow" in what sense? Development? Because I self host a Conduit server and I don't ever notice messages being slow. It would be hard to notice anyway, as in a group chat people usually take some time to type in their responses.
The sync between large groups used to be slow because of amount of data, but Element X and "sliding windows" were rolled out to help with it.
AFAIK, the public Matrix server used to be slow because of a heavy load (I think), but on my self-hosted instance that's not a problem at all.
The experience of using Matrix involves a lot of sluggishness at various points in the client - waiting to decrypt messages or properly sync keys, waiting to join a room or for room search to load - these are the things that have been salient to me using multiple matrix clients with a freshly-spun-up server within the past month.
I hope that at some point a focus of the Matrix project will become why this isn’t being done. A better developer experience would supercharge the ecosystem, IMO.
Matrix should be the default for anyone building a chat app, but for some reason it’s not.
Speaking as the CEO/CTO of Element... the classic Element apps on mobile were buggy, thanks to being a ~10 year old codebase with no shared code between platforms and effectively the 1st generation Matrix client. Which is why we replaced them over the last few years with Element X, with all the heavy lifting shared between iOS & Android via matrix-rust-sdk (effectively a 3rd gen Matrix SDK).
That said, 70% of our users haven't got the memo yet - we'll do a hard-upgrade when the remaining missing features in Element X (Spaces & Threads) are fully out of Labs.
Meanwhile, Element Web is lagging behind Element X - but we're now in the middle of an incremental in-place upgrade (not a big-bang rewrite, thank goodness) to use matrix-rust-sdk - see our talk from FOSDEM last Sunday for the details: https://fosdem.org/2026/schedule/event/DZJVTS-an-element-web...
> That said, 70% of our users haven't got the memo yet - we'll do a hard-upgrade when the remaining missing features in Element X (Spaces & Threads) are fully out of Labs.
This isn't users not getting the memo yet, this is users being faced with an unfortunate choice between a buggy, slow client and a new client that doesn't implement important functionality like Spaces and Threads.
Can i ask why is Element Classic even available on the Google Play Store? If you want people to move away from this?
I've only started my Matrix journey, in the form of writing bots using the matrix Python library. I'm keeping my fingers crossed, as the Matrix protocol could be really impactful.
I’m excited to watch the talk. I’m generally critical of Matrix, but that’s because I want it to succeed. Lately I find you’ve been doing a lot of things right, so I hope you keep going!
They are different, and the biggest reason is (I suspect) that a Zulip workspace is self-contained while a Matrix server is able to federate with other Matrix servers.
Other European institutions are also adopting Matrix, so federation may turn out to be an important feature.
Matrix has threads. So does discord, but discords UI around them basically renders them functionally useless.
Anyway, the first goal listed in this project was to move to European sovereign solutions so Zulip failed at the first hurdle.
Given the (lack of) speed of European bureaucracy, this is likely more a reaction to the US sanctioning the ICC than the more recent Greenland saber rattling, but you'll probably see more of this in the future.
I wouldn't say Discord threads are useless - I do wish the UI made them more obvious, but I'm in many discord chats that use threads all the time.
Matrix has threads in a sense, but in this very thread the project lead is talking about how the new, ostensibly less buggy and more performant flagship client does not yet fully support them.
Element Software SARL and Element Software GmbH however are not. In practice I believe it's Element Software GmbH providing the European Commission deployment of ESS. (Both are owned by the UK topco, but at the current rate we might flip one of them to be the topco instead).
The Declaration for European Digital Sovereignty defined digital sovereignty as the EU and its Member States' ability to act autonomously and to freely choose their own solutions, while reaping the benefits of collaboration with global partners, when possible. The UK is not the EU or a member state.
Part of Russia is in Europe. Do you believe Russian products were considered?
The EU's definition of digital sovereignty included collaboration with global partners. It is obvious why UK companies could be considered more reliable global partners than US or Russian companies. A muddled concept of European was not needed to explain it. If where an open source solution was developed mattered even.
> This is happening for AI leaders across every industry, and the pressure to catch up is increasing.
> Enterprises are feeling the pressure to figure this out now, because the gap between early leaders and everyone else is growing fast.
> The question now isn’t whether AI will change how work gets done, but how quickly your organization can turn agents into a real advantage.
FOMO at its finnest. "Quick before you're left behind for good this time!"
The idea itself has sensibility. It is the kind of AI application I've been pitching to companies, though without going all in on agents. Though I think it would be foollish for any CEO to build this on top of OpenAi, instead of a self-hosted model, and also train the model for them. You're just externalizing your internal knowledge this way.
I've used to do something like this, on a smaller scale and dubbed it "organization as code". As long as you have good enough providers for Terraform/Pulumi you can declaratively specify a lot of the interconnected stuff in a company.
I built this around GitHub as the indentity provider as my interest was declaratively defining repository access control, while also being able to use users public ssh keys to (re)provision services to get them access automatically.
I've done the same thing and I would not call it anywhere near org-as-code either. An organization is much more than a list of responsibilities, people, and compliance requirements.
For the latter, we already have policy-as-code tooling that actually works.
Might be a second language thing. Organization for me is stronger related to the root word organize; label, classify, cluster, etc. than something pertaining to processes and procedures.
I can tell you from years of first hand experience, it certainly makes you far dumber. The biggest factor is loss of memory. Most stoners (myself included) have terrible memory. I used to have near photographic memory, where I’d only have to read things once to remember them nearly verbatim for months after. I genuinely wonder at times where I’d be in my life if I had never got sucked into it years ago. I’m doing great despite the problems it’s caused me, but who knows where I’d be without it.
To be fair, over the past year I've scaled my cannabis use way back and my memory is definitely better. But only like 30% better, which is nice but not a night and day difference. (Memory issues were not why I stopped.)
On the other hand, there's a certain creative groove that's a lot harder to get into now. So there's a tradeoff.
Just one guess, but a lot of memory formation happens during various sleep cycle(s). Persistent cannabis use is commonly associated with lack of dreaming, which suggests it interferes with normal sleep cycles.
I was experiencing lack of dreaming after some time of daily use, and was able to regain my ability to consistently dream (and remember them) by simply ceasing cannabis use by 5pm each day. I don't experience noticeable problems with memory, but YMMV.
A few of my friends are chronic users and have a characteristic that when talking to them even when not high, there is a noticeable pause. It's like when they do those news interviews from someone around the world via satellite. There is the question and then a pause while the person just stares blankly for second, then recognition of the question and the answer. Have you ever experienced this yourself or notice it in heavy users.
As a chronic user, this is something I experience while stoned, but not while sober. I have a lot of behaviors when sober that people tend to classify as ADHD-like and I suspect it is related to this pause. Normally, when sober, I have a tough time staying focused on one thing. I typically have many projects going at once - half of them forgotten. You should see me cleaning my house - the whole thing will look torn apart as I jump from one area to the next, until the very end.
But when stoned, during that pause, I'm finishing with processing another thought that is already in my head. I find it harder to context-switch immediately when stoned. This is very different from my normal experience when sober, where my brain is very "flighty." The other time I notice this type of pause is when I've entered a "flow state," e.g. when deep in a programming project.
Sometimes I can leverage this "focus" into productivity when stoned, but then I am often equally likely to get focused on the wrong thing.
That said, it's well established that marijuana use acutely reduces your reaction time.
> you are witnessing the practice of thinking about the consequences of words before commiting to speech.
It feels different than that though. Its not unusual for someone to take beat before responding, but in that case there are facial and body cues that they heard you and are just getting their thoughts and words in order. But in the case of users, there is an expressionless pause even before that. More like blank pause, then facial cues of thinking, then an answer. That's why I used the 'via satellite' analogy because its like it takes longer for them to even register that they have been spoken to.
I do this a lot. It’s usually a pretty embarrassing pause but usually happens when my mind blanks mid sentence or I start talking again after an interruption.
Then again, I’ve never been able to speak well. I feel like my thoughts go too fast for my mouth.
I would recommend you fiddle with the repeat penalty flags. I use local models often, and almost all I've tried needed that to prevent loops.
I'd also recommend dropping temperature down to 0. Any high temperature value feels like instructing the model "copy this homework from me but don't make it obvious".
i am on mullvad and accessing it fine. if you are on one of the default exit nodes, try switching it. i find the default nodes get blocked by a lot of sites, likely due to malicious behavior of other users.
This ColumnStore is very simple and just do table scans sequentially on every query. It doesn't support indexes and unique constraints. It is almost an append-only serialization file format, but with some columnar concepts.
Is it functionally comparable, discussion threads and all? Or is it much closer to something like Discord?
reply