After five years championing Matrix.org and Element as a Slack/Discord alternative, a dedicated user details a litany of failures – crippling performance issues, fragmented development, unreliable federation, and a catastrophic support breakdown – that forced a return to XMPP. This deep dive exposes the painful gap between Matrix's ideals and its operational reality.
For five years, one user bet big on Matrix.org and its flagship client, Element. Inspired by Mozilla's 2019 adoption and the promise of a truly open, federated alternative to Slack and Discord, they migrated communities, evangelized the platform, and endured its quirks. Today, they're leaving – and their detailed, disillusioned account serves as a stark warning about the state of the Matrix ecosystem.
The Broken Promise: Matrix.org emerged a decade ago aiming to be the SMTP of real-time communication – a federated protocol enabling seamless messaging across providers. While technically similar to XMPP (using JSON instead of XML and baking in WebRTC/E2EE), it garnered significant institutional backing, particularly in Europe (French Tchap, German government tools, Luxembourg's Luxchat4Gov). New Vector, the company behind Element, pushed development hard after spinning out from Amdocs funding.
"Encouraged by the hype, I doubled down... I wanted Matrix and Element to succeed. I wanted to love the platform."
Years of Frustration: Despite the hype and adoption, core user experience issues persisted and, according to this user, worsened:
- Performance Hell: Element clients (both legacy Electron and newer Element X) remained sluggish, especially on launch due to network synchronization. The matrix.org homeserver itself was slow, with noticeable message send/receive lag even in terminal clients like iamb (Rust-based). The web client reportedly took "almost 40 seconds" to load, consuming significant CPU/RAM.
- UX Quagmire: Cross-device verification for E2EE was notoriously buggy until recently. The platform exposed overly complex settings (roles, permissions) while lacking intuitive features users expect (threads, spaces in Element X).
- Development Chaos: Fragmentation plagued the ecosystem. New Vector cycled through tech stacks (Python/Node.js for Synapse, Go for Dendrite, Rust for client SDKs), abandoning projects and leaving community efforts like Conduit and Dendrite struggling. The author highlights stale dependencies in official Dendrite repositories.
- Dying Ecosystem: Third-party clients faced existential choices (fork Element X or maintain old codebases?). FUTO Circles shut down. The once-vibrant landscape of alternatives stagnated.
The Final Straw: Federation Failure:
In early July, the author discovered their primary community channel on the matrix.org homeserver had essentially vanished for them and other matrix.org users, while still appearing active to users on other servers. Attempts to rejoin failed with cryptic errors (403] No create event in auth events). Support tickets went unanswered. Investigation suggested the matrix.org server had lost critical room state data.
"It was as if the matrix.org homeserver had somehow lost the room, without communicating this to the rest of the network... So much for (overly complex) permission models and Matrix federation."
Why XMPP/IRC Won Back a Believer: Faced with unreliable federation, poor performance, a terrible web client, an unfinished native app (Element X), increasing spam, and nonexistent support, the author concluded Matrix.org is unfit for mainstream recommendation. They contrast Matrix's heavy resource demands (Dendrite needing 2-4 cores & 8GB RAM for a "comfortable" small deployment) with the proven efficiency and simplicity of XMPP servers like Ejabberd, built on battle-tested Erlang/OTP.
"Matrix ends up being worse not just in terms of user experience, but also from an administrative and cost perspective... I cannot comprehend why Matrix.org chose to use Python and Node.js of all things, instead of Erlang, or more recently, Elixir."
A Future Foreclosed? While acknowledging Matrix's ideals and continued institutional adoption (especially in European governments tolerant of poor UX for sovereignty), the author sees little hope for mainstream appeal. The platform feels built "for compliance departments and bureaucrats, not communities." Their journey ends where it began decades prior: on the reliable, simple, and interoperable grounds of XMPP and IRC.
"Ideals alone don’t make for usable tools... I’ll be elsewhere, quietly rooting for a comeback, but no longer waiting for it."
Source: Giving Up on Element and Matrix.org (Author's personal blog, July 2025)

Comments
Please log in or register to join the discussion