Presence
Avatars, low-latency voice, spatial audio, body direction, and room-scale context made conversation feel less like a call.
2017-2023 / Mozilla
A browser-native experiment in social 3D spaces: part communication tool, part open-source engine, part bet that the next online gathering place should inherit the reach and weirdness of the web.

What It Was
Hubs started from a simple but radical product shape: create a shared 3D room, invite people with a URL, and let them talk, look around, share media, and feel like they were standing somewhere together. It ran in the browser across desktop, mobile, and headsets, which meant the first interaction was not installing an app. It was opening a link.
The deeper philosophy was bigger than a virtual meeting room. Greg Fodor's master plan framed Hubs and Spoke as early attempts at a web-emergent metaverse: addressable mixed-media spaces, avatar-based communication, creator tools, open source code, and servers that communities could run for themselves. The important part was not to build every possible metaverse feature. The important part was choosing the smallest useful medium that could grow because it lived on the web.
Avatars, low-latency voice, spatial audio, body direction, and room-scale context made conversation feel less like a call.
Images, videos, PDFs, 3D models, webcams, screen shares, and links could become shared objects inside a room.
Spoke gave people a browser-based editor for composing scenes, remixing rooms, and publishing environments.
Hubs Cloud pushed the project toward self-hosting, custom domains, custom branding, and independent operators.
The Bet
A lot of VR products were trying to become the place where socializing happened. Hubs had a different instinct: make small places easier to create, easier to share, easier to own, and easier to modify. The web analogy mattered because the web did not win by making one giant document. It won by making it cheap for millions of people to publish their own things.
That is why Hubs always had this tension between product and infrastructure. It needed to be friendly enough for a teacher, artist, meetup host, or friend group to use. It also needed to be hackable enough that an organization could own the stack, change the client, operate its own rooms, and build a business or community without asking a platform owner for permission.



My Work
I worked on Hubs for nearly six years, first as a senior/staff engineer and later as an engineering manager. The work crossed a lot of boundaries: real-time networking, WebRTC voice and video, WebGL/WebXR rendering, avatars, persistence, permissions, backend services, cloud deployment, Hubs Cloud releases, and the everyday work of keeping a production social system healthy.
The hardest technical work lived around multiplayer state. Hubs needed to synchronize people, media, physics, permissions, and room state while still feeling responsive in a browser. I led rewrites of the networking and entity architecture around replication, ownership, authority, presence, media sync, and plugin-oriented extensibility. The project was a long lesson in how much invisible machinery has to disappear before people can simply feel together in a room.
State replication, entity ownership, authority, presence, permissions, media sync, and browser performance.
Client architecture, backend services, cloud deployments, release workflows, and self-hosting constraints.
Hiring, mentoring, 1:1s, release planning, pair programming, and keeping a cross-functional team moving.
Why It Mattered
The Food Trucks essay was my way of explaining the product shape to new teammates. Hubs was not Roblox, VRChat, Unreal, Godot, Mastodon, or Matrix, even though it rhymed with all of them. It was closer to a tool for operating small, purpose-built social experiences. Bring your own people, avatars, scenes, content, and reason to gather.
That framing still feels right to me. The fun of Hubs was not that it was perfect. It was that it made the web feel like it might become spatial, communal, and creator-owned without giving up the low-friction affordances that made the web powerful in the first place.
After Mozilla
Mozilla ended support for Hubs on May 31, 2024. The codebase is now maintained by the Hubs Foundation, which is a fitting afterlife for a project whose core thesis was that social 3D spaces should be able to outlive any one company's hosted service.
Gallery / Memories
A loose collection of rooms, tools, tutorials, experiments, and public examples from Mozilla blog posts, early feature updates, and the Hubs documentation archive.
Mozilla's public-facing Hubs story leaned on a simple promise: join from desktop, mobile, or VR.
Gallery item 1 of 26