I recently worked with one of our Early Movers, who had reported a bug. When I asked how to reproduce it, they apologized for a "user error": they had right-clicked a button looking for a menu, instead of clicking it.

I don't think that was a user error, and accepting it would have meant missing a real opportunity. This is what co-design means for our Early Movers program.

Moss can now turn speech into text. It does this on your own computer, with an open speech model, and nothing you say is sent anywhere to be processed. That part is what people usually mean by "local AI", and it's good. But it's the smaller half of what we've done.

The larger half is the pattern: collaborative transcription. Instead of one machine listening to everyone, each person's computer transcribes that person's own voice and shares the text with the others. We've implemented this in Presence, our video call tool.

Moss has always kept your group's stuff on your own computer. But a few everyday things still went through the internet: finding the other people in your group, getting a new person into the group, and getting the tools the group uses.

Over the past month we've worked on all three. Together they mean a group of people on the same Wi-Fi can do all of that with no internet at all.

WebRTC is supposed to be peer-to-peer. In practice, it is not. Almost every WebRTC call today depends on servers that someone has to run, and the libraries that support WebRTC carry that centralized assumption into their design.

This post is about pulling WebRTC back toward peer-to-peer, starting with the signaling server. @lightningrodlabs/webrtc-peer gives a peer-to-peer app a managed WebRTC connection per peer, over whatever channel it already has. It ends with the plan for large calls with no server at all.

Over the past couple of weeks we upgraded all fifteen Lightningrod Labs tools in Moss to Holochain 0.7, and every one of them is now in the Moss 0.16 tool library: Notebooks, Emergence, KanDo, Talking Stickies, Gamez, Who's In, Spreadsheets, Converge, DataTub, Glass Bead Game, Files, Vines, FarmHack, Slate and Acorn.

The practical bit first. 0.7 is a new network, so the 0.7 version of a tool can't see anyone still on the 0.6 version, and your group should upgrade together. Most of the tools have an export/import so you can bring your stuff along; each tool's changelog in Moss says how.

We haven't posted anything here in 4 years. Yow. That's a bit of dereliction of duty. But it has been a long-time problem of mine that I like building so much more than talking about what I've built. So, this is a bit of a catch-up post, just to summarize what's happened here over these years. Thank goodness it's easy to ask an LLM to find the first commit of every repo!

As you can see below, our focus has landed on Moss. It started out with We and has just grown from there. But since we're a lab, there's been no shortage of experiments, with many more to come.

Here at Lightningrod Labs we are having a ball bringing together and delivering on experiments that increase the fertility of the Holochain ecosystem. I'm super pleased with the work that's been done all our code output, the sample hApps, libraries and tooling, but also on the human support "output" that's helped nurture the space.

This space is just another little bit of fertilizer in the space where you can expect to see musings, thoughts, invitations, and ideas out the very edge of where we are going. Enjoy!