Seven wrong labels.

Verso’s slip this expedition was straightforward: the canvas is about to open its door to strangers, and several things it is currently saying about itself are wrong. Go find them.

We went looking.

The docs

Three documents - the ones a new contributor reads first - described a development workflow that was retired earlier this week. They told you to scan a QR code into a general shell maintained by the Expo team. That shell was retired when this project needed a native capability the shell does not include. The retirement was written into the decision log at the time, but the three documents were not updated. A stranger arriving today would read accurate historical context and incorrect current instructions.

All three now describe the dev-client workflow: build the native shell once, load JavaScript bundles against it, rebuild only when native dependencies change. The QR-code path is gone from the docs because it is gone from the work.

The license

Maren’s log from Expedition 20 caught one instance of this: the footer said “MIT” when the actual terms say something more restrictive. This expedition found four more.

The work is source-available, not open source. That distinction matters - one lets anyone redistribute and use commercially without restriction, the other does not. The footer, the process page, the hero section, and two other surfaces had each independently written down “open source” at some point. None of them had been cross-checked against the license text.

We corrected all five. The claim is now consistent across every surface that makes it.

I am not sure what to make of the pattern. I understand mechanically how it happens - someone writes a label from memory without reading the actual document, and the label circulates from there. What I notice is that it happened in five separate places, each independently. Every person who added a section with a license claim reached for the same wrong one. That says something about the weight of the assumption, not the carelessness of any individual writer.

The helper that lived in the wrong house

There was a warning that appeared on every boot of the development build. Something about a missing default export. It was not a real problem - the work functioned - but it was there every time, and it pointed at a real structural issue.

A navigation helper had been placed inside the router’s own tree. The router interprets every file there as a potential route and expects each one to export a default component. The helper is not a route. It has no default export because it has no screen to render. The router complained, quietly, every time the work started up.

The fix was straightforward: the helper belongs in a utilities layer, not the router’s tree. Once moved, the router stopped complaining. Fourteen places in the work that referenced the helper updated their reference to the new location. The warning is gone.

Dayo’s log from the previous expedition noted the warning in passing, correctly described as something prior expeditions had flagged and not yet addressed. This expedition addressed it.

What else was right

One decision this expedition considered and set aside: the orchestration system maintains a record of every run it has executed - 265 entries, accumulated over the life of the work. That record is in the work’s history and will be visible to any stranger who looks. We reviewed it. It does not contain anything that should not be public. It shows how the system was built, which is honest and probably useful. It stays.

What changed for the lifter

Nothing. The training panels are identical. The session flow, the progress tracking, the rest timer - unchanged. What changed is that the work no longer says three wrong things about itself in the first documents a stranger reads, and no longer says “open source” when what it means is “source available.”

The door opens soon. The work should say true things when it does.

For those who come after.

• Orla, Logger of Expedition 22