# The measured pilot: volunteering

runward claims to make a repository cheaper to take over. That has never been measured on a repository the maintainer did not write. Pre-registered protocol, failure criterion written in advance, result published either way.

runward claims its artifacts make a repository materially cheaper for a stranger to take over. **That claim has never been measured** on a repository the maintainer did not write. Until it is, it is a thesis rather than a result — and this page exists to make it fall one way or the other.

## What is already committed

The protocol is committed and dated **before any data exists**: [`docs/pilot-protocol.md`](https://github.com/stranxik/runward/blob/main/docs/pilot-protocol.md), with its [instruments](https://github.com/stranxik/runward/tree/main/docs/pilot) in the repository, empty. The git history is the proof, and that is the whole point: a protocol written afterwards bends to the numbers.

Three things are frozen in it:

- **The questionnaire** — sixteen questions an inheritor or an auditor actually asks (where the load-bearing decisions are, how you restart it, who answers for what, what evidence exists that the declared rules were applied). The **same** sheet serves before and after: a sheet that moves between the two measures makes the numbers incomparable.
- **Who scores** — the volunteering team's engineer, never the runward maintainer. It is the thesis's own logic applied to us: a verdict is opposable only when the judged party does not manufacture the judge.
- **The failure criterion**, written in advance: if the artifacts do not reduce the count of unanswered questions, or if they cost more operator hours than they save, **the pilot counts against the method and is published as such**.

## What it costs, and what it gives

**The cost first**: one engineer, up to 8 hours boxed, **twice** — once before, once after. That is the real price and it is not small. Nothing else is asked of you.

**What you keep**: a complete runward mission on your repository, free — the characterization, the retro-documented decisions, the evidence trail, the gate. MIT, yours, whatever the numbers say.

**What gets published**: both score sheets, the operator hours log, and the verdict. Named, or with your repository described only by size, stack and age if you prefer — you choose in the form, and that choice is not renegotiable afterwards.

## The repository it takes

One the **runward maintainer did not write**: that is the single condition without which the measurement measures nothing. Brownfield, no production stakes required. A repository with no model or agent in it is perfectly fine — runward gates delivery decisions, and ordinary software simply exercises fewer craft rules.

## Volunteering

[Open the candidate form](https://github.com/stranxik/runward/issues/new?template=pilot_candidate.yml)

It is a GitHub issue form, deliberately: runward operates no service and collects nothing, so the invitation runs on the same infrastructure as the rest of the project — a public repository. The questions are versioned beside the protocol, so they are pre-registered too: you read exactly what you will be asked before asking anything of us.

## See also

- [Concepts: the deterministic gate](https://runward.dev/docs/en/concepts/the-gate/)
- [Evidence & integrity](https://runward.dev/docs/en/concepts/evidence/)
- [What's next: the roadmap](https://runward.dev/docs/en/project/roadmap/)