the translation — the prediction traces
The prediction is what makes the chain testable
One thing has to survive every gate, every translation, every artifact in this volume: the brief's Prediction. Volume II established it — a baseline measured before any design, a target named, a check date set, an owner. "Forty-seven minutes today; under ten minutes thirty days after release; verified by recording Gal's next cycle." That number is what makes the chain testable end-to-end.
The prediction has to trace. Through the Epic name, the story, the wireframe states, the API contract, the Gherkin scenarios, into the test file, and finally into the production metric. If at any phase the prediction can no longer be traced — if you cannot point at this Epic and say this is the work that will move the 47 minutes, or at this scenario and say this is one of the cases that has to pass for the prediction to hold — then the chain has been broken at that phase. The team is no longer building toward the prediction. They are building something else, however well.
Every gate in this volume asks the same question at increasing resolution: does the work at this phase still trace back to the prediction the brief made?
If the answer is yes, meaning is travelling. If no, it has been substituted — and the cost will surface, weeks or months later, when the prediction is checked and finds nothing in the production metric had moved.
Volume V will be about that check. This volume's job is to keep the prediction alive long enough to reach it.