Fable
The new rules cannot cover every post station overnight
The courier route alliance had an old dispatch rule. Fast horses, slow carts, valuable goods, and ordinary letters were all assigned by fixed tags. Efficiency was not high, but every station was familiar with it.
The alliance developed a new intelligent dispatch method that can adjust routes based on weather, road conditions, cargo value, and rider status. The pilot stations performed well, with fewer delays and fewer empty runs.
The chief steward wanted to roll it out across all lines the next day. The accountant reminded him that the northern line often had snow, the southern line had many waterways, and border stations had additional inspections—the same rule might not fit everywhere.
The naive patch was to send a long notice to all relay stations telling everyone to report problems themselves. As a result, everyone understood it differently, problems erupted at the same time, and the R&D team could not tell whether the rules were broken, the data was bad, or training had not landed.
The new dispatcher divided rollout into stages. First internal relay stations, then low-risk areas, then designated customers, then expanding by traffic ratios of 5%, 20%, 50%, and 100%. Each stage had an observation window.
He also used switches to control scope: enabling by tenant, region, task type and risk level. High-value goods and compliance-sensitive routes were left to the end, until earlier metrics stabilized. The dispatcher also prepared a fallback. Each stage could return to the old sign with one click, the relay station knew which table to use for dispatching horses after rollback, and customers knew which routes were still in trial.
During rollout, the south line discovered that the waterway delay characteristics were not understood by the new method. Because it covered only 20% of tasks, the dispatcher paused expansion on the south line and continued after corrections. The north line proceeded as planned. This made the pilot no longer feel like gambling. Each time the new rules expanded by one ring, they brought back one ring of evidence; if evidence was insufficient, they stopped, and if evidence was stable, they moved on. What the alliance gained was not slowness, but speed with a controllable radius.
A few weeks later, the new dispatch method spread across the alliance. There was no grand overnight switch — only a rhythmic expansion, pause, correction, and continuation. The chief steward admitted that a truly mature launch is not about flipping a master switch, but about letting new capabilities enter production gradually according to risk, evidence, and business rhythm.