Claude Code First, specification version 0.1.0. The plain-language guide
Version 0.1.0. Level 3. Best evidence grade R3/S3.
VG-9 Measure before building
A change made to improve a measurable quantity SHOULD be preceded by a measurement of that quantity, and kept only if a second measurement shows the gain.
| Field | Value |
|---|---|
| Keyword | SHOULD |
| Level | 3 |
| Checked | by judgement |
| Findings | F103, F102 |
| Harness facts | none |
| Threat | internal: designs also survived measurement, and no rate is known |
| Status | active |
Rationale: in three projects a design or diagnosis written before implementation was overturned once it was built or measured (F103, R3/S3). winwright built, measured and reverted WW249: "take the per-code-unit send back out, the rate did not move" [winwright@5012473], and retired two filed tasks whose premise did not survive a measurement taken before building (F102). F103 also records a design that measurement confirmed, so the evidence shows that measuring settles the question, not that designs usually fail; the rule carries SHOULD for that reason.