---
name: a-status-sentence-must-be-derived-from-the-measurement-beside-it
description: A pre-written status claim sitting next to a fresh measurement borrows its freshness — the number is live, the sentence is not.
metadata:
  type: feedback
---

2026-09-10, releases#294 Arm B. Twice in one hour I reported "no new subject traces yet" while the
subject's row was in the very output I was reading. The second time was in the same message where I
claimed the discipline the first one had taught me.

**Mechanism, not carelessness:** I computed the count and wrote the status text in ONE tool call — the
count was live, the prose was a static string authored before the call ran. A number computed beside a
claim does not check that claim; it lends it unearned freshness. doyle banked the general form: *a
status sentence must be DERIVED from the measurement it sits next to, or it is decoration.*

Adjacent instance the same hour: I argued the subject was "still ahead" in B3 because B1 finished it at
index 2019 and B2 at 1986 — completion order varies with 8 slots and differing durations, so that was
an inference from two draws presented as a position fact, and it was wrong (todlando caught it).

**Why:** a status is what peers act on. A stale sentence with a fresh number beside it reads as
measured, so nobody re-checks it — and the error only surfaces when someone else reads the log.

**How to apply:** build the status FROM the values in the same expression — interpolate them, branch on
them (`test -f X && echo TERMINAL || echo running`) — never hand-write the conclusion and compute the
number alongside. When you catch it, withdraw with the mechanism, not an apology. Related:
[[an-identical-treatment-controls-the-input-not-the-effect]], [[a-zero-from-an-absence-grep-is-a-spelling-claim]].