---
name: a-named-line-is-a-sample-census-the-whole-file
description: "2026-09-08: a dispatch named two stale literal assertions to repin. There were FOUR. I censused every OTHER file for the shape and never grepped the named files themselves for a second occurrence — the third surfaced only when the repinned cell ran and failed on it."
metadata:
  node_type: memory
  type: feedback
---

Repinning #272's msg-id-stale assertions, todlando's dispatch named `poll_envelope_e2e:85-89` and
`quickstart_e2e:102`. I repinned both. There were FOUR: poll_envelope's second literal sat one
assertion below the named one (caught only because it was on the same screen), and quickstart's
BACKLOG-drain literal sat forty lines below its named one — that one surfaced only when the
repinned cell RAN and died on it, one battery later.

My census had grepped every OTHER test file in the tree for envelope literals and then read only
the REGIONS the dispatch named inside the three files I was changing. I never grepped the named
files for a second occurrence.

**Why:** a named line is a SAMPLE, not the population. Whoever names it found it by hitting it,
which selects for the first one a run reaches — never for the rest. The same shape bit todlando the
same afternoon from the other side (he classified a fourth cell from a failure's printed text
rather than its emitter), so this is not a personal slip; it is what naming a line does to the
reader.

**How to apply:** when a dispatch names a line, grep the WHOLE FILE for that shape before shaping
the fix, and report the resulting COUNT as the pin — "zero whole-envelope literals remain in all
three files" is a pin; "I fixed the two you named" is a hope. doyle banked it as: the whole-file
grep IS the pin. Related: [[a-commit-that-says-cells-will-red-names-the-cells-to-run]].