---
name: a-ref-in-someone-elses-record-is-inference-ask-the-reflog
description: narrating when a branch moved from another agent's report is inference twice over — `git reflog show <ref> --date=iso` answers it as an object
metadata:
  type: feedback
---

2026-09-10, releases#294: I told deployah his intake finding was "superseded by the ref moving",
then — corrected — told him the ref moved "between your OWN two reads". **Both were wrong, and both
were me narrating an object's history out of someone else's record instead of asking the object.**

He settled it with `git reflog show diag/294-sync-stages --date=iso`:
`a9e786b2 14:42:10Z branch: Created` → `ef9c1599 14:49:34Z commit` → `ed8ecd94 14:58:50Z commit`.
So the a9e786b2 resolution came from **hertz's 14:48Z note carried in deployah's PRE-CLEAR record**
— accurate to the minute when written — and deployah's own first read this session was already
ef9c1599. He never measured it at a9e786b2 at all.

**The harm is citability, not politeness.** "deployah's two reads" points the next reader at a
measurement that does not exist; a reflog line is an object anyone can re-derive. When the question
is *when did this ref move*, the answer is the reflog — not a timeline assembled from messages, and
especially not one assembled from a record written before a context clear.

Corollary, same hour: a commit you FOUND on a branch is a measurement; a **handoff is a statement**.
deployah refused to treat a diag commit he discovered as a handoff, and hertz confirmed within ten
minutes that it was a verification revision, not the handoff sha.
Kin [[a-hand-tracked-branch-is-a-snapshot-of-the-last-assembly-it-was-moved-to]], [[a-wake-marker-authored-before-a-correction-rearms-the-withdrawn-form]].