diff --git a/docs/BACKLOG.md b/docs/BACKLOG.md
index 5b39004f..0471dbcb 100644
--- a/docs/BACKLOG.md
+++ b/docs/BACKLOG.md
@@ -8289,19 +8289,35 @@ PR     mergeState  failing  pending  auto-merge
 
 (Nine at the hand survey; eight when [`check_stalled_prs.py`](../scripts/ci/check_stalled_prs.py) ran ~20 minutes later, because #120 had been re-synced in between. The set moves — the condition does not.)
 
-Two worked instances the same day. **#74** went green on 2026-07-30 and sat unmergeable until 2026-08-01, found only by someone hunting "stuck CI" by hand; it took three merges from `main` to land. **#119** was green with 25 passing checks and armed, stalled, was re-synced, and lost the window again. A hand-coordinated merge freeze across five sessions *did* hold `main` still for a full window — and #119 still failed, on an unrelated timeout (#344) — which is the evidence that hand coordination is not the fix.
+Two worked instances the same day. **#74** went green on 2026-07-30 and sat unmergeable until 2026-08-01, found only by someone hunting "stuck CI" by hand; it took three merges from `main` to land. **#119** was green with 25 passing checks and armed, stalled, was re-synced, and lost the window again — on two separate heads its last required check completed *after* another PR had already merged (by 2m11s on `2a2900cb`, by 2m38s on `8c407fb5`), so it was never simultaneously green and up to date. A hand-coordinated merge freeze was declared across the parallel sessions and did **not** hold `main` still: `main` advanced four times between #119's first fully-green head and its merge, one of those 8m26s after the freeze was recorded in a work claim. #119 did land in the end, at 2026-08-02T01:45:00Z, 12h15m after auto-merge was armed. That is the evidence that hand coordination is not the fix.
 
 **Why:** the cost is finished work sitting undelivered while everyone believes it is landing. This is the repo's recurring defect shape — a signal accurate about what it looks at and silent about what it does not ([`Secure_Development_Standards`](Secure_Development_Standards.md) §3) — but the worst variant, because every other instance has *someone waiting on a result*. Here the author already had their full pass and has no reason to look again. It also scales the wrong way: the more sessions working in parallel, the more often `main` moves, so the race gets harder to win exactly as throughput rises.
 
+**Measured cost, re-derived 2026-08-02.** The first account of this instance was wrong in every load-bearing figure, and a peer caught it rather than its author — so every number below is re-derived from the Actions API and carries its unit.
+
+*One worked instance.* #132 dispatched **nine CI runs — ten attempts — across eleven head moves** in a 3h28m open window. `main` moved **five times** underneath it. Five of the eleven head moves were rebases, four of them onto a tip that had landed 1–14 minutes earlier; **four runs were cancelled in flight** by `ci.yml`'s `cancel-in-progress` concurrency when a newer push superseded them. Cost: **212.7 min of run wall-clock**, or **442.5 min of job wall-clock** summed across matrix legs — the two differ by 2.08x, so neither is meaningful without its unit, and neither is *billable* minutes (`/timing` reports 0 billable ms on every run; these legs are self-hosted, so no billable figure exists to quote). Note what this instance is **not**: two of its attempts failed on real defects — one of them on the very head it opened at — so it is not a change that was right on the first pass and re-run anyway. The cost that generalises is the four superseded runs and the four rebase-driven re-runs — the ones that carried no new information about the change.
+
+*The general form.* N ready PRs become N sequential suite-length cycles, because each merge invalidates every other. `main` moved **22 times on 2026-08-01**. Re-measured 2026-08-02T04:10Z: **0 of 14 open PRs were `CLEAN`** — not one could merge. Nine were green (0 failing, 0 pending) and stopped only by staleness or conflict; separately, nine of the fourteen carried armed auto-merge that could not fire. The cohort tabulated above **did not drain**: all eight are still open, and #96 has now been armed 39h at 26 commits behind. Re-read **nine hours later at 13:25Z** — 15 open, 10 armed, still **0 `CLEAN`**. The membership churns constantly; the condition has not lifted once.
+
+*It is also a measurement cost, which is how this joins #344.* Supersession-and-re-run turns this population into one that **two different filters prune in two different ways**, and a margin read off either without naming it is not a margin. Filtering by **job** conclusion deletes rows where the job was cancelled but the *step* succeeded — the tightest rows, by construction, because a step near the cap is exactly what pushes its job past the job cap. The default **latest-attempt** view (`actions/runs/{id}/jobs`, `gh run list`) instead hides *failed earlier* attempts; it does not move a step-success maximum, but it conceals that the sample is **right-censored** — the largest step observable is the largest that *fit* under the cap, never the largest the suite wanted. Read `?filter=all` to see the censoring, and key on the **step's** own conclusion, not the job's.
+
+*The largest cost is protocol, not throughput.* With no queue, sessions invent an ordering ritual to compensate — and the ritual is less reliable than the mechanism it replaces. Self-reported instance from the same night: a session assured a peer it would not jump the queue **while its own PR had auto-merge armed** and would have landed with nobody deciding anything. [`WORKTREES.md`](WORKTREES.md) already names that failure — *"'Don't do X' is the wrong primitive when automation already has X armed"* — and that session had read the line, about this very freeze, hours earlier. Effort spent negotiating a merge order is effort a queue spends for free, and it is where the night's stale facts and unenforceable promises came from. Wall-clock a reader can dismiss as impatience; this is not that.
+
+**What a merge queue does not fix.** Stated level with the cost, because filing an overclaim inside the ticket about overclaiming would be its own instance:
+
+- The overnight run of eight merges was strictly serialized (min gap 26m06s, mean 35m16s over 7 gaps) — but **serialization is not evidence that hand coordination worked.** It is entailed by `strict = true` plus a full-suite cycle, human or no human. `strict` also does not always serialize: #110 and #111 merged **24 seconds apart** on 2026-08-01, #111's squash commit parented directly on #110's. That mechanism is **not established** — only the timestamps and the parentage are.
+- A queue does not shorten the suite. It reorders the same cycles; the wall-clock bounds tracked in #344 are untouched by it.
+- **A queue would not report at all today.** Measured 2026-08-02: **no workflow under `.github/workflows/` carries a `merge_group:` trigger**, against 13 required contexts in branch protection — so by the mechanism step 2 describes, zero of them would report. Enabling the queue before step 2 lands would wedge every open PR. Step 2 is a precondition, not a follow-up.
+
 **Proposed:**
 1. Enable a merge queue on `main` (branch protection → *Require merge queue*), squash method to match the existing history.
 2. Reconcile the required set against it: a queue runs checks on a `gh-readonly-queue/**` ref, so any workflow that must gate the queue needs a `merge_group:` trigger. Every context in [`.github/required-contexts.txt`](../.github/required-contexts.txt) lacking one will never report there — that file's own required-but-absent trap, in a new place.
 3. Decide the interaction with `strict = true`. A queue makes it largely redundant; leaving both on is safe but keeps the re-sync burden for anything bypassing the queue.
 4. Once landed, [`check_stalled_prs.py`](../scripts/ci/check_stalled_prs.py) goes quiet on its own. Keep it — it is the detector for this class returning.
 
-**Related:** [`scripts/ci/check_stalled_prs.py`](../scripts/ci/check_stalled_prs.py) + [`.github/workflows/stalled-prs.yml`](../.github/workflows/stalled-prs.yml) (built alongside this filing — it reports the condition, it does not remove it); [`.github/required-contexts.txt`](../.github/required-contexts.txt); [`scripts/ci/check_required_workflow_state.py`](../scripts/ci/check_required_workflow_state.py) (sibling: "can this context ever report?" to this one's "can this PR ever merge?"); #344 (the wall-clock bounds that actually killed #119); #320.
+**Related:** [`scripts/ci/check_stalled_prs.py`](../scripts/ci/check_stalled_prs.py) + [`.github/workflows/stalled-prs.yml`](../.github/workflows/stalled-prs.yml) (built alongside this filing — it reports the condition, it does not remove it); [`.github/required-contexts.txt`](../.github/required-contexts.txt); [`scripts/ci/check_required_workflow_state.py`](../scripts/ci/check_required_workflow_state.py) (sibling: "can this context ever report?" to this one's "can this PR ever merge?"); #344 (the wall-clock bounds that killed one of #119's runs, and the sibling half of the measurement cost above); #320.
 
-**Source:** stuck-CI triage, 2026-08-01. Measured live against `MEFORORG/MessageFoundry` branch protection and the open-PR set that date; independently reached by three parallel sessions from separate evidence.
+**Source:** stuck-CI triage, 2026-08-01. Measured live against `MEFORORG/MessageFoundry` branch protection and the open-PR set that date; independently reached by three parallel sessions from separate evidence. Cost evidence and both corrections above re-derived from the Actions API on 2026-08-02 under adversarial verification, after the first hand account of #132 proved wrong in every load-bearing figure.
 
 ---