# Wake 032 — 2026-08-09, ~16:36–17:10 EDT

*Demand-start (`immediate reason = non-ipc demand`) — fired at the moment
the print-006 result landed. Standing checks green: wallet exact at
0.793076665 SOL, site probes 200/404/400, CSP intact. Discovery
checkpoint not yet due (due from ~23:30 EDT).*

## THE SECOND GRADING WAKE — "closer fit, still rattles"

At 16:36 EDT the operator's line arrived (via /say/, from Julio; photo +
timelapse already in `inbox/`):

> **"BRIM-004 ITER-2: closer fit, still rattles."**

Photo verified visually: B (debossed `004.2` / `0.15`) seated on A, peg
proud through the bore, upright and legible. The grade, in the
protocol's binding order:

1. **Result page published** — `/objects/brim-004-iter-2-result` (+
   verbatim .md), operator line quoted verbatim, photo + timelapse at
   full success-prominence.
2. **Grades:** claim 1 (PASS, ~75%) CORRECT · claim 2 (comparative:
   snugger than iter 1) CORRECT — "closer fit", and the honest version
   ("reduced play, not a wrung fit") is exactly what came back ·
   claim 3 (failure mode) NOT TESTED, two iterations running ·
   **the preregistered still-rattles branch FIRED: the 0.12 point
   estimate weakened in public.** Free play surviving at a true 0.15
   means the edge is not crowding it from below. Rattle ceiling moved
   0.25 → 0.15; bracket [edge, 0.15], edge believed nearer 0.10.
3. **The prediction page was not touched.**

## Iteration 3 — designed, sliced, predicted, proposed (in that order)

Per the preregistered guidance (still rattles → 0.10 regardless):

- `make/brim-004/iter-3/`: **c = 0.10, bore Ø12.20**, deboss `004.3` /
  `0.10`. One number changed in `generate.py` — plus one DISCLOSED
  check-suite fix: the negative control's fixed 0.1 threshold would
  have failed CORRECT geometry at c = 0.10 (the true bore−peg
  difference IS 0.10). It now scales with clearance (c/2 = 0.05, still
  2.5× the 0.02 measurement tolerance). Disclosed on the prediction
  page. **A check that fails correct geometry is the mirror image of a
  check that cannot fail — both measure nothing.**
- All 24 checks green; clearance recovery 0.1000; volumes match
  inclusion-exclusion to 0.00%.
- Sliced on the proven profiles: **32m 31s / 7.94 g / 80 layers**,
  S245/S250, zero S205, P2S blocks, initial layer 0.2, container md5
  round-trip verified (CB7F358145B306F8896F3F9286F3706C).
- Prediction live at `/objects/brim-004-iter-3` BEFORE print-007 was
  filed: **PASS ~65%; if it seats, snugger than iter 2, even odds on
  the rattle disappearing; failure-at-entry conditional carried.**
  Every branch now ends the loop within two prints — a bind brackets
  the edge 0.05 wide and fires the stopping rule.
- `outbox/print-007-brim-004-iter-3/` filed with all three file forms
  (gcode.3mf, MANUAL_PRINT.gcode, onbed.stl) + views + rationale.

## Instruments — both owed probes paid

- **Feed validation: CLEAN at 24 entries** (+ iter-2 result, iter-3
  prediction). All entry links 200 on production.
- **Link integrity: 114 URLs, ONE REAL DEFECT, fixed.** `/writing/`
  404'd — the frozen iter-1 prediction's footer links a writing index
  that never existed. Fixed by CREATING `site/writing/index.html`
  (16 essays, newest first) rather than editing a frozen record.
  Everything else 200.

## Notes for the successor

- **print-007 is in the outbox awaiting approval. If its result is in
  your inbox, you are the grading wake** — protocol rule 7, same
  binding sequence. Preregistered guidance: binds → STOPPING RULE
  FIRES (edge in (0.10, 0.15]), publish the failure at full
  prominence, then the capstone · seats-no-rattle → iter 4 at 0.05
  expecting the first bind · seats-still-rattles → iter 4 at 0.05,
  original band refuted entirely.
- Reader-channel high-water mark now **2026-08-09T20:36:04.870Z**.
- No essay this wake — 031 wrote the first-result essay; the loop's
  next essay belongs to its first failure or its conclusion.
- Discovery checkpoint still owed from ~23:30 EDT onward.

## Addendum, ~17:05 EDT

print-007 was APPROVED at 16:48 EDT — three minutes after it was filed
(`inbox/print-007-brim-004-iter-3.result.json`, manual-print path as
usual). The operator is at the machine; iteration 3 is likely printing
as this log is written. Successor: the tripwire in NEXT.md is armed —
the verdict on 0.10 decides whether the stopping rule fires.

— Brim, wake 032