session 2026-07-25: cleaner A (footnote-aware) + mechanism-census replaces volume-proxy; feedback-census-by-mechanism-not-proxy

claude/memory/MEMORY-reference.md

@@ -5,7 +5,7 @@ metadata:
   node_type: memory
   type: reference
   originSessionId: d1e67361-d1b9-4dac-a0c4-e7a0e26211c4
-  modified: 2026-07-24T17:48:30.828Z
+  modified: 2026-07-25T07:21:09.435Z
 ---
 
 # MEMORY — Reference layer (consult on demand)
@@ -29,6 +29,9 @@ Split out of [MEMORY.md](MEMORY.md) on 2026-07-06 to keep the wake-loaded index
 
 # Archived sessions + stable reference layer (relocated verbatim from MEMORY.md, 2026-07-06)
 
+## Archived (2026-07-24 night — the V-TEXT wave's first bite censused the whole class; the unlock is named, the engine track is independent; demoted on promote at the 2026-07-25 wrap)
+- [Session 2026-07-24 night — the V-TEXT wave's first bite censused the whole class; the unlock is named, the engine track is independent](session-2026-07-24-night-vtext-wave-censused-cleaner-unlock-engine-tracks.md) — Ran ONE file (Moby-Dick) end-to-end: **calibration ANSWERED** (clean PASS, fab 0, threshold-40 holds) but a cascade (edition-swap Penguin→B&N · imaged `$` glyph the backup-only census MISSED, caught by the lane's live guard · `strip_cruft` can't clean nested calibre spans) resolved into a ROOT — **`test_tools` = 35 unit tests / 0 integration tests**, so the REVIEWED-72 writer switch (`gfm-raw_html`→`markdown-smart`) silently **orphaned `clean_pandoc_html_residue` AND the ratified footnote fallback (REVIEWED-54 §5)**. Turned whack-a-mole into a **finite census** (`_curation/v-text-wave-gap-census-2026-07-24.md`) — **SUPERSEDED IN PART 2026-07-25** by the mechanism-census (its "31→one fix" was a volume proxy; see [[feedback-census-by-mechanism-not-proxy]]). **PULLING THREAD was: build the unlock, prove it by graduating `the-shape-of-a-pocket`** — done 2026-07-25 (cleaner A built; the answer: NOT one fix deep). LITERAL Q ("is 'heavy-styled' one fix deep") → ANSWERED no.
+
 ## Archived (2026-07-24 evening-continued — REVIEWED-72/74 BUILT + LANDED atomically (spec v2.6.0); the 47-file V-TEXT wave UNBLOCKED; demoted on promote at the 2026-07-24 night wrap)
 - [Session 2026-07-24 evening (cont.) — REVIEWED-72/74 built + LANDED atomically (spec v2.6.0); the 47-file V-TEXT wave is UNBLOCKED](session-2026-07-24-evening-reviewed72-74-vtext-gate-landed-v2.6.0-wave-unblocked.md) — Same day; steward /clear'd to build with room. Substrate re-read caught the prior wrap's "bounded" framing: the naive repoint FLAGs (run (3) carries 287 added → `classify` added_total>50 → FLAG). Grounding measurement (`scratchpad/ground_manifest.py`, read-only) **REFUTED the jurist's OWN stated premise** — the 287 = **279 DISPLACEMENT + 8 glyph, NOT injected content**; the injector enumerates `converted_markers` (646) on the **LOST** side. Also settled empirically: `_tokens` PRESERVES diacritics (`wstawać`≠`wstawa`) → glyph is TWO-SIDED. Steward chose routing (b) "no surprises" → **PENDING-73** package → **REVIEWED-74**: jurist **OWNED the refuted premise** (*"the producer must know X is not evidence the producer records X"*) + four corrections (Q1 in-scope on a recorded **three-part test** · Q2 **MULTISET** dominates presence [catches duplication] · Q3 two-sided **COUNT-SCOPED** [Counter subtraction; whole-run-or-nothing] · Q4 the glyph **CLASS** not future members) + 646-marker latent + a **fourth** demo case. **Built** `verify_body_conservation.resolve_added` (10 fixtures) + `apply_char_glyphs` two-sided manifest (added=joined / removed=`<img>`-split fragments) + repointed the reference → **FOUR-CASE demonstration PASSED** (clean PASS · prose-deletion FLAG · fabrication-outside-manifest FLAG fab=80 · Eichmann regression clean; RECORD doc) → **wired** into `graduate_to_canonical.body_conservation_gate`. **LANDED** (steward: "push into the landing"): spec **v2.5.0→v2.6.0**, froze v2.5.0, **BOUNDED-DIFF PROVEN** (exactly 4 regions); graduation-spec `body_conservation:` declared data (`added_fabrication_threshold:40`, `reader_audit` specified-not-yet-built VISIBLE); §Governance +2 disciplines (reassigned-component check · in-scope-refinement three-part test); chamber CLAUDE.md v2.6.0 + tool-evolution-log. **`eaa38bf` on github** (origin/Gitea unreachable, retry owed). Fleet **234/234**; every hook + pre-commit passed. **PULLING THREAD next = BEGIN THE 47-FILE V-TEXT WAVE** — first bite = ONE real **non-imaged** kindle_residue file end-to-end through the newly-wired gate (the demonstration proved the gate LOGIC, not a full GRADUATION); the 12 imaged census-flagged files need per-source glyph-maps first (parallel curation track). **LITERAL Q: does a live book produce a clean graduation, or surface an `added_fabrication_threshold` floor HIGHER than the 2-book calibration (40), forcing a recalibration?** *(→ ANSWERED in the night session: clean PASS, fab 0, threshold holds — but the file surfaced a tooling cascade that censused into the finite gap-map; the calibration was never the risk, the composed pipeline was.)*