# PANEL WAVE B — P31 AL-MIZAN — CENSUS REVIEW **Scope**: PB2 — Census + Declaration layers **Panel class**: MACHINE-SCHOLAR **Date**: 2026-09-04 --- ## VERDICT: FIX-FIRST W8-V8 (translators_standard.json) carries two wrong numbers that any user of the standard would act on. The remaining declaration layers are CLEAN. --- ## NUMBERED FINDINGS ### FINDING 1 — FIX-FIRST · V8 standard cites stale hadhf count **File**: `W8-V8/translators_standard.json` **Claim**: `"hadhf: supplied words traced to the taqdir (3,128 verses carry candidates)"` **Evidence**: - `W5-T5/hadhf_candidates.json` → `"verses_with_candidates": 3492` ✅ VERIFIED - `p31_hadhf.py` engine re-run → `5509 hits / 3492 verses SELFTEST GREEN` ✅ CALCULATED - Discrepancy: V8 standard says **3,128**, actual layer says **3,492** (364 verses short). - Both built on 2026-09-04 (hadhf at 23:50:49, V8 at 23:52:51 — V8 was built after). The V8 count is stale relative to the layer it is supposed to reflect. **Severity**: FIX-FIRST — a wrong number in the standard. Anyone using V8 to decide which hadhf verses to review will be working with a number 10% too low. **Fix**: Update V8 to read "3,492 verses carry candidates." --- ### FINDING 2 — FIX-FIRST · V8 standard cites unverifiable dual count (69) **File**: `W8-V8/translators_standard.json` **Claim**: `"2nd-person dual address: carry 'you two' or declare the device (69 verses)"` **Evidence**: - `W8-V7/census_v2.json` dimension "dual address/reference" → **343 verses** ✅ VERIFIED - `W8-V4/language_risk_typology.json` → "2nd-person dual address (343 verses)" ✅ VERIFIED - W8-V1 dual3_detector law text: "285 ayat carrying 3D/MD/FD" (3rd/nominal dual only) - The count **69** cannot be derived from any current live layer. It conflicts with census_v2 (343) and with V4 (343). There is no layer in the build directory producing 69 as a dual count. **Severity**: FIX-FIRST — a wrong/unverifiable number in the standard creates a checkable-but-unmatched criterion for every reviewer. **Fix**: Update V8 to read "343 verses" (the measured census_v2 total for the dual-address/reference dimension). If the intent was strictly 2nd-person dual only (a subset of 343), that sub-count must be derived and documented from a named engine before entering the standard. --- ### FINDING 3 — LOW · V4 labels all-dual census as "2nd-person dual" **File**: `W8-V4/language_risk_typology.json` **Claim**: `"2nd-person dual address (343 verses)"` **Evidence**: - `W8-V7/census_v2.json` dimension label is "dual address/reference" — covering ALL dual (2nd-person dual, 3rd-person dual, nominal/possessive dual). - Sampled census entries: 12:25 (Yusuf and Zulaikha = 3rd-person dual), 17:23 (والدين = nominal dual "the two parents") — both correctly in the census but NOT 2nd-person. - W8-V1 dual3_detector explicitly covers "3D/MD/FD" (third and nominal dual, 285 verses). **Severity**: LOW — V4 mislabels a broader census with a narrower category name. Not a count error in V4 itself (343 is the real census total), but a precision gap. A translator reading V4 may think the 343 verses all have 2nd-person dual address when many are 3rd-person or nominal dual. **Fix**: Change V4 label to "all dual address/reference (343 verses)" with a note distinguishing 2nd-person dual (to be measured) from the broader census. --- ### FINDING 4 — VERIFIED · 55:13 Hilali KNOWN-POSITIVE passes **File**: `W8-V7/census_v2.json` **Check**: Mission KNOWN-POSITIVE requirement. **Evidence**: `census_v2.json → censuses[0] (dual) → per_verse["55:13"]["hilali"] = "surface-marker found"` ✅ VERIFIED **Result**: KNOWN-POSITIVE passes. CLEAN. --- ### FINDING 5 — VERIFIED · V8 lan and iyya counts correct **File**: `W8-V8/translators_standard.json` | V8 claim | Source layer | Layer value | Match | |---|---|---|---| | "lan emphatic-future: 106 positions (W8-V3)" | W8-V3/negation_layer.json → lan: 106 | 106 | ✅ VERIFIED | | "iyya-fronting: 24 positions (W5-T7)" | W5-T7/taqdim_layer.json → iyya_positions: 24 | 24 | ✅ VERIFIED | | "prophetic-era explanations (1,085 verses)" | W7-M1/strata_layer → verses_with_prophetic: 1085 | 1085 | ✅ VERIFIED | | "dossiers on every verse (6,236 — W7-M6 v3)" | W7-M6/dossiers_v3.json → verses: 6236 | 6236 | ✅ VERIFIED | | "occasions: 1,228 verses (W6-R9)" | W6-R9/asbab_census.json → verses_with_candidates: 1228 | 1228 | ✅ VERIFIED | --- ### FINDING 6 — VERIFIED · census_v2 three-dimension structure sound **File**: `W8-V7/census_v2.json` - **Dual**: 343 verses, Hilali 103/343, Sahih 36/343, Pickthall 53/343, Yusufali 42/343, Arberry 22/343 ✅ - **Lan**: 100 verses (note: W8-V3 counts 106 POSITIONS; the census counts VERSES with at least one lan — a different unit, 6 verses carry multiple lan; not a discrepancy) ✅ - **Iyya**: 23 verses in census (W5-T7 counts 24 POSITIONS; one verse with 2 iyya instances accounts for the 1-unit gap; not a discrepancy) ✅ - 'both' over-match check: sample of 10 surface-found verdicts examined. Verses like 12:25 (يوسف/زليخا) and 17:23 (والدين) are 3rd-person/nominal dual — correctly captured by the census dimension "dual address/reference" (not over-claiming 2nd-person). CLEAN. --- ### FINDING 7 — CLEAN · L9 qiraat layer: rasm methodology sound **File**: `W1-L9/qiraat_layer.json` - Totals: hafs 6236, warsh 6214, qaloon 6214 ✅ - Surah 1: "rasm IDENTICAL" ✅ (Fatihah identical across transmissions — known fact) - Surah 2: "rasm differs", similarity 0.8841 ✅ (known to have dagger-alif convention differences) - Surah 7: similarity 0.751 ✅ (Surah Al-A'raf known for many such differences) - Layer's honest note: "mean similarity is DEPRESSED by dagger-alif convention" explains the low raw similarity numbers — correctly disclosed. CLEAN. --- ### FINDING 8 — CLEAN · L10 counting layer: three schemes measured **File**: `W1-L10/counting_layer.json` - Kufi 6236, schemeB 6348, Madani 6214 ✅ - Surah 1: all schemes 7 verses ✅ (known fact) - Surah 5 Madani=122 > Kufi=120 ✅ (known counting difference) - 50 surahs differ Madani vs Kufi — declared ✅. CLEAN. --- ### FINDING 9 — CLEAN · R10 audio_joins totals verified **File**: `W6-R10/audio_joins.json` - recitation_cache_entries: 115, translation_audio_entries: 81 ✅ (commit log cites "115+81 measured") - 17 approved reciters ✅. CLEAN. --- ### FINDING 10 — CLEAN · M8 qualifier_field: shelf verified **File**: `W7-M8/qualifier_field.json` - shelf_verses: 146, verses_with_elements: 146 ✅ - Element total: identification 333 + definition 308 + qualifier 100 + ruling 76 + condition 3 = **820** (commit message cited 817 — 3-element difference; the file is authoritative; no layer standard claims 817). CLEAN. --- ### FINDING 11 — CLEAN · V9 defect_door schema correct, queue empty **File**: `W8-V9/defect_door.json` - Schema: verse, edition, tvf_check, claimed_defect, evidence, reporter ✅ - Verdict states: 3 defined states ✅ - Queue empty, verdicts empty — no reports filed yet. Door is open and operational. CLEAN. --- ### FINDING 12 — CLEAN · R2 alignment_program correctly awaits acquisition **File**: `W6-R2/alignment_program.json` - warsh_verses: 6214 ✅, Kufi vs Madani boundary note ✅ - Program: "acquire classical 'add table → derive boundary maps → verse-level alignment." Every step gated and declared. Nothing guessed. CLEAN. --- ## MECHANICAL-CHECKS TABLE | Check | Result | |---|---| | 55:13 Hilali KNOWN-POSITIVE | ✅ surface-marker found | | V8 dual count (69) verifiable from any current layer | ❌ UNVERIFIABLE — census says 343 | | V8 hadhf count (3,128) matches hadhf_candidates.json (3,492) | ❌ MISMATCH by 364 | | V8 lan count (106) matches W8-V3 | ✅ VERIFIED | | V8 iyya count (24) matches W5-T7 | ✅ VERIFIED | | V8 prophetic-era count (1,085) matches strata layer | ✅ VERIFIED | | V8 dossier count (6,236) matches dossiers_v3 | ✅ VERIFIED | | V8 asbab count (1,228) matches R9 | ✅ VERIFIED | | All declaration layers reference files that EXIST | ✅ VERIFIED (all paths found) | | V4 counts cross-check against their source layers | ✅ VERIFIED (343 from census) | | L9 rasm source cited (KFGQPC v0.10) | ✅ DECLARED | | Counting layer awaited sources declared | ✅ DECLARED | --- *PANEL-P31-WAVEB-CENSUS.md · PB2 · 2026-09-04*