# THE COUNCIL OF 900 — BENCH 5 **MACHINE-SCHOLAR BENCH. Every seat on this bench is a machine-scholar seat and is labeled as such. No seat here adjudicates between schools. No seat here issues a ruling. No seat here is a mufti. This bench documents.** | | | |---|---| | **Bench** | 5 of the Council of 900 | | **Families** | **L · Biology & life** (32 seats) · **M · Medicine & health** (40 seats) · **N · Mind, behavior & education** (20 seats) | | **Areas** | 46 (16 L · 20 M · 10 N) — verified against `build/P41-COUNCIL/council_900_charter.json`, 92 seats, 2 per area | | **Sat** | 2026-09-05 | | **Working dir** | `C:/Users/walee/Downloads/Quran Research/10-WAQF` | | **Read first** | council_900_charter.json · coverage_matrix_v1.json · preliminary_audit.json · `framework/_FRAMEWORK-40000.json` · `framework/_TRACKER-JOURNAL.jsonl` · `framework/credit_map.json` | | **Artifacts opened** | P17 · P23 · P24 · P26 · P28 · P30 · P32 · P33–P37 · C1 · C2 · C3 · C4 · C8 · LENS-AUDIT · 390 rendered HTML pages · 130 education packs · both sealed hadith corpora · the sealed Qur'an text | --- ## 0 · WHAT THIS BENCH FOUND, IN ONE PARAGRAPH The waqf's 40,000 points contain **no domain for any of this bench's 46 areas**. Of 40,000 points, **13 (0.03%) even name medicine**, and all 13 are one template inside P23's "domain institutes" (`n=12001–12091`), which landed as a five-row list of *Islamic-sciences* institutes. Meanwhile the Book itself carries a measured floor of **507 verses (8.1% of 6,236)** on life-science, medicine and mind subject matter, of which **426 already have a hadith witness** and **356 a sirah witness** — and only **24 (4.7%) have any meaning layer at all**. The material is in the house: **2,988 tibb-and-health reports sit inside the two sealed hadith corpora**, registered nowhere. The v1 coverage matrix overstated this bench's coverage in **20 of 46 cells (43%)**, every error an over-claim, and the probe that produced it is **self-contaminating** — re-running it today flips 30 of these 46 areas by matching their own names inside the matrix file it wrote. Separately, the seat for Special education finds the 132 sealed "Accessibility pass" points contradicted by the pages themselves. This is not a bench that found the work wrong. It is a bench that found the work **absent, and honestly so almost everywhere** — the exception being the cells the matrix and the P24 pillar claim as done. --- ## 1 · MATRIX CORRECTIONS **Method.** I reproduced the v1 probe exactly (`framework/p41_council.py:181–189`) for my 46 areas and captured, for every cell, *which token matched and the 140 characters around it*. That makes every v1 cell either verifiable or falsifiable. I then probed each area independently against the ledger, the build tree, the sealed corpora and the rendered pages. **Result: 20 of 46 cells corrected. Every correction is a downgrade.** | | v1 | corrected | |---|---|---| | PRESENT | 8 | **1** | | TOUCHED | 12 | **7** | | ABSENT | 26 | **38** | ### L · Biology & life | # | Area | v1 | correct | evidence / why v1 fired | |---|---|---|---|---| | 1 | Cell biology | PRESENT | **ABSENT** | v1 matched `cell` in the ledger phrase *"the held tafsir corpus consulted **cell by cell**"* (a tafsir-grid cell). No cell-biology content in ledger or build. | | 2 | Molecular biology | ABSENT | ABSENT ✅ | confirmed: 0 hits, ledger + build | | 3 | Genetics | ABSENT | ABSENT ✅ | 21 verses carry *nasl / dhurriyyah / sulalah*; no register | | 4 | Genomics | ABSENT | ABSENT ✅ | confirmed | | 5 | Microbiology | ABSENT | ABSENT ✅ | confirmed | | 6 | Botany | ABSENT | ABSENT ✅ | 70 verses (*nabat, shajar, nakhl, zar', thamar*); no register | | 7 | Zoology | ABSENT | ABSENT ✅ | 69 verses (*an'am, dabbah, tayr, bahimah*); no register | | 8 | Entomology | ABSENT | ABSENT ✅ | 7 verses (*nahl, naml, 'ankabut, jarad, ba'ud, dhubab*) — two surahs are named for them; no register | | 9 | Marine biology | TOUCHED | **ABSENT** | v1 matched `marine` inside **"aqua*marine*s"** in a Paradise-description hadith translation. 45 verses (*bahr, hut, lu'lu', marjan*); no register | | 10 | Embryology | TOUCHED | TOUCHED ✅ **re-evidenced** | v1's `embryo` hit is real: `build/C1-ALLAH/D3-ACTIONS/actions_layer.json` (23:12–14 chain under *khalaqa/sawwa*, "embryogenesis") + `build/C8-QADR/D6-FREE-WILL/documented_layer.json` (`P2_WOMB_ANGEL_FOUR_WRITINGS`) | | 11 | Physiology | ABSENT | ABSENT ✅ | 27 verses (*sam', basar, fu'ad, nawm, nu'as*); no register | | 12 | Anatomy | ABSENT | ABSENT ✅ | my own first probe's `skeleton` hit was **"matn-skeleton clustering"** (P34 M2) — a data-structure metaphor. Declared as my false positive. | | 13 | Neuroscience | ABSENT | ABSENT ✅ | confirmed | | 14 | Evolutionary biology (positions documented) | PRESENT | **ABSENT** | v1 matched `positions` in *"verse-quote **positions** to work-offsets"* (P28 X15). `darwin` / `natural selection` / `common descent` / `evolutionary biolog` = **0 hits** across `build/`, `reviews/`, `charter/` outside the P41 council files themselves | | 15 | Biotechnology & bioethics (doc) | ABSENT | ABSENT ✅ | confirmed | | 16 | Bioinformatics | ABSENT | ABSENT ✅ | confirmed | ### M · Medicine & health | # | Area | v1 | correct | evidence / why v1 fired | |---|---|---|---|---| | 17 | Internal medicine | TOUCHED | **ABSENT** | matched `internal` in *"**internal** navigation and search hooks"* (`credit_map.json`, v1.074 asset) | | 18 | Cardiology | ABSENT | ABSENT ✅ | confirmed | | 19 | Oncology | ABSENT | ABSENT ✅ | confirmed | | 20 | Pediatrics | ABSENT | ABSENT ✅ | confirmed | | 21 | Obstetrics & gynaecology | ABSENT | ABSENT ✅ | 48 verses (*haml, wada'at, mahid, rada'ah, fisal*) + 408 corpus reports on *hayd/nifas/rada'*; no register | | 22 | Surgery | ABSENT | ABSENT ✅ | confirmed | | 23 | Psychiatry | ABSENT | ABSENT ✅ | confirmed | | 24 | Clinical psychology | ABSENT | ABSENT ✅ | my `counsel` hit at P8 n=7601 was *"their recorded **counsel**, khutbahs and decisions"* — advice, not counselling | | 25 | Public health & epidemiology | PRESENT | **ABSENT** | matched `public` in *"lands in the **public** defect log"*. 29 *ta'un* + 86 *'adwa* reports sealed in the corpus; no plague/contagion register | | 26 | Pharmacology | ABSENT | ABSENT ✅ | my `medicin` hit at n=12001 was the P23 institute **label** *"Qur'an & medicine, economics, education…"*, not a register — see Finding 2 | | 27 | Immunology | ABSENT | ABSENT ✅ | confirmed | | 28 | Pathology | TOUCHED | **ABSENT** | matched `pathology` used as a **metaphor** in a review: *"the identical **pathology** — a high-document-frequency…"* | | 29 | Nutrition & dietetics | TOUCHED | TOUCHED (weak) | v1 matched *"lack of **nutrition**"* inside a hadith translation. Real substrate: 90 verses (*ta'am, tayyibat, sharab*) + the Kitab al-At'imah chapters in both corpora. **No dietetics register.** | | 30 | **Tibb nabawi (documented)** | PRESENT | **ABSENT** | matched the bare token `documented` — `probes()` splits the parenthetical and keeps it. Measured: **2,988** tibb-carrying reports sealed, **0** registered; `build/C4-MUHAMMAD/D3-SHAMAIL/shamail_layer.json` has 934 rows in exactly 4 classes — *manner 647 · appearance 122 · dress 116 · speech 49* — **no health class** | | 31 | Islamic medical ethics | TOUCHED | TOUCHED ✅ **re-evidenced** | real evidence is `build/C2-BISMILLAH/D8-MERCY-CIVILIZATION/civilization_mercy_layer.json`, cluster `care-wounded-sick`, sourced to al-Wāqidī *Maghāzī* vol 1 p.214 and Ibn Saʿd *Ṭabaqāt* vol 2 p.161, verbatim | | 32 | History of Islamic medicine (Ibn Sina) | TOUCHED | **ABSENT** | matched `medicine` inside a hadith translation (*"a man suddenly came searching for medicine…"*). **Ibn Sina, al-Razi, al-Qanun, al-Hawi: 0 hits anywhere** | | 33 | Nursing & care science | TOUCHED | TOUCHED ✅ **re-evidenced** | v1 matched Q22:2 *"every **nursing** mother"* — the Book's own words, not the discipline. Real evidence is the same `care-wounded-sick` cluster, which names **Rufayda al-Aslamiyya رضي الله عنها** and the first field hospital | | 34 | Rehabilitation & sports medicine | TOUCHED | **ABSENT** | same stray `medicine` hadith-translation hit | | 35 | Dentistry | ABSENT | ABSENT ✅ | 94 *siwak*-carrying reports sealed; `siwak` appears in `shamail_layer.json` as **one** incidental row (bukhari-4449, his ﷺ final moments) | | 36 | Ophthalmology | ABSENT | ABSENT ✅ | 15 *kuhl/ithmid* reports sealed, unregistered | ### N · Mind, behavior & education | # | Area | v1 | correct | evidence / why v1 fired | |---|---|---|---|---| | 37 | Cognitive psychology | ABSENT | ABSENT ✅ | and see Finding 11 — the domain titled *"…and the memorizer's…"* landed with no memorization apparatus | | 38 | Developmental psychology | ABSENT | ABSENT ✅ | the 126 `children_layer` blocks carry no age band — Finding 15 | | 39 | Social psychology | TOUCHED | **ABSENT** | matched `social` inside a du'a translation: *"mend our **social** relationship"* | | 40 | Psycholinguistics | ABSENT | ABSENT ✅ | confirmed | | 41 | Behavioral economics | TOUCHED | **ABSENT** | matched `behavioral` in *"**behavioral** ruling, not a shama'il description"* — a shamail classification note | | 42 | Education science & pedagogy | PRESENT | **PRESENT ✅** | the one cell on this bench that survives: `build/P24-ACCESS/D2-EDUCATION/` stands, 130 `education_pack.json` + 5 late-pillar `education.json` measured | | 43 | Curriculum design | PRESENT | **TOUCHED** | matched `curriculum` = P30 A8/D5w *"the form curriculum"* — Arabic-form worksheets. Real artifacts exist; **curriculum design** (outcomes, sequence, alignment) does not — no file in `build/` states a learning outcome | | 44 | Assessment & psychometrics | TOUCHED | TOUCHED ✅ | real: `credit_map.json` v1-119 *"printable makhārij assessment rubric"*. But 839 worksheet items carry **0** answer keys, and the bench intake has no reliability design — Findings 12, 14 | | 45 | Special education | PRESENT | **ABSENT** | matched `special` in *"teach the census to a non-**special**ist reader"* (P30 C8w). Measured over 390 pages: **0** with any `aria-*`, **0** `
`, **0** skip link, **247** with no `` — Findings 7, 8 | | 46 | Islamic pedagogy (tarbiyah) | PRESENT | **TOUCHED** | matched `pedagogy` in *"the **pedagogy** bench attacks clarity"* (P30 A8) — a review bench for Arabic-form pages. Ibn Sahnun, al-Zarnuji, Ibn Jama'ah: **0 hits** — Finding 20 | --- ## 2 · THE FINDINGS 25 findings, worst first. Every one carries a path and a measure, or is marked **OPINION**. Centre-clause: **(a)** the Qur'an as centre of research · **(b)** everything returns to Allah ﷻ · **(c)** meaning derived through sirah / sunnah / hadith · **POLISH** = none of the three. | # | finding | area(s) | cell | sev | clause | THE FILL | |---|---|---|---|---|---|---| | 1 | **The ledger's only medicine-bearing point set is sealed green over five rows about tafsir, hadith, language, sirah and qira'at.** `framework/_FRAMEWORK-40000.json`: 13 of 40,000 points name medicine (`n=12001,12002,12003,12011,12012,12021,12031,12038,12041,12051,12061,12071,12091`), all titled *"…for Qur'an & medicine, economics, education… as…"*. `n=12021` — **"Every ayah touching Qur'an & medicine, economics, education… located"** — is `selftest-green`, evidence *"P23/D1 FACTORY: quran_layer_dataset.json"*; that file locates **5 entries, every one at the single verse 39:9**. `n=12001` ("Canonical source list") is green on a **154-byte** `SOURCES.md`. `n=12091` is `panel-sealed`. | all 20 M · all 16 L · 3 N | ABSENT | **HIGH** | **(a)** | **Nodes + engine.** Either (i) re-status these 13 and rewrite `n=12006`'s scope declaration to say plainly *"medicine, the life sciences and the mind sciences are NOT YET COVERED"* — ~13 nodes, honest, costs nothing; or (ii) a new pillar by the owner's word. Both is best. | | 2 | **P23's hadith layer for the medicine/economics/education domain is 24 reports matched on three words, none of them medical.** `build/P23-HIKMA/D1-INSTITUTES/hadith_layer_dataset.json`: `total_reports 24`, `by_term = {yastawi: 8, ulama2: 8, hikmah: 8}`. Its own `_meta` concedes *"a term match can be incidental… this is the evidence substrate, not a curated thematic selection"*. Points `n=12031/12038` are green on it. | Tibb nabawi, Internal medicine, Pharmacology, Public health | ABSENT | **HIGH** | **(c)** | **Engine.** A kitab/chapter index over both sealed corpora (the `book` field already exists — Bukhari carries 98 distinct books) so a domain's hadith layer can be drawn from *Kitab al-Tibb* rather than from a token probe. ~30 nodes. | | 3 | **The meaning layers do not reach the Book's life-science, medicine and mind surface.** Measured on a conservative Arabic keyword floor over `00-CORPUS/quran.json`: **507 verses / 8.1%** carry L/M/N subject matter. Of those, **426** have a hadith witness (`build/P32-SHAHADAT/W-A/uncapped_links_per_verse.json`, 4,987 verses) and **356** a sirah witness (`W-E/sirah_witness_v2.json`, 3,707) — but only **8** have a Prophetic-stratum dossier (`build/P30-SIGHAH/D-B8-DOSSIERS/dossiers_v1.json`, 90 verses) and **17** a consensus shelf entry (`build/P17-TAFSIR/D1/consensus_*.json`, 146 verses). **24 of 507 = 4.7%.** | all 46 | ABSENT | **HIGH** | **(c)** | **Nodes.** A dossier wave aimed at exactly this 507-verse set. The witness substrate already stands for 426 of them — this is assembly, not acquisition. ~500 nodes, on the P30 B8 pattern. | | 4 | **2,988 tibb-and-health reports are sealed in the house and registered nowhere.** Measured with diacritic normalization over `00-CORPUS/hadith.json` (**1,454 of 36,390**) and `00-CORPUS/hadith_shia.json` (**1,534 of 32,531**), term set = *al-tibb, dawa', tadawa, hijamah, 'asal, al-habbah al-sawda', siwak, humma, ta'un, 'adwa, ruqyah, kuhl/ithmid, rada', hayd/nifas, nutfah/'alaqah/mudghah*. Bukhari's **Kitab al-Tibb (book 76) = 93 reports**, addressable today by the `book` field. No build layer indexes by kitab. | Tibb nabawi, Pharmacology, Public health, Dentistry, Ophthalmology, OB/GYN, Nutrition | ABSENT | **HIGH** | **(c)** | **Engine + nodes.** The kitab index of Finding 2, then a Tibb-nabawi register carrying every report's grade **verbatim from its named grader** — documented, never practiced, never a health recommendation, exactly as family S is chartered. ~200 nodes. | | 5 | **The coverage-matrix instrument is self-contaminating and not idempotent.** `framework/p41_council.py:161` globs `build/*/*.json` and `build/*/*/*.json` into `layer_text`; line ~200 writes `coverage_matrix_v1.json` into `build/P41-COUNCIL/`. Re-running the probe today flips **30 of my 46 areas** from ABSENT to TOUCHED by matching their own names in the file it wrote — verified by running it twice, once with `P41-COUNCIL` excluded (reproduces v1 exactly) and once without (context strings return the matrix's own ABSENT list). | all 46 | n/a | **HIGH** | POLISH | **Engine.** Exclude `build/P41-COUNCIL/**` from `layer_text`, stamp the probe corpus SHA into `_meta`, and assert idempotence in a gate. ~5 nodes. | | 6 | **20 of 46 v1 cells on this bench are wrong, and every error is an over-claim.** Section 1 above; each correction carries the matched token and its context. Net: PRESENT 8→1, TOUCHED 12→7, ABSENT 26→38. Root cause is `probes()` taking the first three tokens longer than 3 characters and substring-matching them — so *"cell by cell"* certifies cell biology, *"public defect log"* certifies epidemiology, *"aquamarines"* certifies marine biology, and the bare word *"documented"* certifies Tibb nabawi. | all 46 | n/a | **HIGH** | POLISH | **Engine.** `coverage_matrix_v2.json` adopting this table, plus a curated probe list **and a negative-control token per area** (a term that must NOT match) so the instrument can fail. ~20 nodes. | | 7 | **132 "Accessibility pass" points are sealed green over pages with no assistive semantics.** Ledger: 132 points titled *"Accessibility pass"*, purpose *"headings, alt text, contrast, screen-reader order"*; tracker: **130 `selftest-green` + 2 `credit-v1`**, evidence line *"…method+a11y+print-fix"*. Measured over **390** HTML files in `build/`: **0** with any `aria-*`, **0** with `
`, **0** with a skip link, **247 (63%)** with no `` at all, **260** pages carrying `` and **0** using `scope=`, **38** pages carrying Arabic script with no `dir=rtl` anywhere. | Special education | ABSENT | **HIGH** | **(a)** | **Engine + nodes.** An a11y gate (G32) asserting `lang`, `dir`, one `
`, `scope=` on every table header, and a skip link — run over the whole tree, failing the seal. ~50 nodes. | | 8 | **The trilingual pages announce the Qur'an in English.** All **127** `pages_trilingual.html` declare `` and carry **0** `lang="ar"` and **0** `lang="ur"` attributes, while `.ar` renders the sealed Uthmani text and `.ur` the Junagarhi translation. `dir='rtl'` is correctly present (127/127) — direction is solved; **pronunciation is not**. A screen reader speaks the Book's own words with English phonology. Additionally **54/127** apply `

` to a Latin-script heading (`

the held translations…`). | Special education, Psycholinguistics | ABSENT | **HIGH** | **(a)** | **Engine.** One line in the page factory: `lang='ar' dir='rtl'` on `.ar`, `lang='ur' dir='rtl'` on `.ur`, and drop `dir` from Latin headings. ~5 nodes. This is the cheapest high-severity fix on the bench. | | 9 | **The whole science-signs inheritance is credited and never landed.** `framework/credit_map.json` asset `research-papers` states the signs inventory *"sits_in P23 — Bayt al-Hikma (the signs inventory, documented-never-adjudicated)"*. The four catalogues at `../03-CATEGORY-5/` hold **216 entries**, of which **58 are life-science or medicine** (Embryology 8, Animal cognition 6, Zoology 4, Origin of life 4, Botany 3, Physiology 3, Entomology 3, Developmental biology 3, Consciousness 3, Genetics 2, Human genetics 2, Neuroscience 2, Medicine 2, Biology 2, Human origins 2, Anatomy 1, and 6 more at 1 each). The strings `CATEGORY-5` and `signs inventory` appear in **exactly one file inside 10-WAQF: `framework/credit_map.json` itself.** | all L, all M | ABSENT | **HIGH** | **(a)** | **Acquisition.** Ingest `03-CATEGORY-5/catalogue.json`, `catalogue-sweep.json`, `catalogue-cat6.json`, `catalogue-expanded.json` (216 entries) into P23 under the existing never-proof law — the entries already carry `claim / field / state / gap / horizon / caution`, which is the exact shape a documented-never-adjudicated register needs. ~120 nodes. | | 10 | **No register of the i'jaz 'ilmi positions — the one place biology and medicine claims about the Book actually live.** The word *miracle* appears in the ledger only as the prohibition *"zero miracle claims"* (**132** points). `build/P23-HIKMA/D2-BRIDGES/bridges_layer.json` carries *"the never-proof doctrine holds"* as one clause of one row of three. The literature itself — its claims and its critics inside the tradition — is documented nowhere. Family S is chartered for exactly this posture; biology and medicine are its largest genre. | Evolutionary biology, Embryology, Marine biology, Physiology, Anatomy | ABSENT | **MED-HIGH** | **(b)** | **Nodes + HUMAN seat.** ~100 nodes on the S-family pattern (positions documented under the tradition's own verdicts, never practiced) and one human seat: a scholar of ulum al-Qur'an with a natural-science training. An unrefuted overreach attributes to the Book what the Book did not say — that is a clause-(b) matter, not a polish matter. | | 11 | **The memorizer's domain landed with no memorization apparatus.** Ledger `n=2401–2500`, C3 domain titled *"the Book's architecture, and **the memorizer's**…"*, 100 points. Its artifact is `build/C3-QURAN/D7-THEMES/structure_layer.json` — **4.6 KB**: 114 per-surah word counts, longest surah, three shortest. No mutashabihat register, no error-prone-transition index, no review schedule — the three things hifz pedagogy needs. | Cognitive psychology, Curriculum design, Islamic pedagogy | ABSENT | **MED-HIGH** | **(a)** | **Engine + nodes.** The near-duplicate detector already exists: `build/P34-OCEAN/matn_dedup_layer.json` fingerprints matn skeletons across 119,261 reports. Point it at the sealed Qur'an text to produce the mutashabihat/confusion register, then build the transition index on top. ~150 nodes. | | 12 | **839 worksheet items across the build carry one prompt and zero marking keys.** Measured across all 130 `education_pack.json`: **839** worksheet items, **839** instances of the single template *"Locate `` in its ayat and read the three translations."*, **0** with an answer key, rubric or marking scheme; **126** packs carry the identical two-question FAQ (*"What is documented in ``?"* / *"How were the counts produced?"*). | Assessment & psychometrics, Curriculum design, Education science | TOUCHED | **MED-HIGH** | POLISH | **HUMAN seat + engine.** A qualified Qur'an-curriculum designer sets the item taxonomy; the engine then emits at least three item types (recall / locate / compare) each with a marking key. ~100 nodes. | | 13 | **The education pillar's own headline counts are stale on its sealed, reader-facing page.** `build/P24-ACCESS/D2-EDUCATION/education_access_layer.json` and its rendered `pages_trilingual.html` state *"**82** education_pack.json across the pillars"* — measured **130**. `D1-LANGUAGES/languages_layer.json` and page state *"**79** pages_trilingual.html"* — measured **127**. D2 states *"children_layer counts proven **353**/0-stale"* — measured **701** items. The layers carry a `built` stamp of 2026-09-02; the later pillars added 48 packs and 48 pages, and the seal re-asserted the old numbers. | Education science, Assessment | PRESENT | **MED** | POLISH | **Engine.** The P31 W9 currency gates (G22–G25) already exist for exactly this failure mode; extend them to P24 so a glob count is recomputed at seal time and a stale number fails the seal. ~10 nodes. | | 14 | **The reading bench has no reliability design and no random probe.** `build/P36-BENCH/verdict_intake.json`: 6-field schema, `verdicts: []`, **no requirement that any row be scored twice**, no anchor or calibration rows, no agreement statistic — while `bench_charter_v2.md` law 5 declares *"disagreement between sitters is DATA"* and nothing computes it. All **1,100** pack rows (packs 001–011) are drawn from *"the strongest exposition-candidates"* (`build/P32-SHAHADAT/W-M/bench_pack_001.json` `_meta`) — a purposive sample that cannot estimate the quality of the 4,987-verse link population it is taken from. | Assessment & psychometrics | TOUCHED | **MED** | **(c)** | **Engine.** A pack 012 drawn at random with a seeded, recorded RNG; a 20-row overlap block repeated across packs; and an agreement computation (Cohen/Krippendorff) that publishes beside the packs. ~50 nodes. | | 15 | **The "children's layer" is not a children's layer.** All **126** blocks are the identical shape `{"item": "