# 31 — Final Case File Word Count and Authorship Report
**Case:** 23-009185 — 4880 T Street, Sacramento
**Run:** `forensic_production_comparison_2026-05-08`
**Sealed (UTC):** 2026-05-08T17:26:37Z
**Status:** CANON — operator-approved. Wraps the word-count and authorship phase of this run.

This document supersedes the earlier draft `CASE_FILE_WORD_COUNT_REPORT_2026-05-08.md` (removed) and consolidates the locked baselines `28_…`, `29_…`, `30_…` with the today-verified totals into one final, citable artifact.

---

## 1. Inputs (verified this run)

| Production | File | Size (bytes) | SHA-256 | Pages |
|---|---|---:|---|---:|
| **NOV** (Production A — Nov 5–6, 2025) | `case-file-23-009185_2025-11-06_city-produced-redacted.pdf` | 52,451,642 | `eb324451b81cf67af2a792b0432cbae7abe89b93f3b75f0baf8b9f9bc1f223d5` | 590 |
| **APR** (Production B — Apr 20, 2026) | `23-009185_Redacted_Altered_2026-04-20.pdf` | 218,448,538 | `ff6c83546627aacf46e3ce7a413d2e8e22e57ae91b8c1c96877bb3251088f600` | 636 |

The NOV SHA-256 matches the value pinned in `CLAUDE.md` SECTION 2 — file integrity confirmed.

## 2. Methodology (standardized, per `HANDOFF_2026-05-08.md`)

- **Word count:** whitespace-split (`wc -w`) over per-page extracted text on disk.
- **NOV source:** `extracted_text/nov/page_NNN.clean.txt` — Citizenserve native text-layer extraction (clean copy).
- **APR source:** `extracted_text/apr/page_NNN.clean.txt` — Tesseract `--psm 11` OCR of 200 DPI page renders, with `|`→`I` normalization and stripped table-cell labels (`Lovato`, `INSPECTION`, `Information`, `Activity Note`, `General Case`) baked into the clean files.
- **Activity-log structure:** NOV's compact activities table sits on pp 1–10 with a verbose General Case Information section on pp 21–34. APR shifts the GCI section to pp 25–39 (a +4-page offset driven by documents-inventory growth).
- **Lovato authorship classification (TEMPLATED / PASTED / ORIGINAL):** as defined in `HANDOFF_2026-05-08.md`. `28a_*.csv` is the per-row record.

## 3. Total case file word counts (measured 2026-05-08)

| Metric | NOV | APR | Δ (APR − NOV) | Δ % |
|---|---:|---:|---:|---:|
| Pages | 590 | 636 | **+46** | +7.80 % |
| Words (clean) | **165,446** | **185,417** | **+19,971** | +12.07 % |
| Characters (clean) | 1,080,979 | 1,177,598 | +96,619 | +8.94 % |
| Lines (clean) | 22,228 | 58,752 | +36,524 | +164.32 % |
| Pages with <5 words (near-blank) | 1 | 0 | −1 | — |

The lines delta is an artifact of OCR per-fragment line emission in APR versus native text extraction in NOV; word count is the comparable metric across the two pipelines.

### Reconciliation with prior audit figure

`28_NOV_PAUL_WORK_ANALYSIS.md` reports the NOV total as **165,487 words**. The today-measured NOV total is **165,446 words** — a delta of −41 words (0.025 %). This delta is documented in row 2 of the verification table inside file 28 and is within the cleanup-step tolerance the prior audit allowed. No reconciliation action is required; both figures refer to the same NOV production.

The APR figure **185,417 words** in `30_APR_PAUL_LOVATO_OWN_WORDS.md` reproduces this measurement exactly.

## 4. Lovato authorship — locked baselines

Numerators below are the operator-approved figures from files 29 (NOV) and 30 (APR). Denominators are the §3 measured totals. Percentages are recomputed against those denominators.

### 4.1 NOV Production (10 entries, 1,489 words)

Source: `29_NOV_PAUL_LOVATO_OWN_WORDS.md` (sha256 `d892354126d9…329c8fc`).

| Date | Words | NOV pages |
|---|---:|---|
| 03/20/2023 | 116 | 21 |
| 04/07/2023 | 62 | 21 |
| 04/10/2023 | 51 | 1, 21 |
| 04/11/2023 | 50 | 21 |
| 03/28/2025 | 74 | 29 |
| 07/17/2025 | 297 | 31 |
| 08/21/2025 | 589 | 32 |
| 09/16/2025 | 146 | 33 |
| 10/20/2025 | 67 | 33 |
| 10/22/2025 | 37 | 10, 34 |
| **Total** | **1,489** | |

**1,489 / 165,446 = 0.900 %** of the NOV case file.

### 4.2 APR Production — raw (14 entries, 1,954 words)

Source: `30_APR_PAUL_LOVATO_OWN_WORDS.md` (sha256 `725a29afefec…c12a013`). Composition: 10 entries / 1,489 words preserved verbatim from NOV, plus 4 new entries / 465 words.

**1,954 / 185,417 = 1.054 %** of the APR case file.

### 4.3 APR — credibility-adjusted (defensible: 12 entries, 1,521 words)

Two of the four new APR entries rely on Farhad-population neighbor-relayed narrative. The landlord of 4870 T Street (Steven) confirmed in SMS — operator-held evidence outside this folder — that Farhad was a short-term tenant being evicted and that multiple tenants and neighbors had ongoing problems with him. On that basis, the disputed entries are removed:

| Date | Words | APR page | Disposition |
|---|---:|---:|---|
| 01/12/2026 | 269 | 38 | DISPUTED — Starbucks meeting; Farhad-population sourcing |
| 01/26/2026 | 19 | 38 | DEFENSIBLE — permit-expiration process note |
| 01/27/2026 | 164 | 38 | DISPUTED — "neighbor to the West"; same disputed-source population |
| 04/02/2026 | 13 | 12 | DEFENSIBLE — siding inspection approved on permit RES-2603471 |

Defensible total: 1,489 (NOV-preserved) + 32 (defensible new) = **1,521 words / 12 entries**.

**1,521 / 185,417 = 0.820 %** of the APR case file.

### 4.4 APR — defensible new-since-NOV only

**32 words across 2 entries** (01/26/2026 and 04/02/2026) — **0.017 %** of the APR case file. This is the defensible quantum of original Lovato work product across the five months between the two productions.

## 5. Summary table

| Slice | Entries | Words | Denominator | % of case file |
|---|---:|---:|---:|---:|
| NOV — Lovato original | 10 | 1,489 | 165,446 | 0.900 % |
| APR — Lovato original (raw) | 14 | 1,954 | 185,417 | 1.054 % |
| APR — Lovato original (defensible) | 12 | 1,521 | 185,417 | 0.820 % |
| APR — Lovato defensible NEW since NOV | 2 | 32 | 185,417 | 0.017 % |

## 6. Provenance and dependent-files manifest

| File | SHA-256 | Role |
|---|---|---|
| `28_NOV_PAUL_WORK_ANALYSIS.md` | `381534b6e73a…967a5796` | Verbatim copy of operator's prior NLP audit + strengthened enumerative verification block |
| `29_NOV_PAUL_LOVATO_OWN_WORDS.md` | `d892354126d9…329c8fc` | NOV Lovato originals, full verbatim text |
| `30_APR_PAUL_LOVATO_OWN_WORDS.md` | `725a29afefec…c12a013` | APR Lovato originals, full verbatim text + credibility analysis |
| `HANDOFF_2026-05-08.md` | `387eb890048b…0227430` | Run handoff, methodology, do/do-not list |
| `28a_nov_activity_log_renderings.csv` / `_distinct.csv` / `_gci_classified.csv` | (not hashed here) | Per-row source records for the 292-entry NOV GCI corpus |
| `29a_apr_activity_log_renderings.csv` / `_distinct.csv` | (not hashed here) | Per-row source records for the APR GCI corpus |

External evidence relied on by §4.3 only:
- **Steven SMS thread** — operator-held; confirms Farhad short-term tenant / eviction / neighbor disruption. Outside CROSS_EXAM by design (per handoff §"Steven / Farhad situation").

## 7. Scope and non-claims

**This document asserts (PROVEN at the file-on-disk level):**
- The two PDFs at the paths in §1 have the sizes, page counts, and SHA-256 hashes shown.
- The clean per-page text files on disk, summed across all pages, contain the word/character/line counts in §3.
- The Lovato numerators in §4 (10 / 1,489 NOV; 14 / 1,954 APR raw; 12 / 1,521 APR defensible; 2 / 32 defensible-new) are taken from operator-approved files 29 and 30 without modification.

**This document does not assert:**
- That the underlying PDFs contain exactly the word counts shown — these are *extracted* text counts (native NOV; OCR APR). Image-only content, redacted regions, and OCR misrecognition are folded into the totals by definition.
- That the +46 page / +19,971 word delta corresponds to substantive new case-file content. Locating and characterizing the deltas is Tier-2 cross-examination work, out of scope for this report.
- Anything about authorship of the non-Lovato 99.18 %–99.98 % of the case file. That mass is system-generated workflow, hearing-scheduling form letters, fee-cycle machinery, and templated/pasted content, per the methodology of files 28–30.

## 8. What would disprove this report

- A re-extraction producing materially different per-page word totals (>1 % drift on either production) would invalidate §3.
- A SHA-256 mismatch on either PDF against §1 would invalidate the entire report.
- A correction to the locked Lovato baselines in files 29 or 30 would change §4 numerators (denominators in §3 unaffected).
- Operator withdrawal of the Steven SMS reliance would re-attach the 433 disputed words to the APR defensible total, changing §4.3 only.

---

## Plain-list close-out

- **Proven findings:**
  - NOV: 590 pages, 165,446 words. APR: 636 pages, 185,417 words.
  - APR exceeds NOV by 46 pages and 19,971 words.
  - Lovato original work product: 1,489 words / 10 entries on NOV (0.900 %); 1,954 words / 14 entries on APR raw (1.054 %).
- **Strong inferences:**
  - The 165,487 figure in file 28 and the 165,446 figure measured today refer to the same NOV production; the 41-word delta is cleanup-step tolerance.
  - APR defensible Lovato work product, given the Steven SMS, is 1,521 words / 12 entries (0.820 %), of which only 32 words / 2 entries are new since NOV (0.017 %).
- **Unconfirmed issues:** Where in APR the +46 pages and +19,971 words fall (Tier 2 — out of scope).
- **Required next records:** None for this report. Future cross-examination work that revisits §4.3 needs the Steven SMS thread on hand.
- **Highest-value next action:** Begin (when operator approves) the Tier-1/Tier-2 cross-examination using files 28 / 29 / 30 / 31 as the locked baselines.

— End of report. This file closes the word-count and authorship phase of run `forensic_production_comparison_2026-05-08`.