The Word-Count Denominator

RECORD CHAIN

1. Two corpora, independently measured. The case file is 636 pages and 185,417 words, OCR'd from the April 2026 production, which the City rendered through "Microsoft Print to PDF" with no extractable text layer (the format defect carried in Card 47) A R.26-1549 I2. The mailed corpus is 554 pages and 142,152 words; the OCR run completed all 554 pages with 0 processing errors I5 I3. The two streams — the filed case file and the separately mailed pages — total 1,190 pages and 327,569 words. The mailed corpus is dominated by recurring forms; I3 documents no inspector-authorship analysis of it beyond the case-file authorship reports.

2. The inspector-authorship pass has raw and adjusted totals. A page-by-page reading of the April production classified every Paul-attributed entry as template, pasted, or original. The raw original total is 1,954 words across 14 entries I2 I1. Two entries — 433 words total, the 01/12/2026 Starbucks-meeting note and the 01/27/2026 "neighbor to the West" note — were excluded because they rest on a disputed third-party-relayed source, leaving 1,521 words across 12 entries as the conservatively attributable total.

3. The extended property-condition narrative begins at the August 21, 2025 access, and later additions are minimal. Between the November 2025 and April 2026 productions — five months — only 32 words across 2 entries (01/26/2026, 04/02/2026) are conservatively attributable new Paul authorship, 0.017% of the case file I2 I1. The 1,521 words divide at August 21, 2025, the date a compelled backyard inspection put the inspector on the property after two years and five months Card 10. Sub-periods follow the analysis window in I6 and begin at the first Paul entry, 03/20/2023; the three days back to the 03/17/2023 window start contain no authorship:

• Before 08/21/2025 (03/20/2023–08/20/2025, 884 days): 6 entries / 650 words = 0.74 words/day.
• On and after 08/21/2025 (08/21/2025–05/16/2026, 268 analysis-window days): 6 entries / 871 words = 3.25 words/day — a 4.4x increase.

The single 08/21/2025 entry is 589 words — about 39% of the conservatively attributable authored output across the 1,156-day analysis window, and the first extended backyard/interior condition narrative following documented access I1 M035 I6.

4. After removal or swapping of identity fields, 19.1% of tested lines remain, and the residue is process and accounting language. In the mailed corpus, 99,922 of 140,023 line-counted words (71.4%) are recurring stock language I3; the line-counting method yields the 140,023 denominator, while the 142,152-word corpus total stated above comes from a separate counting pass in the same report. I3 identified 36 multi-version form/title groups, and the line "sacramento ca 95819" recurs on 313 of 554 pages I3. After identity redaction, removal, or swapping of recipient and property identity fields, 2,419 of 12,684 tested lines (19.1%) survive. The surviving residue is process and accounting language: fee/invoice/cost 1,145, penalty/order 468, hearing/appeal 429, lien/assessment 342, notice posted/served/mailed 171, inspection/reinspection 57, and 63 lines in the violation/correction category I4. Each penalty order recites the same Level C provision, SCC.1.28.010, addressed in Card 15.

FULL CIRCLE

The measurement has three steps: two independently OCR'd paper streams spanning the 1,156-day analysis window I2 I5 I3 I6; a page-by-page authorship pass producing the raw and adjusted Paul totals I1 I2; and an identity-swap test showing that the surviving residue is invoice, penalty, hearing, lien, notice, and process language rather than property-specific description of 4880 T Street I4.

The City may answer that word count is not a legal standard, that administrative files ordinarily contain automated forms, and that findings can appear in structured activity logs rather than free-form narrative. This card does not treat automation as improper. On the record, however, the structured-log residue isolated by the identity-swap test largely describes case processing, and the detailed backyard/interior authored narrative begins with the 08/21/2025 access-day entry M035 I1.

Across the analysis window, the produced and mailed paper averages 283 words per day, while the conservatively attributable inspector-authored share averages about 1.3 words per day. Related cards carry the legal consequences: Card 10 and Card 11 the fee ledger, Card 15 the unselected penalty predicate, Card 14 and Card 31 the later rate and permit-scope issues, and Card 12 the title-cloud record. Card 52 supplies the denominator those cards rely on.

The method is internal OCR and authorship analysis over City-produced and City-mailed records, and it is open to examination on its own terms. The finding is correspondingly narrow: it measures volume and authorship, and it places the start of the detailed backyard/interior authored narrative at the date documented access begins.