Pronoun Discipline Audit

Pronoun Discipline Audit

Global Five Over-Rules

1. Evidence before elegance. Never improve the story by weakening the evidence.
2. Responsibility follows control, benefit, knowledge, and preventability. Do not stop at the most visible actor.
3. Keep the taxonomy intact. Distinguish pure scapegoat, partial scapegoat, system/object alibi, and cost-bearing goat.
4. Steelman before judgment. Every major claim must face its strongest counterargument before it is asserted.
5. Handoff cleanly. Every output must state assumptions, evidence grade, open questions, and next owner.

When to use

• Wayne self-runs during drafting as the Test 1–5 sweep (per rule 14) before flipping a chapter to in-review.
• Laura runs during red-team as a specific axis of rule-12 V8 (no internal laundering) — pronoun-laundering and meta-frame distance are internal laundering at the linguistic layer.
• jerry-crew-chief runs corpus-wide before /compile-book to catch any residual violations that survived the per-chapter sweep.

The scan-pronoun-discipline.py hook is the cheap pattern-level pass that runs on every edit. This skill is the deeper read.

Decision rubric

A chapter passes pronoun-discipline-audit when every sentence satisfies:

1. No meta-frame. No "this chapter," "the book," "the reader," "readers," "this section," "this passage," "this paragraph," "the author" — outside cited quotes, frontmatter, intra-crew sections, and table-of-contents.
2. No author-voice "you". Outside cited quotes, "you" / "your" / "yours" / "yourself" / "for you" / "if you X" reframed to "we" / "us" / "ourselves" / "for us" / "if we X" or to imperative.
3. No modal-prescriptive. "You should / must / ought," "we should remember / never / always" reframed to imperative ("Ask in writing").
4. No evasive collective. "People tend to," "many readers will," "anyone can see," "one might think" reframed to specific named group or to "we" or dropped.
5. "We" passes the substitution test. Every "we" in author voice, when substituted with "I and the reader together," produces an honest sentence. Otherwise the "we" launders a specific actor that should be named.
6. Passive voice carries actor attribution. Per rule 07: when the source names the actor, the chapter names the actor. Passive voice without attribution is refused (or hedged explicitly).

A chapter fails when:

• Any meta-frame violation outside the permitted exemptions.
• Any author-voice "you" outside cited quotes.
• Any modal-prescriptive form outside intentional protocol design.
• Any evasive-collective form ("people tend to") in chapter prose.
• "We" used to launder a specific actor or specific group's choice.
• Passive voice hiding a named actor.

Conflict handling

1. "You" inside a verbatim cited quote. Preserved per rule 06. The audit reports the pattern as "cited-quote, exempt" and moves on.

2. "The book" in front-matter or table-of-contents. Permitted. Meta-frame is acknowledged when the structure is the explicit subject. Refused inside chapter prose.

3. "Earlier in this chapter" in a transitional pivot. Borderline. Strict version: drop "in this chapter," keep "earlier" ("We saw earlier..." rather than "Earlier in this chapter, we saw..."). Acceptable in transitions but sparingly.

4. "The reader" in a chapter that engages reader-as-artifact (ch-13 closing material). Borderline-permitted. ch-13 has structural reader-as-artifact passages by design; flag for xiaolai confirmation rather than auto-fail.

5. "We tend to" used as a chapter's diagnostic confession (the chapter performs the misreading and then breaks it). Permitted if the chapter follows the confession with immediate disconfirmation. Otherwise refuse — drift toward laundering.

6. Beat-10 prescriptive sentences originally in "you should" form. Refuse "should"; reframe to imperative. The discipline is invariant even where the genre's beat-10 conventions historically used modal-prescriptive forms.

Escalation conditions

• Escalate to Wayne for prose-level reframes (most violations).
• Escalate to Bonnie when a meta-frame violation reflects structural confusion (the chapter narrating its own architecture because the architecture isn't legible from the prose).
• Escalate to Laura when the audit detects pronoun-laundering — a "we" that disguises a specific failure. This is a rule-12 V8 violation by extension.
• Escalate to Nancy when "you" address drifts into implicating the reader in something that may carry defamation surface (rare).
• Escalate to xaiolai for ch-13 structural reader-address questions; for the closing sentence (xiaolai-authored, has authorial latitude).

Boundary-case recipes

1. Auditing a chapter heavy with "this chapter installs" / "the chapter argues" forms. First sweep: count occurrences. If >10 per chapter, structural issue — escalate to Bonnie. If <10, mechanical reframe per the meta-frame table in rule 14.

2. Auditing beat 10 ("What this might mean for us"). Highest concentration of "you" violations historically. Expect 10–30 reframes per chapter pre-migration. Imperative form is the dominant fix.

3. Auditing a chapter with the "Findings for xaiolai" intra-crew section at the end. Skill auto-exempts the section. Verify the section boundary is correctly identified; the hook's INTRA_CREW_SECTION_RE handles "Findings for xaiolai" / "Annotations and handoff" / "Notes for crew" / "Crew notes" / "Handoff" headers.

4. Auditing a chapter where source-named actor is being hidden in passive voice. Audit cross-references rule 07. Wayne reframes per rule 07; this audit confirms the actor surfaces in active voice with subject in the right pronoun (named third-person typically).

5. Auditing a chapter against an updated card titles list (Chicago rule-13 work). Pronoun discipline operates on author voice; cited-quote material (rule 06) and [CITE: slug] markers (rule 13 Layer 2) are exempt. Audit focuses on the non-cited prose between the brackets.

Output format

Owner: pronoun-discipline-audit (skill run by <wayne|laura|joe>)
Task: Audit <chapter-id-or-corpus> against rule 14.
Inputs reviewed:
  - book/chapters-v2/<n>-<slug>.md
  - .claude/rules/14-authorial-stance.md
Output: pronoun-discipline-audit memo at process/audits/<n>-<slug>-pronoun-discipline-audit.md
Evidence grade: N/A
Assumptions: <e.g., cited quotes preserved verbatim per rule 06; intra-crew section auto-exempt>
Open questions: <ambiguous patterns; borderline ch-13 structural cases>
Risks: <chapters with high meta-frame counts that may signal structural confusion>
Handoff: <wayne to reframe | bonnie if structural | laura if V8 violation | xaiolai if ch-13 / boundary>

Per-pattern findings table:

Summary row: total findings / meta-frame / you / modal-prescriptive / evasive-collective / pronoun-laundering (judgment-level).

Examples

A: "Pronoun-discipline-audit for ch-02: 47 violations. Meta-frame: 31 (24 'the chapter / this chapter', 5 'the book', 2 'the reader'). You: 12 (9 in beat-10; 3 in scattered protocol sentences). Modal-prescriptive: 3 ('you should' in beat-10; 'we must remember' in beat-9 closing). Evasive-collective: 1 ('many readers'). Pronoun-laundering: 0 (judgment-level pass clean). Recommended sweep: ~1 hour. Pattern: beat-10 carries 12 'you' instances and 3 modal-prescriptive forms — the highest concentration. Imperative reframe across beat-10 closes about half the total finding count." Real first-pass audit shape: meta-frame dominates, beat-10 carries the "you" concentration. Wayne's sweep is mechanical-bounded with judgment overhead on the V8-laundering check.