Intent Audit¶
Procedure — instantiates Function-Without-Intent Caution
Reviews records, incentives, choices, and communications for evidence that a function was deliberately pursued or knowingly maintained.
Intent Audit is the evidence-gathering procedure that asks a factual question a description alone can never answer: did the actors mean to do this, and did they keep doing it once they knew? It goes outward into the record — decision memos, incentive structures, meeting minutes, private messages, budget approvals, complaints that were logged and then filed — and hunts for concrete traces of deliberate pursuit or knowing maintenance. Its defining move is that it treats intent as an empirical claim requiring behavioral evidence, and it sharply separates two moments: original intent (did someone set out to create this function?) and maintenance intent (once the function became visible, did actors choose to preserve it?). An audit can find weak evidence of the first and strong evidence of the second — a distinction that matters enormously for accountability.
Example¶
A retail company's annual bonus scheme has, over five years, paid out far more to one department than to comparably-performing others. An Intent Audit is commissioned. The auditor does not describe the pay gap (that description already exists) — she looks for evidence of deliberateness. On original intent she finds little: the formula was copied from a template and the disparity appears to be an artifact of how one metric was weighted. But on maintenance, the record is damning: two years in, a compensation analyst flagged the skew in a memo; a manager replied that "changing it now would upset the top team"; the flag recurred the next year and was again shelved. She assembles a maintenance choice record — a dated chain of who knew, when, and what they chose — and an intent evidence check that scores each claim: deliberate original design: weak; foreseeable disparate effect: strong; knowing maintenance: strong. The finding reframes the case entirely: not "someone built a discriminatory bonus" but "someone built a careless one and then repeatedly declined to fix a known problem."
How it works¶
- Fix the specific intent claim. Turn "they intended it" into a testable proposition ("the 2021 formula was chosen to favor Team A") before searching for evidence — otherwise the audit confirms whatever it started believing.
- Separate origin from maintenance. Build two evidence files: one for deliberate creation, one for choices made after the function became knowable. Score them independently.
- Weight the sources. Contemporaneous documents and incentive structures outrank after-the-fact testimony; a logged-then-ignored warning is the single strongest maintenance signal.
- Timestamp the choice chain. Record who knew, when, and what they did — the maintenance choice record is the audit's most reusable output.
Tuning parameters¶
- Evidence bar — how strong a trace must be before "deliberate" is written down. Higher protects against unfounded accusation; too high lets knowing maintenance hide behind deniability.
- Origin-vs-maintenance emphasis — how much effort goes to each. Historical origin can be unknowable while maintenance is fully documented; shifting weight toward maintenance often recovers the more actionable finding.
- Adversarial access — whether the audit can compel documents or only see what is volunteered. The gap between the two is itself a finding.
- Charity setting — how readily ambiguous evidence is read as incompetence versus design. Merton's dictum applies: never attribute to malice what negligence explains, but never let negligence launder knowing maintenance.
When it helps, and when it misleads¶
Its strength is that it converts a purpose accusation into a checkable claim and, crucially, recovers the maintenance finding that a pure origin hunt would miss — the difference the law draws between disparate treatment and disparate impact.[n1] Its failure mode is that intent evidence is sparse, self-serving, and easily destroyed, so absence of a smoking gun gets misread as proof of innocence; conversely, a zealous auditor can read ordinary hedging as conspiracy. The classic misuse is stopping at "no original intent found" and declaring the matter closed — the exact "lack-of-intent excuse" the archetype warns against. The guarding discipline is to always run the maintenance file even when the origin file comes up empty, and to report both evidence bars explicitly rather than collapsing them into a single verdict.
How it implements the components¶
intent_evidence_check— its core operation: scoring, per claim, whether behavioral and documentary evidence supports deliberate design, motive, or foreseeability.maintenance_choice_record— its most durable artifact: a dated chain of who knew about the function and what they chose to do once it was visible.
It does NOT inspect an explanation's internal logic for a missing causal mechanism (causal_pathway_context, explanation_update) — that's Root-Cause Explanation Review, its same-type sibling; Intent Audit reads the external record for what actors actually did, not the reasoning about how the function is produced.
Related¶
- Instantiates: Function-Without-Intent Caution — Intent Audit supplies the intent-evidence side of the analysis.
- Consumes: Evidence Threshold Table — tells the audit how much evidence each intent claim requires before it can be written down.
- Sibling mechanisms: Functional Analysis · Institutional Genealogy · Design History Review · Root-Cause Explanation Review · Attribution Check · Current Use Mapping · Evidence Threshold Table
Editorial Notes¶
Form Classification¶
Form family: Assessment, Review & Assurance
Rationale: Intent Audit operates as a bounded evaluation of existing evidence or work that produces a finding or disposition because it reviews records, incentives, choices, and communications for evidence that a function was deliberately pursued or knowingly maintained
Independent corroboration: The frozen evidence defines Intent Audit as 'Reviews records, incentives, choices, and communications for evidence that a function was deliberately pursued or knowingly maintained', so its operative form is Assessment, Review & Assurance.
Review outcome: Independent reviewer agreement; high confidence.
Origin Attribution¶
Primary origin: Law & Governance
Origin pattern: Cross-disciplinary synthesis
Present-day reach: Multi-domain
Rationale: Establishing purpose, knowledge, and continued maintenance from documentary and behavioral evidence is fundamentally a legal inquiry into intent.
Related originating lineages:
- Criminology & Forensic Studies — Forensic investigation materially contributes evidence collection, corroboration, and distinction between assertion and supported inference.
- History & Historiography — Archival reconstruction materially helps separate original purpose from later knowing maintenance.
- Philosophy — Action theory and the distinction between intended function and mere effect supply the conceptual foundation.
Review resolution: Both independent reviews place the primary lineage in law_governance. The queued differences (alternate_origin_disagreement, origin_mode_disagreement, encyclopedia_synthesis_disagreement) concern secondary metadata rather than primary provenance. The final retains criminology_forensic, history_historiography, philosophy only where a reviewer supplied a formative-lineage rationale; this does not convert downstream applicability into origin. origin_mode=cross_disciplinary_synthesis because the entry's present form deliberately composes methods from the documented lineages. domain_reach=multi_domain records application breadth separately from provenance.
Encyclopedia synthesis: The exact catalogued form synthesizes established practice rather than reproducing a single standard historical label.
Review outcome: Reconciled after independent review; high confidence.
Notes¶
[n1] In anti-discrimination law, disparate treatment (deliberate, intent-based) is distinguished from disparate impact (a neutral rule with an unequal effect, no intent required). The pairing is the legal encoding of exactly this archetype's caution: harmful effect and harmful intent are separate claims with separate evidentiary demands. ↩