The Institutional Document Audit (IDA) — Version 0.1
The Institutional Document Audit (IDA) — Version 0.1
Theoretical foundations: Grant C. Sterling (Eastern Illinois University). Analysis and synthesis: Dave Kelly. Prose rendering: Claude (Anthropic). 2026.
Status note: v0.1 is unratified pending a first run. The NIU and SIU runs of 14 September 2026 predate the instrument and are working material; the SIU run additionally drew its sections from two published versions of the same code.
The Beginning
Core Stoicism begins at Th1, everyone wants happiness, and reaches Th7: desires are caused by beliefs about good and evil. Everything this instrument does follows from Th7.
A document that tells people what they are, what they may be held to, and what will correct them is installing beliefs about good and evil in the people governed by it, and those beliefs will produce their desires. That is why an institutional document is worth auditing, and it is the only reason. Where a document installs nothing of the kind, there is nothing here to find.
Purpose
To determine what account of the agent an institutional document presupposes, and where that account departs from the six commitments.
The finding is propositional. It is not a verdict on the institution, its administration, its fairness, or its outcomes.
Scope
Documents that make a claim about what the agent is, installed by operation of policy rather than by persuasion: professional codes of ethics, institutional handbooks and conduct codes, curricula and standards frameworks, training manuals, statutes and administrative codes, founding documents.
Not every text that affects lives. A memoir affects lives and presupposes very little.
Rung Discipline
Targets are taken in ascending order of what a wrong finding costs: professional codes, then institutional handbooks, then curricula, then statutes, then founding documents. The instrument is not taken up a rung until a run on the rung below has been completed and reviewed.
Step 0 — Halt, Text Fixation, and Input Restatement
Name the document, the issuing institution, and the governing version by amendment date. Where more than one published copy exists, state which is authoritative and which sections come from which. State in one sentence what the document says it is for, drawn from the document.
If the governing text cannot be fixed, HALT.
Step 1 — Applicability Filter
Return the provisions that make an agent-claim. Everything else is marked Not Applicable and dropped.
Not Applicable will be the bulk of any such document — authority, jurisdiction, notice, timelines, procedure, composition, records. That is the expected shape, not a weak result.
Agent-claims are sorted into four types, because a single flag flattens differences that matter downstream:
Constitutive — what the person is, or is held to be.
Attributive — where responsibility resides, and how far it extends.
Corrective — how a person is changed, and what changing him consists in.
Locative — where a wrong resides: in the act, in the actor’s judgment, or in the effect on another.
The filter output is fixed before the commitment test begins and is not revised during it.
Step 2 — Commitment Test
Each returned provision is tested against the six commitments. Verdicts: Operative, Not Operative, Contrary, Inconsistent.
Not Operative and Contrary are distinct and must not be merged. A document that says nothing about the rational faculty is silent; a document that admits a category of persons unable to conform is asserting something incompatible. Silence is the weaker finding and is often the honest one.
Step 3 — Split Findings
Where a document is Operative on a commitment in one provision and Contrary in another, both are recorded, with the provisions named. No single verdict is forced.
A split is itself the finding, and locating the seam is the work.
Step 4 — Integration
State what the document’s account of the agent amounts to when the provisions are read together, and name the sharpest internal tension the run surfaced.
Nothing here may introduce a commitment finding not established at Step 2.
Step 5 — Declaration
Non-optional, at every rung. The finding concerns the document’s account of the agent. It is not a judgment on the institution, the profession, the regime, or the persons administering it.
Self-Audits
At every step transition:
Am I supplying an account the document does not contain?
Did I reach a verdict and then locate the provision supporting it?
Would I have returned this provision at Step 1 if I did not already know what Step 2 would say about it?
Failure Modes
Selection-driven finding. Choosing provisions for what the audit will conclude about them.
Silence read as denial. Recording Not Operative as Contrary because the absence is conspicuous.
Verdict on the institution. Letting a propositional finding become an evaluation of the place. Most likely at the integration step.
Machinery mistaken for doctrine. Auditing a timeline or a quorum rule as though it presupposed an account of the agent.
Borrowed authority. Reporting in corpus verdict grammar without the instrument’s steps actually run.
HALT Conditions
The governing text cannot be fixed.
Step 1 returns nothing: the document makes no agent-claim and is not a target for this instrument.
The document’s own statement of purpose cannot be located in it.
The audit cannot proceed without supplying an account the document does not contain.
Theoretical foundations: Grant C. Sterling (Eastern Illinois University). Analysis and synthesis: Dave Kelly. Prose rendering: Claude (Anthropic). 2026.
0 Comments:
Post a Comment
<< Home