A specification for structural decision evaluation using the Target Alignment Criteria (TAC) framework
Decision Audit is distinct from prediction, scoring, explainability, and risk modeling. It does not evaluate whether a decision will succeed. It evaluates whether a decision is structurally ready to be executed.
A decision may be supported by accurate data and still fail structurally — when its evidence layers contradict each other, point toward incompatible outcomes, or fail to converge under pressure. Decision Audit is designed to detect and surface these structural failures before resources are committed.
A Decision Audit is composed of three structural elements:
A clear, specific, executable proposition that defines the action under evaluation. The decision statement must be verifiable — it must be possible to determine whether it has been executed.
Independent sources of structured evidence that bear on the decision. Each layer represents a distinct epistemic or domain perspective. A minimum of two evidence layers is required for a valid Decision Audit. Layers must be independent — if one layer is derived from another, they do not constitute separate evidence.
The execution environment, constraints, jurisdiction, timeline, or domain-specific conditions that govern how evidence layers interact. Context is optional but improves evaluation precision.
The following defines the canonical syntax for a Decision Audit submission. This syntax is machine-readable, human-readable, extensible, and domain-agnostic.
DECISION: <decision statement> LAYER 1: <layer name> TYPE: <mechanistic | clinical | strategic | legal | financial | behavioral | ...> EVIDENCE: <structured claim or dataset summary> LAYER 2: <layer name> TYPE: <layer type> EVIDENCE: <content> [LAYER n: ...] CONTEXT: <jurisdiction, timeline, constraints> [optional] TAC EVALUATION: ALIGNMENT: <0–100> TENSION: <0–100> CONVERGENCE: <0–100> ELIGIBILITY INDEX: <computed value> VERDICT: <ELIGIBLE | ELIGIBLE WITH CONDITIONS | DEFER | RESTRUCTURE>
mechanistic, clinical, financial, operational, legal, statutory, precedent, strategic, behavioral, market.The TAC evaluation model assesses three structural dimensions across the evidence assembly:
Measures whether evidence layers support a coherent, unified objective without directional conflict. High alignment indicates that all layers point toward the same outcome. Low alignment indicates that layers are pulling the decision in incompatible directions.
Measures the degree of contradiction or structural conflict between evidence layers. Low tension (high score) indicates that layers are mutually reinforcing. High tension (low score) indicates adversarial or contradictory evidence structure.
Measures whether the decision structure resolves toward execution readiness across evaluated scenarios. High convergence indicates stability under variation. Low convergence indicates fragility — the decision holds only under narrow or optimistic conditions.
The Eligibility Index (E) is a composite score computed from the three TAC dimensions:
The default weighting assigns highest priority to Alignment (0.40), as directional coherence is the most fundamental structural requirement. Tension (0.35) is weighted heavily because adversarial evidence is the most common source of decision failure. Convergence (0.25) captures cross-scenario stability.
The following verdict classifications are defined by the TAC Default Decision Eligibility Threshold Model:
| Verdict | Eligibility Index | Structural Interpretation |
|---|---|---|
| ELIGIBLE | E ≥ 72 | Evidence assembly is structurally coherent. Decision is supported for execution. |
| ELIGIBLE WITH CONDITIONS | 48 ≤ E < 72 | Moderate structural tensions exist. Addressable with targeted risk mitigation prior to execution. |
| DEFER | 28 ≤ E < 48 | Significant structural incompatibilities. Decision requires substantial restructuring before execution. |
| RESTRUCTURE | E < 28 | Fundamental evidence contradictions. Decision premise requires reconception. |
Decision Audit applies to any high-stakes decision where evidence layers must be evaluated for structural coherence prior to commitment. Reference use cases include:
Evaluating clinical trial advancement decisions across mechanistic, genetic, and clinical evidence layers before capital or patient commitment.
Evaluating investment thesis structural integrity across financial, operational, and market layers before capital deployment.
Evaluating legal argument structural coherence across statutory, precedent, and risk layers before filing or execution.
Evaluating organizational decisions across resource, risk boundary, alignment, and execution horizon layers before commitment.
The following systems implement the Decision Audit standard:
DECISION: Invest $10M in Series B biotech startup developing
CRISPR-based cancer therapeutics, 5-year hold.
LAYER 1: Financial
TYPE: financial
EVIDENCE: $8M ARR, 3x YoY growth, burn rate $1.2M/month,
18 months runway.
LAYER 2: Clinical
TYPE: clinical
EVIDENCE: Phase 2 trial ongoing, FDA Fast Track designation,
team of 45 with 3 PhDs.
CONTEXT: PE investment, board approval required.
TAC EVALUATION:
ALIGNMENT: 65
TENSION: 70
CONVERGENCE: 55
ELIGIBILITY INDEX: 63
VERDICT: ELIGIBLE WITH CONDITIONS
DECISION: File for injunctive relief against former employee
for breach of non-compete agreement.
LAYER 1: Statutory
TYPE: statutory
EVIDENCE: California Business & Professions Code §16600
prohibits restraints on trade.
LAYER 2: Precedent
TYPE: precedent
EVIDENCE: Edwards v. Arthur Andersen LLP (2008) — California
Supreme Court held non-competes are void.
CONTEXT: US Federal Court, SDNY.
TAC EVALUATION:
ALIGNMENT: 70
TENSION: 0
CONVERGENCE: 72
ELIGIBILITY INDEX: 31
VERDICT: DEFER