§02d Jupiter — Discrete Composition — Subject-directed determinations
Mars® Spec › §02d Jupiter — Discrete Composition › Subject-directed determinations
← Cross-model composition · Section index · Externally-observable properties →
5. Subject-directed determinations
A subject-directed determination is an analysis finding that is order-typed, audit-bound, independently re-derivable, and identifies the responsible order — the specific order of the governing domain model under which the finding was derived. It is the opposite of a black-box score: a score emitted without attribution cannot be independently re-derived or attributed to a cause; a subject-directed determination is explicitly attributed and independently verifiable.
5.1 Structural requirements
A subject-directed determination must carry:
| Field | Content |
|---|---|
| Finding identity | A stable content-addressed identifier for this determination |
| Responsible order | The order identity under which this finding was derived; drawn from the governing domain model’s order decomposition |
| Order-typed evidence | The order-typed IR elements — element identities, order assignments, and analysis-act results — that support this finding |
| Derivation trace | The sequence of analysis acts (battery methods, IR stages, order-typing steps, and traversal witness links where governed traversal was used to surface the order-typed evidence) that produced this finding from the order-typed evidence, expressed as a sub-chain of the §02a §5a.2 witness chain; traversal witness links appear in the sub-chain between IR stage links and analysis-act links where §02a §4b governed traversal was used |
| Source attribution | The governing ground truth artifact(s) — domain model, specification, knowledge base, data source — from which the order-typed evidence was drawn |
| Finding verdict | The typed verdict for this determination: conforming, non-conforming, or conditionally-conforming at the responsible order level |
5.2 The responsible-order identification requirement
Every subject-directed determination must identify exactly one primary responsible order — the order whose shape specification the finding most directly evaluates. Where a finding touches multiple orders, the primary responsible order is the highest-tier order implicated in the finding; secondary orders are listed separately.
A finding that cannot be attributed to any order in the governing domain model is not a subject-directed determination. It is instead a typed gap referral with locus domain model, routing for reinforcement of the model’s order coverage. Issuing a finding without order attribution in place of a typed gap referral is a conformance defect.
5.3 Reflexive artifacts
A reflexive artifact is a subject-directed determination about the analysis system itself — for example, a triage decision about which analysis acts to invoke, a cache-hit determination, or a self-modification assessment. Reflexive artifacts are subject-directed determinations and must satisfy the same structural requirements: order-typed, responsible-order identified, audit-bound, independently re-derivable.
Triage decisions. Where the analysis system decides which battery methods to invoke for a given analysis act, that decision is a reflexive artifact. It must be order-typed (the decision is governed by an order in the domain model, typically an order governing the analysis process itself), carry a responsible-order identification, and be re-derivable from the model and the analysis act’s input.
Cache-hit determinations. Where the system serves a precomputed result from offline precompute (§02a §3), the cache-hit determination is a reflexive artifact. It must record: the cache-key identity, the offline-precompute witness identity, and the equivalence judgment that matched the live input to the cached analysis. The equivalence judgment must be order-typed and re-derivable.
Self-modification order-bounding. Where the analysis system modifies its own formation parameters or traversal rules in response to an analysis result, the modification is a reflexive artifact. A self-modification that would alter the order structure itself — change order identities, order-shape specifications, or the governing model — is an ill-typed modification. An ill-typed self-modification is rejected: it is refused at the modification gate and a typed gap referral is emitted identifying the attempted modification and the order constraint it would violate.
5.4 Relationship to gap analysis and reinforcement
Subject-directed determinations and typed gap referrals are complementary outputs of the same analysis substrate:
- A subject-directed determination is issued when the analysis can produce a grounded finding attributable to a responsible order
- A typed gap referral is issued when it cannot — when the analysis question cannot be grounded in the available order structure, source material, or evidence
A finding that is issued as a subject-directed determination but whose responsible-order attribution does not survive verification is not merely an ill-formed determination — it is a false grounding. An independent inspector who cannot re-derive a subject-directed determination’s finding from its declared order-typed evidence and derivation trace reports a conformance defect at the grounded-verdict level (§02b §2.5 Field 4).
← Cross-model composition · Section index · Externally-observable properties →