§02f Jupiter — Post-Hoc Forensics — Post-hoc artifact attribution
Mars® Spec › §02f Jupiter — Post-Hoc Forensics › Post-hoc artifact attribution
← Reinforcement-action layer · Section index · Composition of the post-hoc forensic layer →
8. Post-hoc artifact attribution
The post-hoc forensic layer may attribute a forensic finding across the relation between one or more knowledge bases and data sources and one or more domain models and specifications. This attribution operates post-hoc over the evidence trail; pre-execution artifact use is governed by §03.
Asset relation. The relation is grounded over assets — a domain model, a specification, a data source, a knowledge base; zero or more of each, at least one overall, combinable without restriction across orders and higher-order orders.
Governing-bound fallback. The governing bound is supplied by a fallback hierarchy: a specification if present; else a domain model acting as the specification; else a domain model reverse-synthesized from at least one data source and/or knowledge base to serve as the specification. The fallback hierarchy is declared in the forensic record; the governing bound used is identified by its position in the hierarchy.
Two-layer grounding join. Where both a knowledge base and a data source are present, the relation is resolved at the order-structured join of knowledge-base governance (which rule or obligation governs) with data-source instance-facts (the case facts). Neither layer alone is sufficient; the join of the two layers is the relation. A knowledge-base-to-data-source divergence on the finding’s referent is a signal contributing to the attribution.
Asset-relative attribution. The forensic finding is attributed across the asset relation — e.g., to a stale or wrong knowledge base or data source relative to the governing bound — as an order-typed, recomputable claim re-derivable by an independent inspector against retained witnesses satisfying the §02a §5a.2 multi-stage witness chain structural requirements. The asset-relative attribution maps to the data causal dimension of the §4 causal attribution. An asset-relative attribution that is not recomputable against witnesses satisfying the §02a §5a.2 requirements is structurally incomplete.
Typed-gap reinforcement. Where the asset relation reveals a deficiency — a knowledge base or data source absent, stale, or in conflict with the governing bound — the forensic layer emits a typed gap {KB / data source} and routes it for reinforcement under §7. The typed gap is a first-class forensic artifact, not an informal observation.
Reverse-synthesized governing bound is not a distinct architecture. Where no specification and no domain model are provided, a governing bound reverse-synthesized from at least one data source and/or knowledge base steps in as the specification for the purpose of this section. The reverse-synthesized bound occupies the third position in the fallback hierarchy; it is not an alternative governing architecture and it does not relax any structural requirement. The asset-relative attribution and typed-gap reinforcement produced under a reverse-synthesized governing bound are subject to the same order-typed, recomputable, witness-bound requirements as those produced under a specification or a domain-model-as-specification. An implementation that treats the reverse-synthesized governing bound as a distinct governing mode with relaxed requirements is not conforming to this section.
Distinction from pre-execution artifact traversal. Post-hoc artifact attribution under this section and pre-execution artifact traversal (§03) apply the same asset abstraction — the same fallback hierarchy and the same two-layer KB × data-source join — to different acts: post-hoc artifact attribution attributes a forensic finding across the asset relation; pre-execution traversal acquires context for a prompt before model execution. The two are not interchangeable and the shared asset machinery is not independently claimable as a general-purpose mechanism separable from either act. An implementation of the two-layer join and fallback hierarchy that is not grounded in one of these two acts — post-hoc forensic attribution or pre-execution context acquisition — is not a conforming use of the asset abstraction.
← Reinforcement-action layer · Section index · Composition of the post-hoc forensic layer →