E.17.0:4.7.3 - Keep empirical and formal evaluation local

When empirical interpretation or replay testing is current, identify separately:

• H_dependencyEvaluator : U.System under A.1 as performer;
• exact RA_dependencyEvaluator : DependencyEvaluationWorkAssignment <: U.SystemRoleAssignment under A.2.1, with H_dependencyEvaluator in HolderSystemSlot, declaration-local assigned-kind domain DependencyEvaluatorSystemRoleKindDomain, and DependencyEvaluatorSystemRole as RA’s assigned-kind value admitted by that domain; the value, assignment, holder System, and Work remain distinct, and neither the value nor assignment acts;
• M_dependencyTest : U.Method under A.3.1 and, when needed, D_dependencyTest : U.MethodDescription under A.3.2; D describes M but is neither method, work, RelationSignature, nor OperationAlgebra, and a separate A.6.1 operation declaration is cited only when typed application is current;
• exact W_dependencyTest : U.Work: A.13 first recovers H as the exact actual performer through the already named obtaining RA; A.15.1 independently admits W as enacting M; because this branch expressly represents precise assignment-bound attribution, F.6 separately relates W to that same RA. F.6 identifies neither RA nor H, and a failed F.6 relation would leave W intact while removing only that attribution;
• exact B_dependencyEmpirical, a C.2.1 episteme identifying the model, calibration, assumptions, and interpretation basis; and
• exact result episteme T_dependency = <G_dependencyTestResult,E_dependent,S_test>, whose ClaimGraph designates exact E_base, predicate, method, conditions, basis, and positive or negative result.

Establish actual participation of E_dependent, E_base, each parameter, and B_dependencyEmpirical during W only through the exact relations that define those participation positions or A.6.1 operation-application bindings. A MethodDescription or compatible SlotSpec establishes no participation. Open a local A.15.PROD claim only when the receiving use needs to say W first constituted T or later completed its declared production; inception, completion, episteme identity, and dependency obtaining remain distinct.

When formal interpretation is current, constitute exact formal-evidence episteme E_dependencyProof = <G_dependencyProof,E_dependent,S_proof> and exact B_dependencyFormal identifying the theory, axiom set, proof semantics, and interpretation basis. Its ClaimGraph designates exact E_base, proof obligation, formal method, basis, and result. Preserve entailment, refutation, malformed input, timeout, and checker failure as different outcomes; neither a refutation nor a checker failure fabricates positive r. The proof episteme is distinct from r and its participants.

If reusable target claims are needed, constitute them separately under C.2.1:

• C_dependencyObtains has c_dependencyObtains as its principal claim and concerns the exact endpoint pair and predicate;
• C_dependencyDoesNotObtain carries a distinct negative principal claim and is not a state of the positive episteme; and
• C_dependencyAdmissibleForSelection concerns exact r under the named use frame and remains distinct from both obtaining claims.

Co-representation in one ClaimGraph does not merge these epistemes. T carries its empirical conclusion locally; E_dependencyProof carries its formal conclusion locally. If a target-claim episteme separately represents one conclusion, use C.29 only when representation correspondence matters—never as truth, use, or r. Mint no duplicate evidence-bearing relation and no new A.10 ontology.

Keep these three cases distinct:

1. exact r obtains while support for c_dependencyObtains is unknown; a selecting system may decline reliance without deleting or reidentifying r;
2. a negative empirical or formal result may support C_dependencyDoesNotObtain without presupposing r, fabricating D, or becoming a positive occurrence; and
3. T may support the claim that r obtains without supporting use-specific admissibility; a later decision method may consume empirical and formal result epistemes in separate declared premise slots and produce a separate C.11 result.

For historical reliance on a claim or result, use A.10 to recover its obtaining premise, decision-use, reference-use, or operation-argument relation for the named bounded use. Recover exact Work and its enacted Method only when that dated Work is itself a current claim or the expressly selected empirical evaluation branch above requires them. Retain that branch’s A.13 performer basis, independent A.15.1 Work admission, and F.6 when precise assignment-bound attribution is consumed. Storage, inspection, citation, attachment, production, graph membership, or adjacency alone does not establish that use. Keep empirical and formal algebras distinct; keep provenance and assurance with A.10, G.6, and B.3. Retain a missing-governor blocker instead of inventing a generic evidence, use, or acceptance relation.