By the time this evaluation ran, the artefact had a large body of tests behind it. It would be easy, and wrong, to present that as evidence that the discipline discriminates.
Those tests shaped the model. Thirty-seven of the forty-three specification cases exist because the model got something wrong and was corrected, and a case written to record a correction cannot then serve as an independent test of it. The evaluation therefore separates three classes and does not let one stand for another.
Conformance evidence is the executable specification, its regressions, the property tests over generated input, and the mutation operators. It answers whether the implementation preserves the discipline the design chapter fixed. It claims nothing about discrimination.
Operational demonstration is the worked integration: a plant historian feeding a model output, adjudicated for two different acts, then re-established by an independent measurement and reevaluated. It was developed alongside the artefact and is therefore exploratory. It evidences that the pieces connect and that the discipline is mechanisable end to end, and it is labelled a demonstration wherever it appears.
Discrimination evidence is what the confirmatory corpus was intended to supply: thirty-nine scenarios authored after the artefact was frozen, from situations external to the design. It is the only class that could speak to whether the discipline tells warranted practice from unwarranted, and it is the class the rest of this chapter concerns.
Its realised role turned out to be different. Only four translations survived exactness review, so the corpus principally became an evaluation of what the artefact can represent, and supplied discrimination evidence over a much smaller set than intended. That is reported as the outcome rather than as the design.
The separation has a cost worth stating. The largest and most rigorous body of testing in this work contributes nothing to the central research question, which is answered instead by a much smaller body of evidence that was harder to produce. That is the correct trade, and it is why the results below rest on single figures rather than on hundreds of passing tests.
What the conformance evidence establishes
The implementation carries 171 tests, 43 specification cases of which 37 are regressions, and 43 mutation operators, all of which the test suite kills. The regressions are the more interesting number: each records a correction the model made during design, so the suite protects against the artefact quietly rebuilding an earlier and worse version of itself.
Mutation testing was used because a passing suite proves only that the tests ran. Reversing a correction and observing the suite fail proves the tests would have noticed. Three mutants survived their first run, and in each case the implementation was correct while the corpus was thin, which is the failure that exercise exists to expose.
None of this bears on discrimination, and the boundary is worth stating flatly. Conformance establishes that the artefact preserves the discipline as specified and resists the faults tested. It does not establish field usability, integration fitness across vendor systems, correctness of supplied provenance, completeness of construction-boundary capture, or operational safety effectiveness.