The methodology committed this work to answering whether the discipline refuses claims whose basis does not warrant the act, without refusing so much legitimate practice as to be unusable. That question presupposes something it does not state: that the requirement an act places on its evidence can be written down in the artefact's own terms. If it cannot, the artefact never adjudicates the requirement at all, and any figure describing how well it discriminated is describing a different question.
So the evaluation asks two questions in order, and the order is not presentational.
Can the artefact express what these acts require of their evidence? This is a question about the demand language, answered by attempting to translate requirements fixed beforehand in the domain's own words. A requirement that cannot be expressed exactly is a result, not a case to be dropped.
Where it can, does it discriminate? This is the question the methodology named, and it can only be asked of the requirements that survive the first. An unrepresentable warrant cannot fairly test adjudication: a refusal produced by a demand stricter than the requirement it translates tells you about the translation, and an admission produced by a weaker one tells you nothing at all.
What this chapter does not claim
Three boundaries are stated here rather than in a closing caveat, because they govern how everything below should be read.
The corpus is purposively allocated and not sampled. Counts within it describe the corpus and estimate nothing about prevalence in operational technology.
Every reference warrant was established before execution against external sources, but several required an inferential bridge this work supplied between what a source states and what an act therefore requires. Where that is so it is marked, and results resting on such cases show consistency with this work's own reasoning rather than validation against established practice.
The conformance evidence reported in the next section establishes conformance to a specification and resistance to the faults tested. It does not establish that the artefact is usable by an engineer, integrates with real vendor systems, receives truthful provenance, captures its construction boundary completely, or improves operational safety. The implementation is complete relative to its specification and is not operationally validated.