Abstract This paper codifies a recurrent heuristic of theory repair: when a theoretical posit remains explanatorily indispensable yet persistently fails to stabilise as a manipulable, localisable entity, it is sometimes methodologically productive to retype it from "bearer" to constraint-role. The aim is practice-facing and methodological rather than foundational. The paper does not settle ontology, defend any particular metaphysics, or claim that constraints exhaust reality. Instead, it offers explicit, audit-oriented criteria (together with scope limits and failure conditions) for when role-replacement is a legitimate move rather than an ad hoc redescription. A key addition is a prospective discriminator: many eventually-confirmed entities temporarily resemble "non-things"; the heuristic therefore distinguishes candidates for constraint-typing from candidates for eventual entity-confirmation by tracking experimental accessibility (manipulability, isolation, controlled intervention) as instrumentation matures. Historical episodes are presented as rational reconstructions for methodological clarity, not as exhaustive historiography. These cases illustrate the framework's categories but do not exemplify deliberate methodological choice; the framework is best understood as prospective discipline for evaluating future retyping proposals, not retrospective rationalisation of historical episodes. The framework operates as a retrospective rationality criterion with prospective disciplinary pressure: it clarifies why certain historical retypings succeeded while constraining what counts as legitimate justification for future proposals. Borderline cases will be contested; the rule-set structures disagreement rather than eliminating it.
Jaimes Chao (2026) studied this question.