Genealogical Trajectory Logic: Counterexample-Guided Theory Reconstruction with Selective Resolution and Deliberate Proxy Use

A proxy can be coarse because consequential differences remain unknown, or because warranted distinctions are deliberately inactive for a bounded task. Genealogical Trajectory Logic (GTL) connects this difference to counterexample-guided reconstruction of theory families. Prospective probes target consequential conflations; Refine links their adjudication to selective inheritance, application conditions, and successor-generation rules. A memory-bounded account proposes a route from costly comparison to cheaper reuse through organization, compilation, and selective retention. Short-, medium-, and long-term memory denote functional roles, not mandatory biological stages. Reduced active load, shorter encoding, and lost information remain distinct. Consolidation must preserve declared answers, observable reopening cues, and responses to registered support changes. Preserving an answer alone does not preserve the ability to withdraw it when its warrant expires. A constructed recovery family demonstrates probe selection, genuine refutation, binary recompression, selective reactivation, and generator repair; negative controls expose silent drift, lost support, and accumulated error. A prospective register separates seven testable working predictions and one recursive extension from exact retention conditions and conformance checks. Each prediction specifies a competing explanation, measurement conditions, and outcomes that would count against the named claim. Matched resources, independent test families, coverage, and full lifecycle costs prevent success from being inferred from short answers, warm caches, or universal abstention. These predictions are not reported findings. GTL offers a methodological specification for retaining the significance and revisability of theory families under bounded resources. Evidence status: no empirical GTL evaluation is reported. The manuscript carries 25 analytic reference cases, and the supplementary archive records 30 independent analytic checks, all passed, together with review notes, edit and revision histories, and SHA-256 checksums. Predictions H1-H7 and the recursive extension X1 remain prospective. Files: the finalized manuscript (v1.0) and a supplementary archive containing the review note, verification scripts, audit results and checksums. GTL is part of the author's Deficit-Fractal Governance research programme.