Devin Bostick is an independent researcher developing the Identity-Persistence Program, a formal research program on the structural conditions under which bounded observers can make coherent, reproducible judgments about identity, persistence, admissibility, declaration, and verification under declared regimes.
The program begins from a simple question: what must already be fixed before a claim of sameness through change can be meaningfully evaluated? Its foundational results derive structural floors for identity persistence, admissibility under transformation, and sufficient regime specification, together with limit results characterizing what bounded evidence can and cannot identify. Within declared finite and compact-metric classes, the work develops corresponding capacity, coding, enforcement, and verification results.
A second layer studies the internal mathematics of declared regimes: demand-relative normal forms, regime equivalence and refinement, recurrence classification, symmetry-compatible quotienting, nested regime composition, reconstructibility and mechanization theory, finite condensation dynamics, and lawful retirement and erasure. These results characterize what may be safely forgotten, what must persist, how downstream demands constrain upstream representations, and which structural features are canonical only relative to a declared demand.
The program’s central outcome is a sharper boundary on bounded judgment. Licensed distinctions determine an information quotient, and derivable consequences are exactly those that respect the distinctions that quotient preserves. For a desired consequence that cannot yet be derived, the theory can identify the exact informational obstruction and the canonical minimal refinement needed to remove it; in finite settings, existing repair theory also determines the minimum discriminating width any sufficient informational realization must carry. What the mathematics does not thereby supply is the realization itself, the institutional authority entitled to introduce it, or the authority to act on the resulting consequence. In this sense, declaration is not treated as a failure of reason but as the authorized introduction of distinctions that the current inferential regime cannot supply for itself.
The program is deliberately bounded. Regime-relative verification is not promoted into final ontology; evidence accumulation is distinguished from unique identification; operational closure is distinguished from self-certifying finality; and formal mechanization theory is kept separate from independent machine verification of the corpus itself. Claims are graded in place as proven, conditional, derived, applied, refuted, open, or outside scope rather than silently promoted.
The research extends into engineering through deterministic decision verification: systems in which consequential verdicts are computed under declared regimes, recorded as independently replayable artifacts, and checked by bounded evaluators. An open-source reference runtime realizes a bounded portion of this architecture; it is an engineering instantiation rather than evidence for the correctness of the broader mathematical corpus. The corpus is DOI-anchored and accompanied by an Orientation that states its dependency structure, reading paths, claim statuses, and explicit boundaries.
Source code: https://github.com/verifyrun/verify-run
.jpeg)
