FlagThis

Abstract

Every authoritative positioning, navigation, and timing (PNT) resilience framework, from the DHS Resilient PNT Conformance Framework (RPCF) to the Resist-Detect-RespondRecover model and Yangs resilient-PNT criteria, defines what resilience means but supplies only self-attestation: a checklist or a maturity Level, with no engine, no measurement, and no evidence. We build the missing measurement layer as an open, deterministic scoring engine over a PNT simulator, emitting perdimension sub-scores each traceable to a scenario and an oracle and each tagged with its honest validation status. We then ask whether a single composite score, or a single maturity Level, is a stable basis for a decision. Across a reference panel of seven architectures built to span genuine cross-dimension tradeoffs, a Dirichlet weighting simplex over the seven RPCF categories, and a five-threat ensemble, the answer separates into two regimes. The composite winner is stable under active denial and under near-equal weightings, flipping in about 1 percent of draws, so a single number is safe precisely where one design dominates; but re-weighting alone flips the winner in up to 22 percent of draws under nominal conditions, where designs genuinely contend, and that instability is a domain application of known compositeindicator sensitivity. The sharper and weighting-invariant failure is categorical: a weakest-link maturity Level (our minimumover-categories operationalization of the RPCF ladder, not the frameworks own rule) is a function of the threat assumed rather than of the architecture, changing for one architecture in seven across the ensemble. A constructed example shows that, because the composite rewards declared techniques, a singleband receiver declaring all seven techniques can outscore a genuinely more resilient system: self-attestation can be gamed by declaration. And apparent fourfold GNSS redundancy reduces, by the definition of a shared common-mode failure domain, to an effective diversity of one. The conclusions hold under a plusor-minus 20 percent perturbation of every driver within the modelled reduction. We argue for reporting per-dimension subscores with provenance and a rank range, not a phantom single number. This is a simulation-derived self-assessment aligned to RPCF v2.0, not a certification.