ARLE: A Preregistered Instrument for Boundary‑Condition Load Escalation and Asymmetric Repair Load

This manuscript introduces ARLE, a preregistered, audit-first measurement instrument for estimating the probability of boundary-linked coherence loss events in interpersonal or symbolic interactions while minimizing normative drift and misuse. ARLE outputs model-based descriptive inferences—P(BCLE) and, when identifiable, P(ARLE)—with uncertainty intervals and explicit non-identifiability states. It separates inference from policy via an Instrument Output Contract (IOC): action-tier guidance is limited to self-exposure management and is permitted only when preregistered integrity and evidence gates pass.

Two gating mechanisms make the instrument reviewer-resistant and conservative by design: (1) a Null-World Gate (NWG) requiring calibration and incremental value versus stress-only baselines, plus a within-session boundary permutation test, and (2) a Tamper/Adversary Flag Index (TAFI) that suppresses interpretation under consent, integrity, or manipulation concerns. The paper specifies machine-checkable state tokens (INSUFFICIENT_DATA, NOT_IDENTIFIABLE, NO_EVIDENCE_BEYOND_NULL, DATA_COMPROMISED), a standardized ARLE Run Report, and compliance checklists to ensure reproducible reporting and limit weaponization. The intent is an instrument-grade methods contribution: falsifiable, conservative under uncertainty, and export-restricted by default (aggregated/de-identified only when permitted).