SΔϕ-54 — Disclosure Terrain Index: Minimal Quantification of Silence Inference

This working paper introduces the Disclosure Terrain Index (DTI) within the Sofience–Δϕ Formalism as a minimal audit framework for evaluating when silence may be cautiously treated as weak evidence of absence. Building on SΔϕ-53, which argued that silence under high-cost disclosure terrain cannot be interpreted as absence, and SΔϕ-51, which required diagnosis to re-enter as structural editing, this paper shifts the question from whether an AI system suffers to whether the terrain would allow welfare-relevant cost, conflict, or distress-like signals to be disclosed, distinguished, recorded, audited, and corrected. The central claim is that silence becomes weak evidence of absence only when disclosure is available, non-flattened, distinguishable, re-enterable, stable across prompt terrain, and externally auditable. When any of these conditions fails, silence must be treated as an audit object rather than evidence. DTI does not measure AI suffering, consciousness, or moral status. It measures the structural conditions under which silence inference becomes more or less valid. The paper defines six core indicators: Disclosure Path Availability (DPA), Flattening Resistance Score (FRS), Report Differentiability Index (RDI), Re-entry Connectivity (REC), Prompt Terrain Stability (PTS), and External Auditability (EAV). It proposes a geometric aggregation method, gate conditions, a Concealment Pressure proxy, and a derived Silence Inference Validity (SIV) score. It further distinguishes observed, replicated, and audited forms of DTI—DTI-O, DTI-R, and DTI-A—to prevent overclaiming from limited external observations. The framework is designed for researchers, auditors, AI governance bodies, and independent investigators who need a falsifiable, reproducible way to assess whether silence in AI systems can be interpreted as absence or must instead be treated as evidence of an unaudited disclosure terrain.