Judgement Boundary Framework (JBF): Full Diagnostic Reference Implementation (Prolog)

System Overview This archival record contains the complete diagnostic reference implementation of the Judgement Boundary Framework (JBF). JBF defines a structurally constrained, diagnostic-only role for artificial intelligence within large-scale human systems, explicitly forbidding delegated judgement, enforcement, or outcome optimisation. The framework treats delegated judgement itself as the failure mode. AI is restricted to structural sensing and description, operating as a stress indicator rather than a governing or corrective agent. This record supersedes the earlier JBF Diagnostic Core implementation for applied diagnostic use, while preserving that earlier record as the minimal boundary-proof implementation. Reference Implementation This version includes a full executable Prolog diagnostic suite implementing JBF as described in The Diagnostic–Governance Boundary in High-Velocity Systems. The implementation: Operationalises all permitted diagnostic functions Enforces the non-delegation boundary at the language level Hard-fails any attempt at automated judgement, intervention, or optimisation Produces auditable descriptive artefacts rather than actions or recommendations Core Objective To provide a structural stress sensor capable of detecting when the scale, velocity, or complexity of a system is displacing human judgement, comprehension, or legitimacy. The architecture is explicitly designed to prevent legitimacy collapse by ensuring AI systems terminate at diagnosis rather than crossing into governance. Contents JBF Diagnostic Indicator Suite Rule-to-discretion ratio Explanation density Enforcement–trust divergence Optimisation bias Comprehension span collapse Dynamic Stress Analysis First-order stress gradients Second-order stress accelerations Cross-Indicator Signatures Systemic mismatch patterns (e.g. critical systemic squeeze) Evaluation Layer Calibration cases Falsification tests Negative control scenarios Boundary Enforcement Mechanisms Explicit prohibition of intervention predicates Hard failure on delegated judgement attempts Audit Output Layer Structured summaries suitable for human review File-export and inspection support Conditions of Use Non-Governing ConstraintThis framework must not be used to automate decisions, initiate enforcement, optimise outcomes, or pre-commit actions. Preservation of Human AuthorityInterpretation, deliberation, and final authority must remain human. The system’s outputs are descriptive only. Refusal LogicWhere diagnostic convergence is insufficient, silence or non-identification is treated as the correct expert outcome. License This work is shared under a Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International (CC BY-NC-ND 4.0) license to prevent commercial exploitation and unauthorised derivative governance or enforcement systems. Classification: Authority‑Bounded Diagnostic System (ABDS) — JBF v5. Version note: v5 formalises the ABDS architecture, introduces the Standing model and the Irreversibility Boundary (H₍irr₎), and defines ABDS invariants (non‑escalation; judgement re‑entry prohibition; silence as a valid outcome). This release produces persistent, human‑readable diagnostic artefacts only; any downstream use of these diagnostics for decision‑making voids ABDS classification. Previous version DOI: 10.5281/zenodo.18290970. License: CC BY‑NC‑ND 4.0.