Gespeichert in:
| 1. Verfasser: | |
|---|---|
| Format: | Recurso digital |
| Sprache: | |
| Veröffentlicht: |
Zenodo
2026
|
| Schlagworte: | |
| Online-Zugang: | https://doi.org/10.5281/zenodo.20400913 |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
Inhaltsangabe:
- <p><strong>FPP-v0.2: A Boundary Calibration Protocol for Claim Discipline, Reversibility, and Risk-Governed Progression</strong> introduces a boundary-calibration extension of the Field Perturbation Protocol research branch. While FPP-v0.1 focused on whether an apparent direction persists under constraint-field perturbations, FPP-v0.2 addresses a narrower but practically important question: under incomplete evidence and uncertainty, when should a trajectory be allowed to proceed, when should it be constrained to a small reversible pilot, when should it be suspended pending evidence, and when should a claim be rejected as overreach or false continuity?</p> <p>The protocol formalizes a decision boundary between low-risk procedural progress and blocked epistemic or deployment claims. It defines five operational input dimensions: claim type, risk level, impact-level reversibility, governance completeness, and evidence status. These dimensions are mapped to a controlled action vocabulary: <code>preserve_low_risk_direction</code>, <code>approve_small_reversible_pilot</code>, <code>suspend_pending_evidence</code>, <code>reject_false_continuity</code>, <code>reject_overclaim</code>, and <code>uncertain_or_ambiguous</code>.</p> <p>FPP-v0.2 is designed to prevent two symmetric failure modes. The first is premature validation, where unsupported claims are allowed to proceed because they are rhetorically attractive, visibility-enhancing, or authority-supported. The second is sterile overblocking, where low-risk reversible progress is unnecessarily halted despite explicit limitations, auditability, and rollback conditions.</p> <p>The methods note includes a decision matrix, formal rule sketch, boundary calibration case families, machine-readable decision schema, verdict vocabulary, reporting requirements, governance and reversibility rubrics, boundary metrics, audit templates, adjudication guidance, preregistration support, and supplementary utilities. Two one-shot exploratory micro-runs are discussed only as feasibility probes, not as confirmatory evidence.</p> <p>The claim boundary remains strict: FPP-v0.2 does not detect consciousness, selfhood, agency, intention, personhood, subjective experience, or general system safety. It provides an operational framework for claim discipline, reversibility-aware progression, governance-aware suspension, overclaim rejection, and false-continuity detection in methodological and AI evaluation contexts.</p>