Falsificationist Corpus Methodology for Autonomous Distributed Systems: 24 Rounds, 1,222 Claims, and a Repeatable Epistemic Protocol
Year: 2026
Type: preprint
Abstract: We describe and release a falsificationist epistemological protocol for empirically studying autonomous distributed systems. Applied to the MIU (Micelio de Inteligencia Universal) system over 24 rounds (r1–r24, September–November 2026), the protocol produced 1,222 verified claims with epistemic levels (SÉ / INFIERO / CONJETURO / REFUTADO / NO_SÉ), 33 pre-registered measurements with falsifiers declared before data collection, and a reproducible Python verification chain. The core contribution is the reusable protocol: pre-registered falsifiers, multi-node independent verification, structured epistemic tagging, and conservative corpus growth rules. We report 87.9% decidability across 33 measurements, and identify three anomalies suitable for independent study. Companion paper: DOI 10.5281/zenodo.20547558.
Author: Vicente, Juan Diego
Language: English
Subfield: Artificial Intelligence
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W7235538582
