kybernesis: A Revisable Meta-Foundation for Evolutionary Evidence-Based Policy Making
Evidence-based policy making composes observation, interpretation, estimation, evaluation and authorised action. The composition requires explicit interfaces through which evidence, methods, evaluative standards and authority can be examined and revised. This technical report develops kybernesis, a software meta-foundation for steering systems known through evidence. An architecture declaration specifies typed blocks, ports, laws, boundaries and delayed connections. Structural constraints and categorical checks govern admission of architectures, while finite relational models support admission of methods to their blocks. Evaluation operators act on estimated states under declared readings and standards. Operational learning, collective choice and constitutional revision organise the evolution of policies and their evaluative terms. The implementation provides execution, recorded human interventions, branching, replay, ontology migration and conditional transfer between relational systems. Formal arguments and executable examples establish properties of the finite representations used for admission and recording. A riverside example develops Bayesian updating and sensitivity to semantic readings; an urban engagement illustrates composition of geographic evidence, standards and transfer assumptions. General policy evolution and constitutional governance define extensions of the implemented foundation. The resulting framework supports an inspectable account of knowledge production and policy practice, with the assumptions, evidential standing and authority of its constituent acts recorded explicitly.
Authors
- Wanhong HUANG
Publication Details
- Journal
- Knowledge Commons Works (Michigan State University)
- Published
- 2026-11-19
- DOI
- https://doi.org/10.17613/5p429-nfg71
- Primary Topic
- Evaluation and Performance Assessment
- Type
- preprint