Corrigible Goodness Under Constraint: Authoritative Renderings, Misfit, Answerability, and Revision in Good-Oriented Systems

David T. Swanson · 2026

Abstract

This paper argues that under finite, mediated, and fallible conditions, no system can responsibly pursue the good through fixed targets alone. Real systems do not act on the good in unconstrained contact with reality. They act through authoritative renderings: operative targets, proxies, representations, procedures, thresholds, files, scores, and other structures that acquire standing over cases. Because these renderings are selective and structurally vulnerable to misfit, any system that claims to pursue the good must keep them corrigible, answerable, and revisable relative to reality and the subject-processes they govern. The constitutive danger of finite good-oriented systems is therefore not target use as such, but closure in authoritative renderings. Target closure is one especially important form of that broader danger: once the renderings through which systems actually govern harden into self-authorizing practical authorities, systems can optimize misfit rather than the good. Correction is therefore not an optional moral virtue but a structural necessity of good-oriented systems. The paper sharpens this claim by distinguishing declared good, authoritative rendering, operative target, and real good; by showing why governance over subject-processes intensifies vulnerability to misfit; and by arguing that corrigibility is not real unless misfit can become detectable, travel through a live challenge path and usable correction channel, and reach a locus with authority to revise what is actually governing the case within a meaningful burden and time profile. The framework does not provide a final metaphysics of goodness or a complete ranking of goods. Its narrower aim is to identify what must be true of systems that claim to pursue the good under finite conditions.

Citation

@misc{Swanson2026CorrigibleGoodness,
  author       = {Swanson, David T.},
  title        = {Corrigible Goodness Under Constraint: Authoritative Renderings, Misfit, Answerability, and Revision in Good-Oriented Systems},
  year         = {2026},
  publisher    = {Zenodo},
  doi          = {10.5281/zenodo.19105820},
  url          = {https://doi.org/10.5281/zenodo.19105820}
}