Generative AI can change local work capacity faster than organizations update the systems that govern work. Drafting, coding, summarizing, replying, searching, and analysis may become faster, while review, validation, integration, approval, budget ownership, or decision authority becomes the new limiting step.
This working paper introduces normative governance as a design framework for AI-mediated work under moving constraints. It builds on the preceding Operator Judgment papers, which developed an explanatory account of task-specific expertise, conversion costs, candidate work, local metrics, constraint movement, and governance lag. The central claim is that AI governance becomes defensible when it tracks accepted work, conversion capacity, active constraints, denominator validity, decision rights, budget ownership, validation capacity, and update timing.
The paper develops the Moving-Constraint Governance Cycle, a governed adaptation procedure linking capacity-change detection, candidate-work review, denominator review, active-constraint diagnosis, authority/budget/validation alignment, and update records with monitoring triggers. The framework avoids both unrestricted automation and excessive procedural drag by treating governance as adaptive alignment around the work that now governs value.
The deposition package includes the final PDF, DOCX, Markdown source, citation metadata, package manifest, checksums, source note, and transcript supplement.
Publication Date: 2026-06-20