We present CANONIC: a governed intelligence system that compiles digital artifacts into a scalable evidence ledger. Large language models generate prose faster than humans can verify, a phenomenon termed 'slop' by Oxford Languages, naming it the Word of the Year for 2025. CANONIC governs the admission of content akin to how a compiler determines program well-formedness: mechanically, through grammar. The governance process reduces to three axioms (Triad, Inheritance, Introspection), which map onto compiler theory's syntax, scope-resolution, and type-system layers, with admission being a decidable, linear-time check. We then conduct a pre-registered cross-provider benchmark to assess whether structural admission can effectively prevent slop. The results indicate that no prose-reading gate reliably distinguishes reliable from unreliable content. Slop is not a property that an algorithm computes; it is a verdict of domain expertise. Thus, the governance layer does not determine slop; it ensures the record is auditable—every claim anchored to a definition, a commit, and an evidence window, making the process reproducible and checkable end to end.
Blogger's Review: CANONIC's governance mechanism provides a novel approach to digital content review by leveraging compilation principles. However, the findings show that algorithms alone cannot effectively assess content quality, highlighting the importance of domain expertise in governance, thus warranting further exploration of hybrid governance models.