Tag
The author reflects on the paper 'Self-Revising Discovery Systems for Science' which proposes a new agentic architecture using strongly-typed DAGs, schema migrations via Kan extensions, and an MDL gate to distinguish genuine discovery from simple retrieval or search.