Canonization and the Overhang

Kellan Elliott-McCrea responds to David Bessis's "The fall of the theorem economy," drawing parallels between mathematics and software engineering through two borrowed terms.

Canonization (from Alex Kontorovich) is the work of turning a local, one-off formalization into library-quality work: general, reusable, coherent, efficient, and compatible. It's what turns something that works into something the future can build on — the inverse of "vibe coded" throwaway software. Elliott-McCrea frames its absence as "the photo negative of technical debt": features that resist change and drain morale signal missing canonization.

The Overhang (from Bessis) is the unrealized latent value sitting in past creative work, waiting to be unlocked by connecting existing ideas. LLMs, capable of memorizing entire corpora and pattern-matching at scale, are well-suited to searching this space in ways human cognition can't.

The core warning: if the industry rewards only production (theorem-proving, coding) while neglecting canonization and the training of new practitioners, the Overhang will deplete. Review and cleanup earn no status; vibing code into existence does. "Generation easy. Canonization hard." The task is to reimagine what teams reward so practices don't leave people or the Overhang exhausted.