TMS-based architecture models how claims retract as evidence shifts
letandrewcook · x · 2026-11-06
The author proposes an architecture for handling derived propositions when their evidentiary basis changes:
- A Claim c is modeled with a set J(c) of justifications under OR-of-AND support semantics, with an acyclic dependency graph
- Doyle's Truth Maintenance System propagates support loss when sources mutate or drop out
- Symbolic dependencies are separated from geometric position, since dense embeddings encode proximity but not opposition along a specific quality; each dimension defines an ordered value space with projection, comparison, normalizer, prototypes, provenance, and version
- An accompanying figure shows the Claim state machine: proposed/accepted/contested are retrievable; retracted claims are kept for audit; superseded ones stay reachable from successors
A symbolic-geometric hybrid design for revocable, auditable reasoning.
More from Research
- Google DeepMind shows AI agents accelerating cryptanalysis of encryption schemes — sebkrier · 2026-11-07
- Frank Noe shares DeepQMC results for rubredoxin Fe(SH)4 cluster spin-crossing surfaces — FrankNoeBerlin · 2026-11-07
- DeepQMC benchmarks excited states of C2 and methylenimmonium against high-level references — FrankNoeBerlin · 2026-11-07
- Noé group releases transferable DeepQMC for excited-state potential energy surfaces in one run — FrankNoeBerlin · 2026-11-07
- Luger's New Springer Chapter Traces AI's Pre-History from Myth to Dartmouth — ArtificialOther · 2026-11-07
- Neel Nanda, after supervising 20+ papers, publishes a guide to writing great ML papers — CSProfKGD · 2026-11-07