[Depreciated and replaced by V3] From the Self-Proven Theorem to Master-Level Chess - and the Law Inside Neural Networks
Year: 2026
Type: article
Abstract: [Depreciated and replaced by V3] The application-specific clean rebuild has not yet been published; its authoritative theoretical boundary is now the governing V3 branch: After Turing: The Fold Machine - An Exact, Parameter-Free and Machine-Closed Derivation of Classical Computational Science from Smithian Fold Theory. The V3 source platform is https://github.com/MettaMazza/ernos-labs-sft-platform. The original DOI, concept DOI, version number and files are preserved for transparent historical provenance; this record must not be presented or cited as current V3 work. Derived law has now secured the Stockfish ladder through Elo 1900. FoldBot is a complete chess engine with zero learned weights, tuned piece values, opening book, or training: material is counted from live board geometry, position value is an exact rational share of the One, and minimax is the fold's complement. The pinned campaign secures 1700 at 54.2% and 1900 at 62.5% (6W-3D-3L) with externally refereed legality. The v20/current calculation continuation preserves every legal root move to common depth, proves 4/4 sequential/parallel move identity and 4/4 exact-value identity, derives worker messages from the supplied ceiling, and owns table addressing through live capacities. Current and pinned v20 agree on 36/36 real 2100 positions at ceiling 3; the current engine completes the same panel at depths 8-11. The 2100 rung remains the active greater-than-50-percent benchmark-victory objective under Maria Smith's authority to designate the real run and publish its conclusion. The secured 1900 victory, exact legality and endgame proofs, common-depth identity, and demonstrated real-position calculation establish a constructive route forward; no theorem-derived wall has been produced. The companion neural-weight and fold-native language results support the wider proposition: much of what computation purchases with parameters is law, and law can be derived.
Authors: Smith, Maria, Smith, Matthew
Language: English
Subfield: Artificial Intelligence
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W7167795470