top of page

                                                     A Science of Chess ?










Chess is more than a game—it’s a living laboratory for studying strategy, complexity, and intelligence. Today’s engines play far beyond human ability, revealing patterns and ideas once hidden from view. By learning from them, we can search for the game’s universal features—those that define its very structure—and others that illuminate its inner workings as a complex system. This quest continues a vision once expressed by Richard Réti: the dream of a true science of chess.

This page is under construction, I am happy to hear your comments and suggestions
 

combined_chess_graph_same_size.png

News & In the Press

New paper about how much engine evaluation is useful for human brains

Blog article on chessable.com written by Karel van Delft and Dries Wedda 

Blog article about the fragility of positions on chess.com

Article "How a new science of Chess is emerging" Discover Magazine (28/10/24)

 

Article "Complexity physics finds crucial tipping points in chess games" in Ars Technica

Papers

 

Chess variation entropy and engine relevance for humans

      M Barthelemy

      arXiv:2505.03251

Fragility of chess positions: Measure, universality, and tipping points

      M Barthelemy

      Phys. Rev. E 111, 014314, 2025

      arXiv version​

 

Statistical analysis of chess games: space control and tipping

      M Barthelemy

       ArXiv.2304.11425

Blogs and websites
 
On chessable many resources about scientific approaches to chess are available:
- https://www.chessable.com/science/
- chessable blog: https://www.chessable.com/blog/
 
Other few blogs exist:
- Interesting references can be found on the site https://www.chess-science.com/en/chess-and-science/
- A blog on scientific approaches to chess: https://lichess.org/@/NDpatzer/blog
Interesting podcast on France-Inter (in French) about Chess and AI -Interview of the GM Laurent Fressinet
Data

The standard format for games is the Portable Game Notation (PGN) format, which includes metadata (e.g., date, location, opponent) and the moves in algebraic notation (e.g., a2a4, b7b8). Useful sources are:
 

Codes
 
bottom of page