Reproduce “Assessing Game Balance with AlphaZero: Exploring Alternative Rule Sets in Chess” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Nenad Tomašev, Ulrich Paquet, Demis Hassabis, Vladimir Kramnik
The authors train separate AlphaZero agents under nine altered chess rule sets and use self-play to examine how rules affect outcomes and opening preferences. In 10,000 one-second-per-move games per variant, No-castling chess draws 84.41% of games, versus 88.20% in Classical chess. This model-based comparison suggests rule changes can alter play, but does not establish human outcomes or isolate causes.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.