Assessing Game Balance with AlphaZero: Exploring Alternative Rule Sets in Chess

  • 2020-09-09 15:49:14
  • Nenad TomaĊĦev, Ulrich Paquet, Demis Hassabis, Vladimir Kramnik
  • 61

Abstract

It is non-trivial to design engaging and balanced sets of game rules. Modernchess has evolved over centuries, but without a similar recourse to history,the consequences of rule changes to game dynamics are difficult to predict.AlphaZero provides an alternative in silico means of game balance assessment.It is a system that can learn near-optimal strategies for any rule set fromscratch, without any human supervision, by continually learning from its ownexperience. In this study we use AlphaZero to creatively explore and design newchess variants. There is growing interest in chess variants like Fischer RandomChess, because of classical chess's voluminous opening theory, the highpercentage of draws in professional play, and the non-negligible number ofgames that end while both players are still in their home preparation. Wecompare nine other variants that involve atomic changes to the rules of chess.The changes allow for novel strategic and tactical patterns to emerge, whilekeeping the games close to the original. By learning near-optimal strategiesfor each variant with AlphaZero, we determine what games between strong humanplayers might look like if these variants were adopted. Qualitatively, severalvariants are very dynamic. An analytic comparison show that pieces are valueddifferently between variants, and that some variants are more decisive thanclassical chess. Our findings demonstrate the rich possibilities that liebeyond the rules of modern chess.

 

Quick Read (beta)

loading the full paper ...