A New Self-Play Experiment in Computer Chess
Author(s)
Heinz, Ernst
DownloadMIT-LCS-TM-608.pdf (1.138Mb)
Metadata
Show full item recordAbstract
This paper presents the results of a new self-play experiment in computer chess. It is the _x000C_rst such experiment ever to feature search depths beyond 9 plies and thousands of games for every single match. Overall, we executed 17,150 self-play games (1,050{3,000 per match) in one \\calibration" match and seven \\depth X+1 , X" handicap matches at _x000C_xed iteration depths ranging from 5{12 plies. For the experiment to be realistic and independently repeatable, we relied on a state-of-the-art commercial contestant: Fritz 6 , one of the strongest modern chess pro- grams available. The main result of our new experimentis thatit shows the existence of diminishing returns for additional search in computer chess self-play by Fritz 6 with 95% statistical con_x000C_dence. The dimin- ishing returns manifest themselves by declining rates of won games and reversely increasing rates of drawn games for the deeper searching pro- gram versions. The rate of lost games, however, remains quite steady for the whole depth range of 5{12 plies.
Date issued
2000-05Series/Report no.
MIT-LCS-TM-608