One Jump Ahead pp 334-363 | Cite as

Dissension Among the Ranks

  • Jonathan Schaeffer


The match was over, and now it was time to get on with the rest of my life. But I couldn’t. The realization that we had come so close haunted me. If we hadn’t lost game eighteen, then maybe the match would have been all even going into game thirty-nine. We wouldn’t have fiddled with the program, and the loss in game thirty-nine wouldn’t have occurred (without the changes we made to Chinook, the correct d2-c3 would have been played instead of the d4-e5 loser). Instead of desperately trying for a win, the match would have been all even with two games to play. We came that close to holding the Terrible Tinsley to a drawn match. It all came down to game eighteen and the sudden change in fortunes for both sides.


Silicon Graphic World Championship Search Depth Weak Side Deep Search 
These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.


Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.


  1. 1.
    Tom Landry, in “Silicon Graphics World Draughts Championship,” the daily bulletins of the 1992 world man-machine checkers championship.Google Scholar
  2. 2.
    “The Mechanical Grandmaster,” Economist, August 29, 1992. ,Google Scholar
  3. 3.
    Tony Buzan and Barry Buzan, The Mind Map Book, BBC Books, London, 1993, p. 285.Google Scholar
  4. 4.
    CNN interview, shown on October 10, 1992.Google Scholar
  5. 5.
    Once the databases were complete, I had Chinook analyze that game. As it turned out, the program had the right answer in 1992, but possibly for the wrong reason.Google Scholar
  6. 6.
    Peter Jansen, my friend from my Carnegie Mellon days, did his thesis on this topic. He created a chess endgame program that tried to trick the opponent into a mistake. Peter did an excellent piece of work, but the implementation overhead of his ideas has prevented them from finding their way into strong programs. See Peter’s Ph.D. thesis: Using Knowledge About the Opponent in Game-Tree Search, School of Computer Science, Carnegie Mellon University, 1992.Google Scholar
  7. 7.
    Rob Lake, personal communication, February 1, 1996.Google Scholar
  8. 8.
    Unfortunately, 1992 was the last year for this event. To David Levy: it was a great idea; please resurrect it.Google Scholar
  9. 9.
    Before the match, I tried to see whether we could automate the Colossus-Chinook games. I wanted to write a program that transmitted the moves between the two, eliminating all human intervention. That way we could have the programs play each other twenty-four hours a day. Unfortunately, the differences in the computer operating systems, UNIX and Microsoft Windows, made this difficult to do on short notice. Today, many computer-only tournaments have been automated so that the games are played without any human assistance.Google Scholar
  10. 10.
    I had completely forgotten about this anecdote until Martin Bryant wrote me about it, August 5, 1996.Google Scholar
  11. 11.
    The dogs are named Caissa (after the mythical goddess of chess) and Casanova (because of his attitude towards female dogs). We have two cats as well, Mischief and Mistake, so named because I felt it was a mistake to get them since they would cause a lot of mischief. In 1994 we added another cat, Miscellaneous. All the animals look forward to the rewards that they get when Chinook wins a game and when the Edmonton Oilers hockey team wins a playoff game (unfortunately, a rare occurrence these days).Google Scholar
  12. 12.
    The following few pages are almost all e-mail messages. I felt that it was necessary to express people’s opinions in their own words so that there could be no misrepresentations.Google Scholar
  13. 13.
    E-mail sent on August 10, 1993.Google Scholar
  14. 14.
    E-mail sent on August 11, 1993.Google Scholar
  15. 15.
    E-mail sent on August 15, 1993.Google Scholar
  16. 16.
    E-mail sent on August 15, 1993.Google Scholar
  17. 17.
    E-mail sent on August 15, 1993.Google Scholar
  18. 18.
    After the book decision, Norm kept in touch with us and followed Chinook’s progress. However, he stopped working on the project.Google Scholar
  19. 19.
    Of interest is that this is the opposite of what we see in chess. Chess programs do better against human opponents with faster time controls (such as speed chess, five minutes a side). In checkers, because of the ever-present danger of traps that require a deep search to uncover, I felt that the program performed better against humans given slower time controls.Google Scholar
  20. 20.
    Yes, the same Limburg as in Limburger cheese. Without a doubt, Limburger cheese is the foulest-smelling cheese I’ve ever had the “privilege” of sampling. It must be an acquired taste—there’s no other rational explanation for sacrificing one sense (smell) for another (taste). In 1996, the university renamed itself as the University of Maastricht.Google Scholar

Copyright information

© Springer Science+Business Media New York 1997

Authors and Affiliations

  • Jonathan Schaeffer
    • 1
  1. 1.Department of Computing ScienceUniversity of AlbertaEdmontonCanada

Personalised recommendations