Man vs. Machine: Challenging Human Supremacy at Chess

by Karsten Müller, Jonathan Schaeffer

Overview

Man vs. Machine Technology continues to advance at a rapid pace. It may sound quaint today, but not so long ago, computers battled humans for supremacy at the game of chess. The challenge of building a computer program capable of defeating the best of human-kind at chess was one of the original grand challenges of the fledgling field of artificial intelligence. On one side were dedicated scientists and hobbyists who invested decades of effort developing the software and hardware technology; on the other side were incredibly talented humans with only their determination and preparation to withstand the onslaught of technology. The man versus machine battle in chess is a landmark in the history of technology. There are numerous books that document the technical aspects of this epic story. The human side is not often told. Few chess players are inclined to write about their man-machine encounters, other than annotating the games played. This book brings the two sides together. It tells the stories of many of the key scientists and chess players that participated in a 50-year research project to advance the understanding of computing technology. “Grandmaster Karsten Müller and Professor Jonathan Schaeffer have managed to describe the fascinating history of the unequal fight of man against machine in an entertaining and instructive way. It evoked pleasant and not so pleasant memories of my own fights against the monsters. I hope that their work gives you as much pleasure as it has given me.” – From the Foreword by Vladimir Kramnik, 14th World Chess Champion

Product Details

ISBN-13:	9781941270974
Publisher:	Russell Enterprises, Incorporated
Publication date:	11/05/2018
Sold by:	Barnes & Noble
Format:	eBook
Pages:	480
File size:	26 MB
Note:	This product may take a few minutes to download.

About the Author

International Grandmaster Karsten Müller is recognized as one of the world’s top endgame experts. He is the author of many books on endgames and chess tactics. He is the author of over a dozen chess books published by Russell Enterprises.
Jonathan Schaeffer is a Professor of Computing Science at the University of Alberta in Canada. For over 35 years he has been doing research in artificial intelligence using games and puzzles to demonstrate his ideas. Schaeffer’s scientific paper, “Checkers Is Solved,” was a runner-up in Science’s breakthroughs of the year for 2007, was named by Nature’s readers as the ninth Most Important Achievement of the Year, and was one of the New York Times’ “Ideas of the Year.” Schaeffer’s other game successes include Phoenix, a program that tied for first place in the 1986 World Computer Chess Champio

Read an Excerpt

Chapter 7
2750 (1996-1997)
There was no doubt that IBM wanted to see a rematch. Internal IBM assessment
indicated that the match generated enormous and favorable publicity for the company, especially for the company's stock price. With that much value at stake, the decision to invest in the DEEP BLUE team way back in 1989 was beginning to look like a bargain!

The DEEP BLUE team did a postmortem analysis of the match and came to several
conclusions:
• Search: No major issues arose. The searching capabilities appeared sufficient. The program was tactically strong.
• Knowledge: Kasparov clearly had a greater positional understanding of the game,
and that was where major improvements were possible for the program. This meant
1) identifying more chess knowledge that would add value to the program,
2) putting that knowledge on the chip, and, most importantly,
3) tuning the program so that the knowledge was applied in the right circumstances with the appropriate weighting.
• Openings: One can never do too much opening preparation! However, there were special cases that were not properly planned. For example, the team should have had their opening book prepared for a line to play as Black in their “must win” game 6.

• Match strategy. Consider possible match scenarios that might arise and decide in
advance how they will be handled. For example, the team did not know how to respond to Kasparov's draw request in game 5.
The chess knowledge aspects were the most challenging. Sometimes DEEP BLUE
would play a “bad” move and the team would dive into the program's innards trying to figure out what when wrong. But sometimes things were not as they seemed (Campbell 2005):
There were examples of that in the games against Kasparov — moves that seemed
counterintuitive or just wrong, just plain wrong. But there was a logic to them, and you could reconstruct that logic. I saw plenty of examples of that as we were programming and preparing DEEP BLUE to play the match against Kasparov – the two matches. Many cases where we were very upset about a move that it played, and we
would assume it was a bug and conduct a deep investigation and find out it had to play that move – there was no choice. Any other move would have lost, or significantly worsened its position. But intuition sometimes let us down into thinking we knew what was going on, when we didn't always. Further, even identifying possible errors was hard.

Editorial Reviews

"...entertaining and instructive...It evoked pleasant and not so pleasant memories of my own fights against the monsters. I hope that their work gives you as much pleasure as it has given me.” —From the Foreword by Vladimir Kramnik, 14th World Chess Champion

From the Publisher