How Computers Play Chess

In 2009 I was a member of the Organizing Commitee of the World Computer Chess Championship and I had the honour to know David Levy [...]

Is a very interesting man, very kind and a real patriarch of the chess computer.
He talks about the computer chess story and was a pleasure listen him.

This book is a deep explanation about this subject until the 90's.

Excellent book -- it's the highlight of my extensive computer chess book collection.

I read it in high school in the early 1990s after some failed attempts to write my own chess engine. Highlights were clear sections on alpha-beta pruning, killer move heuristics, hash tables for position transposition detection, and quiescence search. This was actually my first introduction to those topics, so for me this doubled as a "random topics in computer science" text as well as computer chess book. I really appreciated all the details, good diagrams, and the clear communication style. The historical perspective is also top-notch; it's great to play through games of early chess programs. I especially liked the diagrams of very early games where the boards used were smaller than 8x8 in order to reduce computation time.

I cant recommend this highly enough if you are interested at all in the history and details of computer chess. Sure, its focus is on the pre-Deep Blue era, but this book has the most important details. I'm confident that by implementing the algorithms in here on 2015-era hardware, a decent programmer would have a chess program that could beat almost anyone without much work.

I was expecting that this new April 2009 reprint of the book, 18 years later, would have included some of the state-of-the-art techniques on how computers play chess these days (circa 2010).But, aside some interesting historical perspectives in the central chapters of the book, there is little new in this book when compared to David Levy's "The Joy of Computer Chess" 1984; that was a huge breakthrough, at that time, on how computers used play chess (IMHO).
It is an interesting book if you are searching the early history of "How Computers used to Play Chess"; but it is not a book on how today (2009~2010) computers play chess.

This excellent book was and still is one of the state-of-the-art books in computer chess literature, dealing with both a very interesting history of computer chess, including many rare examples of the play of early programs, and several recent examples of the best actual programs, as well as discussing in sufficient detail the quite complex technical aspects of programming a computer to play chess.

In "The Challenge Is World Champion Kasparov", we are introduced to one of the first encounters between Garry Kasparov and the former incarnation of Deep Blue, then called Deep Thought, through the detailed discussion of the proceedings and analysis of both games. A game between Karpov and Deep Thought is also carefully analyzed.

Next, "The Early Ideas" present historically the pioneer works of Shannon, Turing, Zuse, and many others, whose theorical works provided the basement for writing procedures to allow a machine to play chess.

Then, in "The First Working Programs", we see Bernstein, Kotok, McCarthy, and other AI specialist, as they struggled to implement Shannon's ideas to make Jurassic computers play some passable chess. Several games between both
computers and humans are discussed.

After these preliminary attemps, "The Formative Years" discusses more advanced programs, such as Greenblatt's MacHack VI program, Botvinnik's Pioneer, and specially Slate & Atkin's Chess program and soviet Kaissa, focusing both on
the internal of the programs and on relevant sample games.

The following chapter, "The Challenges for the Levy Bet", tells us all the details of the famous Levy bet, nicely commented by co-author David Levy himself. The best games between him and Chess are commented, as well as a particularly beautiful miniature of Blitz against Belle.

As the field advances, "The Computer Becomes a Master" discusses the ever increasing achievements of the new generation of stronger hardware-assisted chess programs, such as Belle (written by Ken Thompson, who also has
developed many Endgame Databases) and Cray Blitz (written by Bob Hyatt, who is also the author of Crafty, a strong freeware chess program), which use their incredibly fast underlying hardware to compensate for their lack of chess sophistication. We can also read all about how the first International human Masters began to know defeat against them on a regular basis.

The next step, the defeat of strong human Grandmasters, is introduced in "Eyeball to Eyeball with Grandmasters", where we see several commented games between the strongest chess programs, such as Deep Thought and Hitech, and human grandmasters such as Miles and Larsen. Also, microprocessor
commercial chess program Mephisto has a close encounter of the 3rd kind against macroprocessor non-commercial Deep Thought, and far from ashamed,
beats him hands down !

Once those historical details have been dealt with, the book enters into a discussion of the more advanced chess techniques there are, such as "Endgame Play and Endgame Databases", an area pioneered by Ken Thompson's Belle, where computers have conquered new grounds, and become invincible players. The development of a K+R vs K database is discussed in detail enough to allow anyone to program it, and then both games of the mini-match between grandmaster Walter Browne and Belle, the former trying to mate the computer with K+Q against K+R, are commented in detail.

A very technical chapter follows, "Search Techniques Used by Chess Programs", where the most advanced techniques are explained, such as Minimaxing, Alpha-Beta prunning, Iterative Deepening, and a large, detailed, and complete
explanation of Hash tables, with many diagrams and examples, to make it crystal clear. Other aspects such as Time management, Evaluation functions, Move generation, etc. are thoroughly discussed as well.

The next chapter, "The Evolution of Computing Systems for Chess Programs", explains what lies ahead: faster processors, chess-specific hardware,
multiprocessors, and makes dire predictions on the increment of playing strength all these advances will bring.

Once these almost unearthly machines have been shown, it is the time for down-to-earth-ones, the ones everyone can buy, and "Commecially Available
Chess Computers and Software" introduces them all, from the primitive, very early Chess Challenger, to Mephisto Almeria announcing mate in 7 to a 2350
ELO player under tournament conditions.

On "Writing a Chess Program" gives a concise advice on how to write a chess program oneself, and by way of comparison shows a table with the ELO rating of the best chess programs as compared to that of their programmers and more chess-profficient technical advisors.

Finally, closing the book with a gem, "Stop Press" shows commercial program Mephisto Portoroz defeating former World Champion Anatoli Karpov during a simultaneous exhibition. That such a machine, which anyone could buy, without
any special ultrafast hardware, can defend successfully against as superb a grandmaster as Karpov, says much about how far computer chess has
progressed.

The book closes with an extensive bibliography given in "Additional Reading", and some information on the ICCA, given in "Appendix A: The International Computer Chess Association", and a table with complementary data in "Appendix B: Results of Major Tournaments".