For some time I have been interested in computer chess and am in the early stages of planning my own chess-playing program. In particular, I am interested in writing a highly tweakable chess-playing engine that will allow me to see how applying different search and evaluation techniques affects the overall speed and quality of the evaluations generated by the engine. I searched for a while for good resources that do more than just mention different concepts, looking more for detailed explanation. I stumbled on this book and decided to give it a chance. I'm very pleased that I did so.
This book is not original material. Rather, it is a collection of previously-written articles, compiled together by David Levy who is a well-respected authority on computer chess topics. He has divided the material into nine sections that make locating information easy: The Introduction, Famous Chess Programs, Position Evaluation, Tree Searching Techniques, Analysis, Writing a Chess Program, Special Purpose Hardware and Software, The Endgame, and Games Played by Chess Computers.
Most online sources mention different techniques (such as alpha-beta searching) but offer precious little in explaining what the techniques actually are or how to implement them. All too frequently, references are made to some open-source engine that implements one or more techniques but is so poorly documented (or is documented in another language) such that understanding is just out of reach. Where this book is different is that is not only explains the various techniques and ideas but more often than not it gets very detailed and offers diagrams and pseudocode to clearly present ideas in terms an interested individual can access.
There are several places within this book where games played by computers are analyzed and explained in detail, which at first may seem a little wasteful but in context with the rest of the material makes understanding a little more interesting. The games are from older programs and do not represent the quality of play seen in today's modern chess programs but they still provide a good reference point and demonstrate the ability of computers to search and evaluate.
This brings up a good point. Much of the material, while still applicable today, is regarded as older (1988 timeframe). Since the first printing of this book there has been appreciable progress in computer chess strength and playing ability. This may not be the most up-to-date material but it is still all very relevant and instructional and is still foundational for modern computer chess understanding.
Computer Chess Compendium is an exceptionally useful reference for anyone who wants to really understand how computers play chess. This is the reference for which I have been looking and I cannot say enough good things about it. If you want to really get a good understanding of how computers "think" when playing chess, or perhaps you are like me and want to write your own engine, this is a fabulous resource worth much more than the asking price.