This is the latest in a growing list of books that focuses on using computers to improve your chess play, and while HOW TO USE COMPUTERS TO IMPROVE YOUR CHESS and CHESS SOFTWARE USERS GUIDE had their good points, the third time is the charm. MODERN CHESS ANALYSIS expands upon many of the concepts introduced in those earlier books in a way that should allow most serious players to improve their chess analysis and play.

While not a well-known over-the-board player, Robin Smith has qualified for the correspondence chess grandmaster title and has twice been USA correspondence champion. He has also won a world correspondence chess championship semi-final. The use of chess software to assist in finding relevant games, plans, and positions is particularly important in world class correspondence chess, and it is not surprising that a strong correspondence player can provide a range of methods for assisting chess analysis via chess software.

The book starts with some excellent hands on advice on what to look for in purchasing a computer for use as a chess aide. The author discusses the critical nature of processor speed, suggests that the AMD Athlon is currently superior to Intel Pentiums of the same price, and explains the advantage of choosing a multi-processor machine. He also touches on a variety of other useful concepts, such as the amount of memory needed to use hash tables (use of a portion of RAM to detect transpositions, which speeds analysis by not requiring the same position to be analyzed over and over via differing move orders). Players in the market for a computer to use for chess-related activity will probably find the book worth its purchase price based on this five page introduction alone.

The depth of the discussion compared to other similar books is demonstrated in the first chapter, which deals with the relative strengths of computers versus human players. Anybody who has read or observed computer versus human games knows that computers are superior calculators and humans are better at long range planning. While this chapter validates those basic concepts, it does so with a depth and breadth not found in other books. For example, the author compares and contrasts the relative strengths of various chess playing programs in the areas of depth of calculation, positional calculation, and assessment of positions.

I found the explanation of how differing computer programs come up with their numerical assessment of a position to be particularly eye opening. In most cases, this evaluation is an aggregate score, assessing, for example, in the case of the Crafty chess playing program, material, development, pawn evaluation, passed pawns, passed pawn race, king safety, and piece interaction.

Of course, while these are important aspects of a position, they do not all weigh equally in most instances. The author illustrates the types of positions where computer assessment is inaccurate because these various features do not really explain what is happening on the board. Likewise, the book shows positions where even very strong players have missed ideas that the computer will pick up. As the author notes, players often get blinded to alternate ideas in positions because they have been so wrapped up in their own analysis.

The first chapter (which may be the book's best) also includes an interesting section on exceptions to the chess “rules” that computers generally use in assessment and deciding upon moves. For example, exchange sacrifices, even of the positional variety, have become very common in modern chess, but most computers eschew them and misevaluate them, particularly in the middlegame. The author also argues that any unusual material imbalance is difficult for the computer. As he notes, “a human will often have a better chance of sorting out unusual material imbalances than a computer. This is because the computer has nothing but its rules for the value of pieces, and its search, to determine which exchanges are good or bad.”

The first chapter sets the tone for the rest of the book, because it helps to explain when computer-assisted analysis may be useful, and when it should be at least treated with caution. The second chapter builds on the discussion by examining various methods for using the computer for analysis. The author argues that the best use of the computer is as a sparring partner, where the human player comes up with plans and uses the computer to check on their implementation. With this method, the strengths of both the human player (in planning) and the computer (in calculating) are maximized. By showing how computers assess a variety of positions from grandmaster games, the book discusses how computer analysis can (or cannot) assist in understanding positions. Along the way, there are lots of useful basic tips – for example, many computers generate bizarre PVs (their lines of thinking). While the first move these generate is the best move, if you go far beyond that, you can often end up with nonsense.

There are lots of practical tips, such as explaining how to use backward analysis (starting from a position with a large edge and moving back to a starting position) to find new branches worth analyzing, and multi-variation analysis mode to examine alternate tries worthy of analysis.

It is evident from all the examples the author provides that this has been a well researched book. In many cases, the author discusses the relative merits of various programs (Fritz, Junior, Crafty, Chess Genius, Chess Tiger, Chessmaster, Shredder, Hiarcs, etc.) and compares their methods of analyzing and handling various positions. As the author notes, there can be substantially differing results based on the program. For example, the author shows one sharp position where a couple of strong programs find the correct move in a couple of seconds, while a similarly strong program won't find it after more than 24 hours of analysis. This and other examples explain the benefit of running more than one engine on a position.

More than anything, I was struck by how often the book acts as a cautionary tale about reading too much into computer insight. For example, in the line after 1.e4 c5 2.Nf3 e6 3.d4 cxd4 4.Nxd4 Nf6 5.Nc3 d6 6.Be3 a6 7.f3 b5 8.Qd2 Nbd7 9.0-0-0 Bb7 10.g4 h6 11.h4 b4 12.Na4 Qa5 13.b3, the author notes that “by this point most programs begin seeing a slight advantage for Black, and will start suggesting alternative moves for Black, compared to what human GMs have tried. For example, Fritz seems to like 13…e5. Is Black really better in this line? No! Computer evaluations cannot be trusted, even in tactically rich and complicated middlegame positions such as this one.” The author's suggestion is to conduct an “engine tournament” among strong programs as a way to test these evaluations.

In the above position, the author's 8 engine double round robin revealed that white scored 31 wins, there were 8 draws, and 17 black wins, for a 63% overall result for white. As the author notes, the first two moves after the novelty, 14.Nf5 d5 are common to most of the games. Things then get “crazy. But the crazy complications, with the knight menacingly posted on f5, tend in the end to favor White…The lines are just too deep for the programs to see, and the position too hard for either computer or human to evaluate clearly.”

The remainder of the book examines computer use in the opening, middlegame, and ending. In each chapter, there is the same focus on practical application of use of the computer, examples of how computers cope with positions from high-level contests, and direction on areas where computers may not be up to the task. Throughout, the author's grasp of various programs and chess software is impressive – there is, for example, a discussion of the Bookup program, which can be used for opening study. While most authors concentrate just on the Chessbase program (because that is the database most utilized by strong players), Smith provides insight that branches out from the ordinary. This is what helps to make this an extraordinary book.

This is a well written and produced book, which has become the norm for Gambit Publications. It is not, however, a mainstream book – it is mostly written for strong players seeking to do specific kinds of research and analysis. In fact, the many areas where computers are still lacking point out the danger for the average player in relying too much on computer-generated analysis. One must be very cognizant of the shortcomings discussed in these pages before embarking on computer assisted chess analysis.

In conclusion, the use of computers in chess has become a topic of interest to most players, and this book does a better job than any other at explaining how to maximize that use for the serious player. There are hands on tips galore, and any player interested in purchasing a computer for chess-related study will benefit from the introduction alone. In short, if you plan to use your computer for chess study at a high level, you will need this book.