Pachman's book is a classic on strategy that gives all the most important principles with his own games and those of Nimzovich, Alekhine, Smyslov and Botvinnik as examples. He covers concepts like play in the center and on the wings, P-structures and how to play with the pieces. He also discusses topics like exchanging pieces, lead in development, coordination of the pieces, etc.
While browsing through it the other day I was intrigued with a position from Pachman vs. Donner that was played in the 1955 match between Czechoslovakia and The Netherlands. As black, Donner had a P-majority in the center, but Pachman stated it wasn't favorable for him because white was able to attack it by piece pressure combined with P-pressure from the flanks and finally destroy it by the advance of the d-Pawn which he had held back in reserve.
It looked like the game would be very instructive, but when I began playing over it with the help of Stockfish and Komodo it became apparent that Pachman's analysis was at odds with these two monster engines. In fact, in a couple of lines Pachman's analysis was so tactically faulty as to render his conclusion on the line meaningless.
I have found many other similar examples in old books. Obviously in any chess book errors are unavoidable, especially those written before modern engines were invented, but does that mean those old books have no value? I don't think so.
We don't play engines; we play humans, at least you do if you play over the board. What's important are ideas and the author’s ability to communicate them to his readers. Old books like Pachman's, Euwe’s two books on the middlegame, Lasker's Manual of Chess, Bronstein's Zurich International Chess Tournament and the game collections of great players of the past are timeless classics that never grow old.
Jeremy Silman probably said it best when he advised that when an engine disagrees with assessments of the authors of these great old classics the best thing to do is concentrate on what the author is trying to share with you.
I like playing over games using an engine because it gives me a chance to try out a whole bunch of different moves and see why they don't work. Sometimes there will be tactical mistakes that refute what the author has written. When that happens or when the engine is saying things are equal but the author claims one side has an advantage, what I usually do is run a Shootout in my Fritz GUI. Playing the game at 17-25 plies results in five games and gives a pretty good indication of the true evaluation. What I usually do then is look at the game played at 25 plies and try to get an idea of the general drift the play took.
For example in the above mentioned Pachman – Donner game, at one point in his analysis Pachman commented that white's central position combined with the open g-file gave him a decisive K-side attack. It wasn't so; in fact, the opposite was true as indicated in Shootouts with a couple of engines and all the games were won by black! Even so, Pachman was correct in his overall assessment because had white played correctly he likely would have won. As Shootouts indicated, white had better control of the center and his K-side was adequately defended so the correct strategy was to commence playing on the Q-side which lead to victory in every case. That was because black's K-side attack was stalled and he was too tied up to defend against the Q-side attack because white controlled the center.
It's been pointed out (by Silman, I think) that a lot of modern instructional books which rely heavily on engine analysis often have a lot of engine generated variations that have no instructional value. They are included simply because they may be a few centipawns better. The fact is that in many cases humans are not capable of finding a move that an engine may suggest. Humans rely on patterns, concepts and ideas and are not capable of calculating millions of positions. Instructional manuals are designed to teach us to understand concepts that we can repeat. When GM Gregory Kaidanov is analyzing with an engine and it suggests a move that he would never consider and there likely is no realistic possibility of playing or even noticing it in a real game situation he ignores it even though it may be the absolute best move.
Looking for some good chess books that are enjoyable to read and from which you can actually learn something? My advice is 1) take a few minutes to learn descriptive notation and 2) buy those cheap but well constructed Dover chess books!
Play over the games and try to absorb the ideas the author is trying to teach. Even if Alekhine, or Botvinnik or Fischer missed something that Stockfish spotted in a fraction of a second you can still learn something from those guys. I believe even if your move choice was limited to moves those legends rejected you could probably make a living playing chess.