The diagram shows the position after Kasparov's 23...Nc5 in the fourth game. He has sacrificed a pawn and obtained good compensation in the form of pressure down the f-file. Now look at the next few moves: 24.b4 Nd7 25.Qd3 Qf7 26.b5 Ndc5 27.Qe3. Both moves of the white b-pawn provoked horrified reactions among all the human spectators. In an already loose position, White was voluntarily wrecking the shelter around his king. While 24.b4 had the merit of depriving Black of the use of c5, pushing the pawn on another square smacked of complete incomprehension.
But there is another possibility. A human feels complete revulsion at the moves b4 and b5 because of the weaknesses they create. Now just suppose that Deep Blue had calculated sufficiently deeply to be sure that those weaknesses cannot be exploited. Furthermore, suppose that it had also calculated the results of White's not playing b4 and b5, and had concluded that Black's initiative on the K-side can only be countered if White creates an open line on the other wing.
In that case, we would have to agree that b4 and b5 is the right plan, even though it runs totally against our sense of what a good move looks like. Furthermore, we would have to admit that our whole mode of paying chess - by developing strategic judgement and positional sense - is nothing more than a poor substitute for massive calculating power.Reuse content