the Horizon effect—how to beat your computer

Recently, I've run into a few articles and lectures by Yaacov about the reversed stonewall attack; specifically how it is useful against computers because of the "horizon effect." This is best described as the movement of pieces behind the lines of your pawn wall, ultimately leading to a breakthrough on the opponent's king that is subtle and not readily obvious.

The following game is one in which I crush Sigma Chess. Although it is not a stonewall attack, it is a closed game with a good example of the "horizon effect" that is so useful against computers—It bears keeping in mind that I've had great success with the stonewall attack and this strategy against human opponents as well.

Checkers is for tramps (W) 1576 Sigma 1700 (B)
King's Indian Defense, mar del plata variation

An early f4 is the main idea for White in this position, and if Black is an attacking player, she either needs to play the knight to the wing early and play f5 ASAP, or place the knight safely on e8 immediately after castling in order to prepare the f5 pawn push.

Here, the pawn push on the king side is starting to take shape, and the computer calculates that taking is incorrect because it gives me open lines to the king, which technically is correct, however, the computer does not look far enough ahead to see that the pawn advance threatens to push it off the board fifteen moves later, when the bishop and knight will work in tandem around the king. Therefore, gxf5 in this position makes more sense than the passive Nh5, despite the fact that it creates a weakness in front of the king.

Now, gxf5 is too little too late. The computer begins to recognize the severity of the spatial crunch. Here, h4 for White is a necessary measure to prevent fxg5 followed by a collapse of white's aggressive pawn wall. In the near future, white's main idea here is to dump the knight on e2 in exchange for black's knight on the rim, with the hope of opening a line to the king and bringing the dark squared bishop into the attack.

White's king steps up to facilitate the exchange of the knights, with the long-term expectation of opening up a line along the g or h file.

While Bxh7 seems better at first glance because the king escapes immediate mate, it's actually losing after 24 … Bxh7, 25 Qxh7+ Kg8 26 Qe3 Kf8 27 h5 … when the passed pawns on the g and h files become "self propelled." At this point, Black resigned, however, I offer the following continuation for the purpose of flushing out the complete collapse of Black's defense.

The pin is the final nail in the coffin, and Black's position, which has been hopeless for ten moves, is now 1 move from mate.