Independent Pursuits: Chess

IN THE old days of computer chess, programmers sometimes came across a curious anomaly as they began working with faster machines. Sometimes, the same program would play worse when it analysed more deeply. What was happening, it seemed, was that the machine picked the best move from a superficial look at the position, then talked itself out of it by calculating too much. Humans often do the same.

Look at the diagram position, from Summerscale-Levitt, British Championship 1998. It is White to play. A reasonably strong player might see nothing wrong with the move 1.Qh3, threatening a devastating attack down the h-file. A slightly stronger player would see the possibility of 1.Qh3 Nf2+ 2.Rxf2 and now Rh6 or Rh8 wins the white queen.

Now put yourself in Black's position after 1.Qh3. A decent player would see the threat and make a haven for his king with 1...Rg8 to meet Qh7+ with Kf8. A better player would find the idea of Nf2+. But what, in each case, would a very strong player conclude?

Here is how the game went: 1.Qh3(!) Nf2+ 2.Rxf2 Rh8 (2...Rh6 is met by 3.Qxh6+! Kxh6 4.f6 when the threat of Rh2 mate forces Black to give back his queen.) 3.f6+! Qxf6 (3...Rxf6 loses to Rxg5+) 4.Qxh8+ Kxh8 5.Rxf6 Rxf6 6.Rxg5 and White's extra piece was enough to win the game.

So why is the exclamation mark consigned to brackets after 1.Qh3?

Well, after 1.Qh3 Rg8! Black has a better chance of saving the game. The best move may well be 1.Qh4! when White retains the option of exchanges on e4 after 1...Rg8. But then we would have missed all the fun.