Chess

DEEP Thought, the IBM chess super-computer project, like all players with hopes of becoming world champion, analyses rivals' games and rudely points out their errors. It must have been gleefully whirring its cogs at the eighth game of the Spassky-Fischer match, where it has proved that Spassky's blunder was actually a good move; his real mistake came two moves later than he thought.

The diagram position was the scene of the crime. Spassky played 31. Rc1] The question marks are because Spassky clearly overlooked his opponent's reply; the exclamation mark is added because it should not have mattered.

Play continued 31 . . . Ncb3 32. axb3 Nxb3 when White, threatened with Nxd2+ or Rxc1+ appears lost. Spassky's 33. Rc6? Nxd2+ 34. Rxd2 just surrendered without a fight.

The move discovered by Deep Thought is 33. Qc3]] After 33 . . . Rxc3 34. Rxc3, White threatens both Rxb3 and Rc8+, so 34 . . . Nc5 is forced, when 35. b4 wins the knight, leaving White on top after 35 . . . Qa7 36. bxc5 dxc5 37. d6]

Black does better with 33 . . . Nxc1 when 34. Qa3 confuses matters. White threatens Qxa6, while also having chances of picking up the cornered knight. 34 . . . b4 35. Qa4+ also leads to advantage for White. Black's best is probably to try to open lines for his queen with 34 . . . f6, but the position remains very obscure.

Yesterday's answer: 1. Ba1] Bxa1 2. Kf8] mates next move.