The reputation of chess computers rose — and fell — repeatedly during Vishy Anand’s successful defense of his world championship title this month.
Chess-playing programs, like the popular Fritz, proved invaluable in calculating variations, such as the winning Anand attack in the 12th game. Fans following that final game in real time online knew the dramatic finish before it happened.
But in other games fans were mystified when computers saw huge endgame advantages that didn’t pan out.
For example, some engines claimed Anand had a huge edge when he had a king, two rooks and a pawn to fight challenger Veselin Topalov’s king and queen in Game 8. But computers have a problem visualizing perpetual check, and the game was quickly drawn.
In the previous game, computers kept saying that Topalov had a winning edge when he had an extra pawn in an endgame with bishops of opposite color.
But in “B’s-of-opps” endings, an extra pawn is easily blockaded. Exploiting the pawn typically requires visualizing a plan — a human strength — rather than calculation, a computer’s strong suit. When Topalov won, thanks to an Anand blunder, some machines — as well as some GMs — were caught off guard.
Source: NY Post