A position from the recent computer chess match between the programs Rybka and Houdini. Houdini, as black, sacrificed first one pawn, then another, and just now a third for more active pieces. I would be getting distinctly nervous as black here. Sure, black’s pieces are more active, but three pawns is a lot. If black can’t extract some tangible gains from his superior pieces, he’ll soon find himself in a hopelessly lost endgame. Indeed, there are several ways to regain a pawn immediately, options I’d be considering intensely. But Houdini played the icy cool …Kf7. I can see why this move is useful — it gives the rooks free passage along the back rank — but why it’s the best move at this moment, with pieces hanging all over the board, is beyond me. The conclusion proved Houdini right. It coordinated its pieces into an attack against Rybka’s exposed king, forcing the win of a piece, and then the game.
The computers have come a long way. Early chess programs were notorious for over-valuing material advantages (having more pieces than the opponent) at the expense of positional considerations. The newest, best programs play at the opposite extreme, sacrificing pieces and pawns for dynamic advantages. Rybka and Houdini are arguably the two strongest chess-playing entities in the history of the universe. It would have been easy to imagine their encounter, a more perfect brand of chess, would be dry, clinical, and boring, but in the event it was dynamic, even romantic. Another reminder that the most rational strategy is not, as is often supposed, the most conservative.