In February of 1996, Kasparov played Deep Blue in a six-game, full-length regulation match sponsored by the Association for Computing Machinery in celebration of the fiftieth anniversary of the computer. Before the match, Kasparov was quoted as saying, "To some extent this is a defense of the whole human race." When he lost the first game, his computer adviser, Frederick Friedel, openly acknowledged that Kasparov was devastated (see figure 5.1). Even though he rebounded to win the match, Time magazine called the first-game defeat an event larger than "world historical. It was species-defining."

Other grandmasters refuse to play against computers at all. Why? Perhaps because the idea of computer superiority in an arena as cerebral as chess is so disorienting; in Western culture, many consider chess the ultimate test of the human intellect. (It is interesting that Kubrick originally filmed the "chess scene" with a five-in-a-row board game called pentominoes but chose not to use it, believing that viewers would better appreciate the difficulties involved in a chess game.) Mathematicians have estimated that there are more possible chess positions than there are atoms in the universe. Therefore, skilled chess players must possess the ability to make difficult calculations and recognize a seemingly infinite number of patterns. Yet excellent chess play also requires imagination, intuition, ingenuity, and the passion to conquer. If a machine can beat a man at a game requiring as much creativity as chess does, what does that say about our "unique" human qualities?

For now, at least, we can rest assured that even though the best computers are better at chess than 99.999999 percent of the population, they do not actually play chess the way humans do. In the history of man's rivalry with machines, only one grandmaster-level computer has appeared to play like a human -- and that computer is our fictitious friend HAL.


How HAL Plays Chess

By analyzing the game between Frank and HAL, we can uncover a number of clues about how HAL plays chess. As I explain in more detail below, HAL appears to play chess the way humans do. Even more important perhaps, the game reveals that HAL is not simply mimicking the way humans play; he actually understands how humans think.

The game in the screenplay is a real one played by two undistinguished players in Hamburg in 1913. Kubrick, a former Washington Square Park chess hustler and aficionado, selected a clever checkmate but was careful not to employ one too complex for viewers to grasp. He picked a position from a fairly obscure game -- one obscure enough not to appear in the 600,000-game database of Deep Blue. I eventually located the game after being directed to an article written by Grandmaster Larry Evans on January 12, 1990 (HAL's birthday).

Evans makes the crucial point in his article that HAL should have said "Queen to Bishop six" (not three). HAL used the so-called descriptive notation system that describes moves from the viewpoint of the moving player. This contrasts with the algebraic-notation system used in the game score (see Appendix to chapter), in which moves are described from White's viewpoint. HAL used the incorrect viewpoint when giving his fifteenth move. Was the notation error a deliberate foreshadowing of the machine's fallibility or merely a writer's oversight? This is a question only Kubrick can answer. If Poole had been a little more attentive, he might have realized sooner rather than later that the HAL 9000 was indeed capable of error. But, like most chess players, he was focusing on the actual moves; he was not looking for errors because he had never even considered the possibility that HAL was capable of making one.

To better understand how HAL chooses to play against Frank, it is important to have some sense of Frank's chess background. Although the movie does not disclose his chess rating, it is easy enough to speculate about his skill level. He is a highly educated man who holds a doctoral degree, most likely in a field such as aerospace engineering or robotics. We can surmise that, as second in command on a top-secret space mission of unprecedented importance, Frank is well above average intelligence. Because he is a full-time astronaut, it is unlikely that he would have time to compete in professional chess tournaments; yet he clearly knows something about chess, for his game with HAL follows opening theory for eleven moves (see Appendix for a complete account of the game). Frank's chess rating may be in the expert range, making him strong enough to engage in an interesting game but certainly not experienced enough to handle HAL. (See figure 5.2 for an explanation of the rating scale.)

In the game itself, Frank plays White and HAL is Black. Frank chooses an unusual but perfectly sound variation of the well-known Ruy Lopez, or Spanish opening. HAL responds with very aggressive play, creating a situation that makes it very difficult for Frank to find the best moves. By the time the movie picks up the game, Frank has already made the losing move, and he goes down without much of a fight.

The game provides sufficient evidence that HAL plays chess the way humans play chess. Early in the game HAL uses an apparently nonoptimal but very "trappy" move. The choice creates a very complex situation in which the "obvious" move is a losing blunder. If Frank had been able to find the best move, he would have gained the advantage over HAL. In leading Frank into this trap, HAL appears to be familiar with Frank's level of play, and we can assume that HAL is deliberately exploiting Frank's lack of experience. < /P>

The interesting point here is that present-day chess programs do not normally play trappy chess. They are almost always based on the minimax principle, which assumes that the opponent always makes the best move. (I discuss this principle in more detail later in the chapter.) A machine like Deep Blue, therefore, would only play the optimal move found in its search. The ability of HAL to play trappy moves is a sign of a sophisticated player who is familiar with the opponent's strengths and weaknesses.