IBM's Gambit Declined

by Mickey Z.

Dissident Voice
January 25, 2003



Just as we're submitted to television spectacles of sprinters racing giraffes, Dylan Loeb McClain of the New York Times informs us "Garry Kasparov, the world's top-ranked player and the former world champion, will play a $1 million, six-game match against a chess program called Deep Junior" ("If a Machine Creates Something Beautiful, Is It an Artist?" January 25, 2003). This will be the first time Kasparov has "matched wits against a computer," as the Times calls it, since losing to Deep what appeared to be one long commercial for IBM in 1997.


"Whether Mr. Kasparov wins or loses, clearly chess computers have reached a point where they can compete against, and sometimes beat, the world's best players," McClain says before reminding that even Kasparov "admits that the point at which computers consistently play better than humans is probably not that far off."


We also learn from this article that in 1997, a Stanford University pitted a

human against a computer to see which could compose music in the style of

Bach. "The computer won," the Times reports without elaboration.


The question McClain draws from all this is: "If computers become better than humans at chess, does that mean that computers are being artistic or that chess is essentially a complicated puzzle?"


For me, the questions lie deeper.


Just as we commonly and unselfconsciously describe an airplane as "flying," we may innocently use the word "think" or "play" or "compose" to define the actions of a programmed computer-should we feel impressed enough with the similarities to adopt such a metaphor. However, while very few people truly believe that airplanes fly (or that submarines don't "swim," they "set sail"), the concept of a machine out-thinking (or out-playing) a human has gained widespread acceptance thanks to the venerable game of chess.


In the 30 years since Bobby Fischer launched chess onto the pop culture radar screen, technicians have spent countless hours designing computerized chess programs. Lost amidst the hype surrounding Big Blue's "victory" over Kasparov is the simple reality that computers can't think. Hence, no matter who "wins" when man confronts machine, such a match-up can tell us precious little about the human mind and even less about chess playing.


Noam Chomsky, institute professor of linguistics at Massachusetts Institute of Technology, has written and lectured widely about issues of human intelligence, like airplanes flying, submarines setting sail, and machines playing chess.


"All of these things are questions about sharpening and altering usage, they're not questions of fact," Chomsky says. "Well, there's a completely separate question which shouldn't get confused with this, and that's the question whether simulation might teach us something about the process that's being simulated. So, you could ask the question, would some chess-playing program, for example, teach us something about human thought. The answer to that is 'certainly not.' "


The reason for this is simple. The best chess-playing programs succeed by abandoning any notion of simulating human thought and instead relying on the unique capacities of computers.


Even McClain, in the Times, concedes this point. "People rely on pattern recognition, stored knowledge, some calculation and that great unquantifiable - intuition," he writes. "Computers, on the other hand, have a database of chess knowledge but mostly rely on brute force calculation, meaning they sift through millions of positions each second, placing a value on each result. In other words, they play chess the way they attack a large math problem."


"The better computer programs are ones that simply use the fact that computers can do things extremely fast," says Chomsky. "So you can get a bunch of grandmasters to sit there for year after year and program in answers to just about every possible problem anybody can think of.... So, as simulation, this is a stupid topic. It's hard imagine a more stupid topic if you're trying to learn anything about humans."


Snowdon Parlette is the author of The Brain Work-Out Book: Aerobics for the Mind. "Computers operate in a slavishly logical fashion and are unable to do what we call reason," he explains. "Human brains act more like a large committee coming to a consensus after having considered all the relevant information and taking input from many different sources."


"Incidentally," Chomsky adds, "a computerized chess program has no other purpose either, as far as I can see, except maybe taking the fun out of playing chess."


Well, that and selling plenty of software.


Your typical desktop computer is the result of a few decades of work and is composed of roughly five million transistors while the human brain-at least two hundred thousand years in the making-has more synaptic connections than the number of known stars in the universe. Thus, like a dolphin racing a submarine across the Pacific or an eagle pitted against an F-16 from coast-to-coast, the man vs. machine "contest" is utterly meaningless.


Such spectacles would be best left to Rupert Murdoch's network, not only because the supposed simulation teaches us nothing about the act being simulated but, more importantly, the brute force of a man-made machine will never be a match for the living and creative beauty of a dolphin swimming or an eagle flying...or a human being thinking or playing.


Mickey Z. is the author of The Murdering of My Years: Artists and Activists Making Ends Meet and an editor at Wide Angle. He can be reached at: