THE idea of a machine that can play chess has captivated man's imagination for centuries, but serious attempts to create one began only about 30 years ago. And it is only this decade that enough progress has been made to suggest they may soon outstrip all human competition. ``I think it's inevitable,'' says Murray Campbell, one of the creators of Deep Thought, the current state-of-the-art model. ``We're hoping that we're going to be developing the ultimate chess-playing computer ... in the next few years.''
Long before technology like this existed, men dreamed about it. There were fictional stories on the subject, and from 1770 to 1834 an ``automaton chess player'' was exhibited throughout the world, only to be exposed as a hoax with a human player hidden inside the machinery.
As for the real thing, the first major efforts occurred in the early 1960s with teams headed by former world champions Max Euwe of the Netherlands and Mikhail Botvinnik of the Soviet Union. Various universities and scientific laboratories in the United States also got involved. But progress was slow and erratic at best.
The problem is far more difficult than most laymen imagine. Because of the finite nature of chess (64 squares, 32 pieces), one might assume that modern computers could calculate all possible moves and outcomes. In fact, there are an estimated 10 to the 120th power (a 10 followed by 120 zeroes) possible moves.
Most of those moves would seem useless or ridiculous even to a novice human chess player, however. The trick is to program the computer to evaluate moves, giving value to the pieces and to such concepts as control of space, mobility, king safety, etc. Because of the complexity of chess, this has proved very difficult.
In the 1980s the picture began to change. A program named Belle, developed at Bell Laboratories, was the first computer to achieve a master's rating. A team at Carnegie-Mellon University (CMU) headed by chess master and computer programmer Hans Berliner created a program called Hitech that won several tournaments against human opposition (including the Pennsylvania State Championship). Then also at CMU a group of graduate students developed Deep Thought, which has stunned the chess world with a string of successes over the past year and a half and is now getting a crack at world champion Gary Kasparov. (See story above.)
The original Deep Thought team consisted of Feng-Hsiung Hsu of Taiwan, who devised the chip that gives the computer its incredible calculating speed; Campbell, an expert chessplayer who put in most of the evaluation function and search techniques; and Thomas Anantharaman of India, who did most of the programming. They have been joined by Andreas Nowatzyk of West Germany, and Peter Jansen of Belgium.