The world was stunned in May 1997 when Deep Blue, a chess programme running on a high-powered computer, defeated world champion Gary Kasparov in a six-game series. Some called this a victory of machines over man. Is that really so?
Certainly not. Look at it this way. Deep Blue calculates at 200 million moves per second, Kasparov at (according to IBM `statistics') three. And since we could arguably say that they are equally matched players, does this not mean that Kasparov's human weapons --- namely intuition and ingenuity --- are sixty million times better than a computer's number-crunching? In a matter of speaking of course. But we certainly have all the more reason to marvel at the human brain.
So much for the philosophy. How does Deep Blue figure out which move to play next? It obviously considers several billion possibilities. But it also uses a series of complicated formulae that take into consideration the state of the game. These formulae take into consideration things such as material value (e.g. queens are more useful than knights), position (e.g. can you attack more squares than your opponent?), safety of the king and the pace of the game.
Deep Blue also keeps a record of several past matches to see how it can make best use of what's available. As one would expect, it tends to be a very thorough player. Kasparov found this out to his cost when he tried to play a very calculating game against the programme --- he got beaten. On the other hand, when he played some unorthodox moves, he had the computer totally flustered.
Let's leave 1997 behind and return to 1950, when the first chess program was written by the mathematician Alan Turing (who incidentally committed suicide by eating a poisoned apple). Eight years passed before a chess program managed to beat a human --- albeit one who had been taught the game a few minutes before the match!
Computer chess tournaments were organised. There were some amusing incidents, as in 1982 when the American program BELLE was confiscated by the U.S. State Department when it was heading to the Soviet Union to participate in a tournament. The charge was that it was a transfer of American technology to a foreign country! Anyway, such tournaments are still on, for instance the World Microcomputer Chess Championship is held every October.
In 1985, three doctoral students (Hsu, Campbell and Anantharaman) created the chess-playing program Chiptest. This would develop into Deep Thought, a program that shared first place with Grandmaster Tony Miles (58) in the 1988 U.S. Open championship and defeated the brilliant 16-year Grandmaster Judit Polgar in 1993 in a 30-minute game.
The other programme that had risen into prominence in this period was Fritz. It was especially good at fast chess. Fritz 2 (which can be played on ordinary PCs) defeated Kasparov in a 5 minute game in Cologne in 1992. Kasparov also lost to Fritz 3 the following year in a blitz tournament in Munich two years later. The programme also defeated top Grandmasters Vladimir Kramnik, Viswanathan Anand, Boris Gelfand and Nigel Short. Grandmaster Robert Huebner refused to play it! Kasparov did get his revenge in November 1995 when he beat Fritz 4 in London with a win and a draw.
At a University of Zimbabwe seminar in 1997, Professor Ingo Althofer from Germany told us of an interesting way of playing computer chess. Take two computers running chess programs and keep them informed of the current state of the game. At every juncture they each suggest a move. If they coincide the move is carried through, otherwise the decision of which of the two moves to play is made by a human (Prof Althofer himself, who is a more than decent chess player). This three-player system merits a ranking significantly better than any of the three individual players --- in fact, enough to get it into the top few world players!
Back to Deep Blue. Funding of its progress was taken over by the computer company IBM, resulting in its name (initially that of a computer in a science fiction book) was changed to Deep Blue to reflect the company colour. But why would IBM spend millions on a game-playing computer? Chess hardly bring the bucks rolling in. Nor can it be just the advertising. IBM say that the ``massively parallel, special-purpose computing like that found in Deep Blue could certainly be of great use to people if applied to finance, medicine, education, etc.'' In other words, the technology behind Deep Blue can be applied to more lucrative fields such as modelling economic markets, developing new drugs for pharmaceutical companies and uncovering hidden patterns in large databases.
Any remaining ideas about IBM's generosity were crushed when the company refused the defeated Kasparov's request for a rematch, since they felt they had learnt all they needed to know from their chess project. Not that there's anything wrong with that, as Gabriel Silberman (the head of IBM's Deep Blue project) told University students in Vancouver at a lecture delivered in mid-January 1998. He said that the media hype surrounding the project had falsely attributed ``autonomous ability" to the computer, ignoring the fact that its intelligence had to come from real people --- its designers and programmers. He was especially critical of a recent news magazine's characterization of a possible future match (against Gary Kasparov or another chess Grandmaster) as the human brain's last chance to show it was still master of machines, and emphasized that the computer was flawed and quite capable of making mistakes.
Final question: why chess? Why can't computers play other games? They do. For instance, a program called Chinook has beaten the world draughts champion. But games like this tend to be too easy for computers since there are far less possibilities than chess. Other games, like the Chinese game Go, tend to be too hard for computers. Chess offers the right amount of challenge for machines. As for me however, I think I'll stick to Noughts and Crosses.
That such relationships exist is vividly illustrated at any maths conference --- there are always chess games on the go after-hours! And we are convinced there would have been some potentially brilliant mathematicians at the recent Zimbabwe Open Chess Championships --- if we only knew who! Obviously there are close similarities in the logical complexities involved in each discipline. But, as if to show that both maths and chess involve qualities of imagination and intuition that go far beyond mere logic, there have been great mathematicians who were no good at chess, and vice-versa.
A similarly elusive affinity exists between maths and music --- more of this in our next issue. In contrast, not many people (excepting the Italian Renaissance period with its excitement over the discovery of the laws of perspective) have been talented at both maths and painting. Even Escher, the Dutch painter whose fascinating works embellish the offices of several mathematics professors around the globe, once mentioned that he never thought of mathematics while painting.