Advertisement

P. C. CORNER

When current U.S. chess champion Patrick Wolff, of Somerville, lost a game to his opponent on Saturday, he was less than thrilled about his defeat.

But his opponent wasn't happy, or remorseful. In fact, his opponent showed no emotion at all.

That's because Wolff was playing against a computer chess player. The game was one of 36 played between six humans and six machines in the Fourth Harvard Cup "Human Versus Computer" Chess Challenge, held at the Computer Museum in Boston. The final result was 27 games to the human masters and the remaining nine to the computers.

The human participants included Wolff, who counts in his career victories a defeat of world champion Garry Kasparov of the former Soviet Union, Boris Gulko, a former Soviet Union chess champion, and Joel Benjamin, who ranks third among U.S. grandmaster.

The computer challengers were four software programs and two chess board machines, all commercially available. In fact, the organizers say they deliberately excluded research chess-playing systems because they wanted to demonstrate that even the $49.95 chess program can prove a formidable opponent.

Advertisement

"The inclusion of only commercial programs brings our event a lot closer to the expectations of spectators, who are also consumers," says Daniel H. Edelman '91, former president of the Harvard Chess Club.

"There is really not a lot of difference between these programs and big research systems," says Edelman, who organized the first Harvard Cup tournament with Chris Chabris '88 four years ago.

By consensus, the world's best chess-playing machine is Deep Blues, formerly known as Deep Thought, a special hardware-software system built exclusively to play chess by a team of young researchers at IBM.

In a widely publicized match between Deep Thought and Kasparov in 1989, the world's best human player beat the computer twice. But Deep Thought did manage to beat a number of grandmasters on various occasions, including Robert Byrne, chess columnist for The New York Times.

The chess machines featured in this year's Harvard Cup were admittedly far less powerful than Deep Blues, but they performed better than ever, winning 25 percent of all possible points. The four software packages--"Kasparov's Gambit," "BattleChess 4000 SVGA," "Socrates Exp," and "M-Chess Professional"--all ran on 60MHz Pentium-based PCs on loan from Intel Corp.

"This year we have more sophisticated software and hardware," says Edelman. "The Pentium is a great leap in hardware technology for our event."

Of course, the fastest computer can't play good chess if the software lacks "intelligence." But more processing power certainly means a better chance for the machine to prevail.

There are basically two ways for a computer to play chess. One, known as the brute force method, tries to analyze many possible moves from the current board configuration and pick the best one.

The second technique involves more human-like thinking patterns. In determining which move to make, the computer not only looks ahead but knows which scenarios provide the best chance of winning. By analyzing the board mathematically or searching its database for classic playings, the computer is able to come up with a winning strategy that may even involve setting up traps.

Humans are very good at combining both methods, and can carry out many mental calculations in parallel and at incredibly high speeds. Computers, in contrast, lack intuition and a feel for the game and are forced to rely heavily on their forte of fast searching and storing large amount of information to win.

As all the games in Harvard Cup this year restricted each player to 50 minutes to make all his moves, or lose the game automatically, the computers are handicapped by the limit on the number of possible moves they can consider. A very smart program may be good at estimating which moves are optimal or near-optimal, but most systems, like their human counterparts, play much better without time constraints.

Knowing that computers rely mostly on the brute force technique to make their moves, some human chess masters have no doubt figured out how to take advantage of the machines' naivete. But most are awed by how well the computers can play.

"They have a lot more respect for the machines than they did a decade ago," says Edelman, now a student at Columbia Business School, "and it's no longer considered a humiliation to lose to a computer."

But will computers someday beat humans in all games? Most human chess masters doubt, or at least convince themselves to doubt, that computer players will ever be as good as the best humans.

As the machines get better and better each year, however, it becomes difficult to imagine that humans will always be able to dominate the world of chess, especially when today's researchers try to build programs that can mimic the way grandmasters play--with help from the latter.

"We expect that very soon the computers will play better than the best human minds," acknowledges Edelman.

Haibin Jiu '94, associate photography chair of The Crimson, is the former president of the Harvard Computer Society. His column appears Tuesdays.

Tags

Advertisement