1978 | Not Bad – for A Machine…

David Levy against Chess 4.7
David Levy against Chess 4.7

This article by Jim Kaplan was published in “Sports Illustrated” September 18, 1978.

David Levy won his 1968 bet that no computer would beat him in the next 10 years, but Chess 4.7 didn’t roll over. When David Levy sat down in a soundproof, air-conditioned booth in Toronto to play chess against a computer 725 miles away in Arden Hills, Minn., nobody laughed. Indeed, the outcome of the match would have as much significance for the future of chess as anything Viktor Korchnoi or Anatoly Karpov might be doing across the board from each other in the Philippines.

Each time Levy, an international master from Great Britain, touched a piece, his opponent, a Control Data Corporation CYBER 176 computer using a program called Chess 4.7, sensed the vibrations. Whirring into action, the machine absorbed the new position, made its calculations and spewed back its answer on a TV screen. Small lights would flash on the squares of the playing board in front of Levy—one underneath the piece’ to be moved, another showing the square the piece would occupy.

Fine Chess Products from Around the World

Ten years ago, the idea of an international master—or even a moderately accomplished chess player—getting a serious game from a computer was considered ludicrous. In August 1968 Levy, then the Scottish champion, made what he calls a small bet with four professors interested in computer development: no computer would be able to beat him in the next decade. It seemed like a safe proposition. No opponent was even found for him until 1977, when he beat the Sovet Union’s best program, KAISSA, and Chess 4.5. Later that year, an improved version of the latter program, 4.6, won the world computer-chess championship. To win his wager, by now grown to 1,250, Levy began a six-game match late last month against Chess 4.7, an improved version of 4.6.

There was not much doubt that Levy would win his bet, but what made the match significant was how far computers had come in the 10 years. Earlier this year, 4.6 had beaten one of Korchnoi’s seconds, Michael Stean, in speed chess and toppled former U.S. champion Walter Browne in a simultaneous exhibition. (Stean cabled Levy to beware “the iron monster.”) Even so, no chess computer had beaten or so much as drawn with an international master, to say nothing of a grandmaster, in a one-on-one, regulation-time match.

But by the 12th move of the first game, history was in the making. Levy had taken an inordinately long time—almost 8 minutes—to make a predictable pawn move. During that period, 4.7, analyzing possible moves at the rate of one every 337-millionths of a second, had looked at 1,508,192 different positions. When Levy finally moved, 4.7 had an immediate response—a stunning knight sacrifice. Even more astonishing was its next play, a queen thrust that pinned two pawns and threatened checkmate in two moves.

“I was devastated,” Levy said later. “Against a grandmaster I would have resigned immediately.” Now it was time for some peculiarly human savvy. Drawing on his knowledge of computer “psychology,” Levy pressed ahead. He knew that computers are strong on basic tactics but tend to overemphasize material gain at the expense of less tangible positional advantages. Sure enough, 4.7 retreated to defend its pawn structure and, in doing so, weakened a winning position.

David Levy vs Chess 4.8., Hamburg 1979
David Levy vs Chess 4.8., Hamburg 1979

At this juncture, however, 4.7 suddenly signaled its human masters that it wasn’t feeling well, that something was wrong inside. “There are two million transistors in the computer,” explained a man from Control Data. “If one goes bad, the machine doesn’t work.”

The machine was fixed just in time to prevent it from going over the time limit, and the match ended in a mutually satisfactory draw. “I’m just happy I didn’t lose,” said Levy. The computer people were happy that they had produced a history-making draw.

Playing the white pieces, 4.7 made an unusual opening in the second game. Levy promptly transposed into the Dragon Sicilian Defense, of which he is the world’s leading author. He then prevented the machine from castling and had it checkmated in 54 moves; as early as Move 23, 4.7 was sounding like a loser. “Oh, you had that,” the screen said after one Levy move. Ahead by a queen and a rook, Levy began joking. “A gentleman would offer to resign,” he said. “My opponent is no gentleman.” Meanwhile he was moving somewhat carelessly toward mate. The irony was that the machine, which is superb in mathematical end games with few pieces on the board, could have beaten itself faster than Levy did.

In Game 4 over the Labor Day weekend Levy decided to play the machine on its own terms—and got stung. He chose a risky Latvian Gambit, stuck with it, and lost in 56 moves. For the first time in hand-to-transistor combat, a computer had beaten an international master.

In the fifth game Levy went back to his normal style and closed out to win the bet. At 33 he is an anomaly—a player of less than grandmaster status who can make both a living and a reputation from the game. Trained in math, statistics and physics at St. Andrews University, Levy has written three books on computer chess, owns two publishing houses and is working on a business deal that, he asserts, will be a “tremendous worldwide success.” Levy is an engaging raconteur with a dry sense of humor, whose wide range of interests includes both music and poker. “I enjoy playing poker more than chess,” he confesses. “It has as much content and more psychology.”

But computer chess remains his entree to celebrity. “In 10 years computers will beat grandmasters,” he says. Others disagree, pointing out that Bobby Fischer has recently annihilated one of the world’s best, the MIT Greenblatt program. These critics assert that chess is too creative to be conquered by machines. Still, chess-playing computers are here to stay. Small models now on the market can be set to play beginners or intermediates up to the 1300 level (an expert is rated at 2000 or above). Another machine teaches end games. So far, humans have withheld part of the game from the computer, the act of moving the pieces. David Slate, one of 4.7’s programmers, could change all that. He is working on the ultimate chess computer—a human-sized robot that will move the pieces itself. “Now that,” says Levy, “would be unnerving.”

This article by Jim Kaplan was published in “Sports Illustrated” September 18, 1978.

Fine Chess Products from Around the World

Share
This entry was posted in Collectors Items. Bookmark the permalink.

Comments are closed.

Fine Chess Products from Around the World