The History of Computer Chess – Part 5 – Levy's bet...

The History of Computer Chess – Part 5 – Levy's bet...

GM Ginger_GM

We're half-way through my series on moments in the history of computer chess! So far, we've chartered the course of computers and chess from the automata of the 18th century to the early computing developments by chess-fan teams at IBM and Los Alomos in the 1950s and 60s. It was in the late 60s and throughout the 70s that computer chess gets good! 

In 1968 at a conference on Artificial Intelligence at the Department of Machine Intelligence and Perception at Edinburgh University, International Master David Levy was in attendance. Levy is a chess writer and founder of the International Computer Games Association, an organisation that organises various tournaments and competitions for chess engines.

It was at this conference in 1968 that Levy met John McCarthy, who we were introduced to earlier in this series. McCarthy had earlier attended the famous Dartmouth Conference in 1956, having coined the term 'artificial intelligence' in 1965. Now at Stanford University, he was a Professor of Artificial Intelligence, a university he would stay at for some 40 years. A world's leading authority in AI, McCarthy was in Edinburgh ahead of publishing his paper on broad philosophical problems in AI.

One evening during this conference there was a cocktail party, and it was there that Levy, then the reigning Scottish chess Champion, and McCarthy, a known chess fan, agreed to play a friendly game against one another. After Levy won the game, McCarthy remarked that although he himself was not yet strong enough to beat an IM, he believed that within ten years time there would be a computer chess program that could defeat a master. Levy, then the reigning Scottish Chess Champion, was taken aback! How could this amateur player, as smart as he was, make such a bold claim that a computer would beat a master within years, not even decades.

Politely, Levy told McCarthy to 'put your money where your mouth is', and so the bet was born. With £500 (then worth around £1250 in today's money) at stake, Levy would front his own share if proven wrong, with McCarthy joining ranks with host Professor Donald Michie to split their £500 stake.  “You know, I was earning £895 a year,” remembers Levy, “But I was so confident.”

IM David Levy

There were three challenge matches in the next decade until the conclusion of the bet, each two-game match being won by Levy, who would win the first game, after which the computer's teams would default the second. "Until 1977, there seemed to be no point in my playing a formal challenge match against any chess program because none of them were good enough," wrote Levy, "but when CHESS 4.5 began doing well it was time for me to defend the human race against the coming invasion."

The fourth and final challenge came in the decade anniversary of Levy's original bet with McCarthy, when in late August of 1978 Levy travelled to the Canadian National Exhibition in Toronto, Canada. There, Levy was to play a six-game match against Chess 4.7, the descendant of Chess 4.5 that he had taken note of in the year previous. The exhibition match drew an audience, who watched silently as Levy, adorned in tuxedo, and Chess 4.7, housed safely in a glass booth, drew their first game.

Levy then bounced back in the second and third games, scoring a win, and only needed a draw in the remaining three games to secure victory, and more importantly the stake of his £500 bet. But round four saw a devastating loss for Levy after he gambled on changing his style to challenge Chess 4.7 and take it out of the book; instead of his usual closed, quiet games, Levy experimented with the Latvian Gambit, and after a series of minor mistakes in the middle game, and with Chess 4.7 looking at the position and moves some nine or ten plies deep, he crashed out of a rook vs bishop endgame.

Now Levy had to draw at least one of the final two games, and after winning the 5th game, Levy had won the match the computer's team defaulted the final round in defeat. A somewhat unexciting finale and conclusion to the bet! Six months later, Chess 4.7 would make another appearance at 9th annual North American Computer Chess Championship, crashing out after a loss to a newer computer called BELLE, who convincingly won the championship undefeated.

Levy though continued to play exhibition matches against computers well into the 80s, and for 21 years after his original bet with McCarthy, he remained undefeated until 1989 when he lost to Deep Thought at a match in London hosted by the British Computer Society. It was then less than ten years until 1997, when IBM's Deep Blue would famously become the first engine to beat a reigning World Champion in a tournament – a story we will return to in the next edition of my series on moments in computer chess history!