Checkmate in 37 moves: Deep Blue's silicon victory over Kasparov

In 1996, IBM's Deep Blue computer defeated chess world champion Garry Kasparov in 37 moves. The victory marked a turning point for humans and machines.

listen Print view
Lead image Checkmate

(Image: heise medien)

5 min. read
By
  • Detlef Borchers
Contents

On February 10, 1996, the then chess world champion Garry Kasparov lost the first chess game against the IBM computer Deep Blue. The “Man vs. Machine” competition, held under tournament conditions in Philadelphia and watched by 3,000 spectators on-site and millions online, caused a sensation.

Although Kasparov still won the match series 4:2 (one loss, two draws, three wins), the rematch in 1997, where Kasparov grandly announced he would defend the “honor of humanity,” was won by the machine 3.5:2.5. It was the worst defeat of his illustrious career.

Zahlen, bitte!
Bitte Zahlen

In dieser Rubrik stellen wir immer dienstags verblĂĽffende, beeindruckende, informative und witzige Zahlen aus den Bereichen IT, Wissenschaft, Kunst, Wirtschaft, Politik und natĂĽrlich der Mathematik vor.

To commemorate the historic defeat, Deutsche Bundespost issued a postage stamp in 2021 a postage stamp that was intended to comfort passionate chess players who were defeated by their smartphones. After all, apps with access to databases have long since reached a playing strength that surpasses that of Deep Blue, not to mention the self-learning chess intelligence Alpha Zero.

Videos by heise

In 1996, it looked different: experts believed that no computer could beat the reigning world champion before the turn of the century. Kasparov himself had already accepted a simultaneous match against 32 chess computers in June 1985, which he won 32:0.

Garry Kimovich Kasparov (*April 13, 1963 in Baku, Azerbaijan), photographed here in 2023, is considered one of the strongest players in chess history. His defeats against Deep Blue demonstrated how powerful computer technology had become.

(Image: CC BY 4.0, European Union)

In the same year, computer science student Feng-hsiung Hsu began developing a chess computer named ChipTest at Carnegie Mellon University. In 1987, it won the championship of chess computers. Hsu persuaded computer scientist Murray Campbell to join his idea of building a chess computer that could beat a grandmaster. When the two developers moved to IBM Research in 1989, the name of their project also changed.

Initially, the chess computer was called Deep Thought, before it was renamed Deep Blue after IBM's nickname “Big Blue”. At IBM, programmers Arthur Joseph Hoane and Jerry Brody joined the project to develop a chess computer based on the IBM RS/6000 with its parallel processors. They enlisted the American grandmaster Joel Benjamin, author of Unorthodox Openings, as an expert for the opening game.

The prototype of Deep Blue was completed in 1995 and could calculate two and a half million positions per second. Work began to persuade the eccentric world champion Garry Kasparov to compete – for example, with a print advertisement for the chess match in which the chess computer allegedly made a Spock-like declaration of love: “I am a powerful supercomputer, Mr. Kasparov. I do not have emotions. Ah, what the heck, I love you man.”

IBM's Deep Blue.

(Image: CC BY 2.0, Christina Xu)

Kasparov accepted and began developing ten new opening variations at the time, as he had to assume that all his previous openings were stored in Big Blue's memory. He used Chessbase Fritz 4 as his training partner. For the first game in Philadelphia in front of a large audience, Deep Blue, connected from New York, played with white pieces.

After a cautious start by the computer, Kasparov thought he had the game under control until Deep Blue's 23rd move unsettled him, and he suspected human intervention. He then switched to attack, but Deep Blue parried skillfully and had reached a position in the 37th move where Black was hopelessly trapped. Kasparov reportedly spent a sleepless night pondering his mistakes.

The game Deep Blue (white pieces) against Garry Kasparov (black pieces) from 1996 in a GIF animation. The IBM computer needed 37 moves to win.

(Image: CC0. Morn)

After winning the first match in February 1996, Kasparov accepted a rematch for which Deep Blue received an immensely extensive “update” with doubled speed and quadrupled VLSI-based “chess chips.”

The “update” did not improve things for Kasparov: he turned the so-called “friendly match” of 1996 into a match of the century in 1997, in which he wanted to defend the “honor of humanity.” Vladimir Kramnik, his student and successor as world chess champion, scoffed that humanity consisted of more people than just Kasparov.

In the rematch games against Deep Blue, Kasparov caused great displeasure by repeatedly accusing the IBM team of manual intervention in the game. At a press conference, he spoke of a “Maradona moment of chess history”, alluding to Diego Maradona's unpunished handball goal, which initiated Argentina's victory over England in the 1986 World Cup final.

Only with the perspective of many years did his view of the games with Deep Blue change, as he reflected in his book Deep Thinking. Kasparov, who meanwhile made a name for himself as a critic of artificial intelligence, wrote in his book in 2017: “Anyone who wants to put themselves in my shoes 20 years ago just needs to get into an autonomous car.”

(dahe)

Don't miss any news – follow us on Facebook, LinkedIn or Mastodon.

This article was originally published in German. It was translated with technical assistance and editorially reviewed before publication.