Deeper Blue is a super computer used in the Jeu of failures. This computer is the changing of another older Supercalculateur, Deep Blue. Deeper Blue faced (just like its elder previous year) the world champion of the time, Garry Kasparov at the time of a match in six parts which proceeded in New York the 3,4,6, May 7th, 10th and 11th 1997. Kasparov had to be inclined against Deeper Blue, on the score from 3.5 to 2.5 (a victory for Kasparov, two for Deeper Blue and three null).

Deeper Blue calculated 300 million blows a second. It was conceived by the firm IBM, more precisely by Feng-hsiung Hsu, at the origin of the project, and Murray Campbell; the large Masters of failures Miguel Illescas, John Fedorovich, Nick De Firmian and Joel Benjamin helped with his design, in particular on the book of openings of Deeper Blue.

Summary of the parts

Part 1 : Kasparov gained the first part, rather easily according to the opinion of all; Deeper Blue had not badly played, but had played of the weaker blows at certain times, which made it possible Kasparov to gain the part. According to certain commentators, “it swept the machine out of the chess-board”.

Part 2 : A true surprise awaited Kasparov on this part… It made in kind arrive at a position blocked, complex, of kind to push the computer to be analyzed less better (the more there are possibilities and the more the computer spends time to reflect, or reflects by pushing the analysis less). But Deeper Blue arrived at a difficult position for Kasparov; however, it was not lost for as much: the computer had certainly a small advantage, but not enough important to gain. Kasparov sacrificed a pawn then, thinking of obtaining sufficient compensations to leave themselves this difficult position and to possibly ensure the null one. But the computer did not take the pawn “offered” (Kasparov thought that it would do it) and played in place a master stroke, a strategic blow as those which played Karpov in its time, which ensured the profit of the part for a long time. Kasparov had not thought that the computer would find a blow also strong, and it lost the part some blows afterwards. This part destabilized Kasparov completely: it had gained the first part easily, and during the second the computer had played much better! Its team even came from there to suspect a cheating… It was to some extent the pivot of the match, and according to much, the best part of failures ever played by a computer.

Part 3 : Although destabilized by its lost part, Kasparov was able to concentrate and ensured null.

Part 4 : Kasparov did not manage to have a tangible advantage, and the part arrived in a null final of Towers; consequently the null one was signed after 5 hours of play.

Part 5 : A finale Turn + Riding, with a pawn passed for Kasparov, but the part was null by perpetual failure.

Part 6 : A drama. Kasparov, tired physically and psychologically, did not even want to play this sixth part. It did not lose this part because the machine was stronger than him, says one, but simply because it was tired too much and cracked: it is one of the disadvantages of the human players who to feel tiredness after nine days of competition… Kasparov gave up with the nineteenth blow, and left the room by raising the shoulders of a resigned air. Deeper Blue had gained the match.

Anecdote

• the project Deep Blue was not always called thus. Indeed, at the time of its creation, its name was Deep Thought, a reference to the book The Hitchhiker' S Guide to Galaxy Douglas Adams; called “Big Blue” with IBM, the computer then evolved/moved in Deep Blue, then in Deeper Blue.