New York, May 11, 1997.
World chess champion Garry Kasparov stood up from the board. It was the sixth game, and only the nineteenth move.
There was no person sitting across from him. In that seat was IBM’s computer, Deep Blue. It was the first time a reigning world champion had lost a match to a machine under standard tournament conditions.
Act 1. A graduate student’s machine
The story begins with a graduate student at Carnegie Mellon University. As a doctoral student, Feng-hsiung Hsu built a chess computer called ChipTest, which was followed in 1988 by Deep Thought. After receiving his doctorate in 1989, he and his colleague Murray Campbell moved to IBM Research to carry on the work.
Deep Thought lost to Kasparov in 1989. IBM held an internal contest to give the machine a new name. The winner was a play on IBM’s nickname, Big Blue: Deep Blue.
Act 2. Philadelphia, 1996
Their first match was held in Philadelphia in February 1996, over six games.
Deep Blue won the first game. It was the first time a computer had beaten a reigning world champion in a game under normal tournament conditions. The world was stunned.
But Kasparov quickly worked the machine out. Over the remaining five games he won three and drew two, taking the match 4–2. The human had won.
IBM did not back down. It agreed to a rematch a year later and rebuilt the machine.
Act 3. New York, 1997
The Deep Blue that returned was a different machine. With 30 processors and 480 custom chess chips, it could evaluate 200 million positions per second, twice as many as the year before.
May 3, game one. Kasparov won. So far it looked like the year before.
But near the end of that game, Deep Blue played one strange move. It was move 44, and it seemed to have no purpose. It was later reported to be the result of a bug: unable to settle on a move, the machine had fallen back on a fail-safe.
Kasparov did not see it that way. In that move he saw a depth he could not understand.
Act 4. What broke was not calculation
Deep Blue won game two. Kasparov later said he had seen “unusual creativity” in the machine’s moves. He did not play the game out to the end. He resigned.
Only afterwards did something else come to light. The position he had given up was one he could have drawn. There was a way to force a draw by perpetual check. The world champion had not seen it.
Kasparov suspected that a human was behind the machine’s moves, that some grandmaster had chosen them. IBM denied it, saying the only human intervention took place between games.
Games three, four and five were all drawn. The score stood at 2½–2½. Everything came down to the last game.
Act 5. Nineteen moves
May 11, game six. Kasparov chose the Caro-Kann Defence.
Early in the game, Deep Blue threw away a knight. It gave up the piece to drag the king into the open. Kasparov’s position collapsed almost at once.
He resigned after the nineteenth move. The world champion had lost in fewer than twenty.
Final score: 3½–2½. Deep Blue had won.
Epilogue. The winner that disappeared
Kasparov demanded a rematch. IBM refused. Deep Blue was dismantled after its victory. IBM later published the machine’s log files.
The two racks that made up Deep Blue are now in museums: one at the National Museum of American History, the other at the Computer History Museum.
People talked about this match for a long time as the day machines became better at chess than humans. That is true. But the record of those six games shows something a little more complicated. At the turning points of the match there was a move produced by a bug, a man who read deep meaning into it, and the moment he gave up a game he could have drawn.
Nineteen years later, another match would be held in Seoul. That time the game was Go, and that time the human took one game.

