The Machine at the Table
It ended in nineteen moves. The most famous short game in modern chess — one hour on the clock, a knight sacrificed on move eight, and the world champion pressing resign.
Two Heavyweights, One Rematch
May 11, 1997, New York's Equitable Center. Garry Kasparov, world champion since 1985 and 34 years old at the time, had never in his life lost a multigame match to a single human opponent. He was about to lose one to a machine.
The machine was Deep Blue, IBM's chess supercomputer: 32 parallel processors running 480 custom chess chips, heavy enough to weigh 1.4 tons and fast enough to survey about 200 million board positions every second, with a memory bank of 700,000 grandmaster games. Across the table from the greatest human player alive sat silicon built by a team led by designer Feng-hsiung Hsu and helped along by grandmaster Joel Benjamin. It felt no nerves, and it never tired.
This was a rematch. A year earlier, in Philadelphia, the same contest had ended 4–2 in Kasparov's favor — though not before Deep Blue won the very first game, the first time a computer had ever beaten a reigning world champion under standard time controls. Kasparov had slowed it down there. This time, that plan would fail.
The 1997 rematch ran May 3–11. Kasparov won game one. Deep Blue struck back in game two. Games three, four, and five all finished drawn, and the world's press — which treated the whole affair as a symbolic test of whether machines were catching up to human intelligence — watched the score sit even at 2½–2½ going into the final game.
Nineteen Moves
Then the machine stopped being patient.
Playing White, Deep Blue reached for the wild, forcing chess that human strategists usually manage to avoid. Kasparov, a Sicilian player against humans, instead chose the solid Caro–Kann — a move he would later judge a dubious opening, picked to drag the machine out of its depth. It went badly. The computer sacrificed a knight on move eight, tore apart Black's pawn shield, and forced the black king out into the open. Kasparov's pieces tangled, his queenside collapsed, and on move 19, facing the loss of his queen for nothing more than a rook and a bishop, he resigned. The game had lasted barely more than an hour. "I lost my fighting spirit," he said afterward.
Final score, 3½–2½ — Deep Blue. It was the first time a reigning world champion had fallen to a computer over a full match under tournament conditions.
The Argument That Never Ended
The fallout lasted far longer than the chess. Kasparov claimed he had seen "superior intelligence" that bordered on the human and suggested someone had helped the machine between moves — charges IBM denied and no one ever proved. IBM retired Deep Blue, declined Kasparov's demand for a third match, and later put parts of the machine on display at the Computer History Museum.
Was Deep Blue smarter than Kasparov? Almost certainly not. It was faster. It worked through carefully chosen evaluation rules, searched without fear, and looked at roughly 200 million positions every second. On that one New York afternoon, it was just enough — and it changed the story of chess forever. The royal game would never belong to human beings, and human beings alone, again.