No. 3005: DID KASPAROV MEET A SUPERIOR INTELLIGENCE?
by Andy Boyd
Today, the wrong move. The University of Houston presents this series about the machines that make our civilization run, and the people whose ingenuity created them.
The year 1997 saw a breakthrough unlike any other in the history of computing. It was the year that world chess champion Garry Kasparov lost a match to IBM’s supercomputer Deep Blue. But why the loss? Was Kasparov outmatched, or was Deep Blue lucky?
Chess master Garry Kasparov. Photo Credit: Wikimedia Commons
Kasparov was no stranger to chess-playing computers. Only a year earlier he’d defeated Deep Blue. But even more, Kasparov understood how computers played. Humans play chess by evaluating how a handful of moves might play out into the future. Computers do much the same thing, but can evaluate millions of moves and trace their repercussions farther into the future. Still, even with all that computing power, computers need guidance about what makes a move “good.”
The computer Deep Blue. Photo Credit: Wikimedia Commons/Jim Gardner
That’s where people behind the scenes came in. They developed algorithms that allowed Deep Blue to calibrate itself by studying hundreds of thousands of games played by the finest chess players in history. In addition, grandmaster chess players aided programmers by sharing their experience and insights about the game. Deep Blue played with the knowledge of grandmasters and the speed of a powerful computer.
So when Kasparov sat down to play chess, he knew his opponent was good — perhaps superhumanly so. And near the end of the first game, Kasparov was witness to an extraordinary move; a move so outlandish the world champion was dumbstruck. By all human standards the move was nothing short of crazy. Deep Blue went on to lose the game shortly thereafter.
A diagram of a chess board. Photo Credit: Wikimedia Commons
The move appeared so crazy Kasparov spent the evening fussing over it. What was Deep Blue thinking? Perhaps the computer realized the game was lost and threw it to confuse the reigning champion. Or perhaps Kasparov was being set up, lured into a sense of overconfidence. Ultimately, with the help of his own chess-playing computer, Kasparov reasoned that Deep Blue was seeking to tempt him into a long, losing sequence of moves. What appeared an inexcusable blunder was a sign of just how deeply Deep Blue was thinking. The machine was performing at a level Kasparov couldn’t begin to comprehend.
The game was the last Kasparov would ever win against his mechanical nemesis. Deep Blue would win two of the remaining matches with three played to a draw. Author Nate Silver has argued that the computer’s unorthodox move in game one badly shook Kasparov’s confidence, convincing the grandmaster that “the counterintuitive play must be a sign of superior intelligence.”
Deep Blue’s move clearly had an impact on Kasparov, though exactly how it affected his play will never be fully understood. But one thing we do know. Following Deep Blue’s victory, its designers divulged the logic behind the computer’s unconventional move. It seems … Deep Blue had a bug.
I’m Andy Boyd at the University of Houston, where we’re interested in the way inventive minds work.