Thursday, April 21, 2005


Game Over: Kasparov and the Machine

I saw this at the Alamo downtown tonight. In a nutshell, it covers the 1997 media frenzy rematch between Garry Kasparov and Deep Blue and the largely ignored allegations of cheating by IBM. I was never interested in these matches since there is a finite (albeit astronomical) number of plays that can be made in chess, so theoretically a computer can be built that could never lose. However, I am interested in security design, scientific integrity and stock market manipulation. Since this incident raises all three issues, I'm interested.

Security Design
It would be easy for IBM to cheat. There are these things called computer networks. They allow computers to be controlled at a distance. At the press interview after the pivotal second game loss, Kasparov made reference to "The Hand of God" interfering with the game. In security speak, IBM was the trusted authority since they were solely responsible for making sure the computer didn't cheat. No one else was allowed to audit the system. Another way of describing a trusted authority is that it is the person or entity that you give permission to violate your security.

Scientific Integrity
This issue is simple but also the definitive issue and curiously the most neglected one. Kasparov beat Deep Blue in 1996 and had no problem accepting a rematch. Scientific method requires that claims are repeatable. So does the sports world. Just because the Red Sox finally win a World Series doesn't mean that every year henceforward they can say they are the best baseball team without playing. But that is exactly what IBM has done. But even more ludicrous is that they refused to share the logs to show that the computer actually did the work in the second game (Anyone who has ever taken a math or computer programming class knows that they have to show their work to get credit for their answer. Otherwise, it is assumed they received outside help). They are supressing the evidence of the fix! And what's amazing is the sheer audacity of the smug tool of a guy from IBM who won't shut up about his responsibility for the victory. He says that it is petty for Kasparov to raise the possibility of cheating when he has no evidence! Well how about coughing up those logs? Otherwise, you have zero scientific integrity. And this further demonstrates why IBM would not accept a rematch. After an allegation of cheating, a rematch would require that IBM would no longer be the sole trusted authority.

Stock Market Manipulation
Of course, this was a naked PR stunt, but a major one. In 1997, in the midst of the Internet boom, IBM needed its image revitalized. As a result of the spectacle, IBM shareholders did well (because their horse won J.M Keynes so-called beauty contest). On more than one occasion, the film used anonymous whispers to suggest that the stock may have shot up 15% as a result. However, this isn't true unless you count preceding weeks and assume foreknowledge contributed to this earlier part of the stock's rise. I'm skeptical about the film's anonymous 15% suggestion.

People often find it tough to swallow the possibility of foul play here. First, it would require a complicit media. That just seems unimaginable! Second, Pete Rose was barred from baseball for betting for his own team. To suggest IBM was after monetary gain would be just as unsportsmanlike!

IBM promoted the match as "Man vs. Computer" and just about all of us bought the line. In restrospect, it is easy to see the match was really "Man vs. Corporation". Man didn't lose to a computer. He lost to a corporation.

Greetings, Professor Falken... I became disillusioned with competitive sports long ago.

Perhaps with the exception of Lance Armstrong and his perpetual yellow jersey status. He wins even when the French "fans" throw stuff at him.
Post a Comment

<< Home

This page is powered by Blogger. Isn't yours?