infinite improbability

You know you're a geek when you go to see a documentary about chess (aka. a review of said chess documentary)

The Sarasota Film Festival (motto: we're not Sundance or Toronto, but wait ten years) started this week, which is kind of cool because it offers the chance to see films that the average person probably wouldn't normally get too see. Considering that the first full fleged festival was only in 2000, it's not on the level of any of the really prestigous film festivals yet (though they are trying to get there), although it has increased it's stature a lot in the last four years--Jennifer Love Hewitt's attempt at being seen as a serious actress premiered here yesterday. Anyway, enough of that, the point of this blog entry is to write about the documentary I saw this morning, Game Over: Kasparov and the Machine about Garry Kasparov's 1997 chess match against IBM's Deep Blue.

On a technical note, I don't know whether it was intentional or not, but the camera man couldn't seem to keep the camera in focus, and the lighting was bad for a lot of the interviews, which was kind of distracting. I was a bit worried when the trailer for the film, which was shown first, said that it was from the producers of the docu-fiction known as Bowling for Columbine, and as I get into below, it was rather more biased than a documentary should be.

What Game Over: Kasparov and the Machine does best is to delve into Garry Kasparov's psyche during the 1997 competition against IBM's Deep Blue. You see him becoming more and more paranoid, and increasingly unravelled, all because in the second game, Deep Blue made a move that seemed too human for his preconceived notion of chess computers. Kasparov thought then, and still does, that IBM cheated.

Game Over tries to seem unbiased, but it is clear that the director thinks that IBM cheated. However, they give no real evidence to support the cheating claim, only intimations that IBM's security surrounding the computer room was because IBM really had grandmasters hidden in there overriding the computer on certain key occasions, and Kasparov's assertion that the computer didn't play like a computer usually does at one point in game two. In game two, Kasparov played a game that was designed to trick the computer, attempting to sacrifice a pawn in a situation where previous computer chess programs would have taken the pawn, leading to the computer's eventual loss. Deep Blue didn't take the bait, and Kasparov was so rattled because the computer seemed to play like a human that he didn't even see that he could have played Deep Blue to a draw and ended up resigning. That game psyched him out so much that he was unable to recover, and after playing games 3,4, and 5 to draws, lost game 6 horribly.

The question of whether IBM cheated all comes down to that single move in game two, where the Deep Blue made the move that any human would make but that had, up to that point, tripped up computers. Joel Benjamin, a chess grandmaster on IBM's programming team explained in the documentary that they knew that chess computers always got tripped up in that situation, and consequently spent a lot of time and effort programming Deep Blue so that it wouldn't make the mistake that other computers do. If you believe Benjamin's assertion, then the case is clear, IBM did not cheat. Unfortunately, the director quickly moved on and never mentioned IBM's explanation for the rest of the movie, preferring to cut between shots of the chess playing hoax of the 19th century, The Turk, and shots of Deep Blue, hinting that Deep Blue was really controlled by a human as well. As someone who has an understanding of programming, the explanation by IBM makes perfect sense--if you knew what you were doing, it would not be terribly difficult to put something in the code so that, if thus and so conditions are reached, then do thus and so--in other words, tell the computer what to do if a situation like the one that Kasparov created in game 2 ever happened. This isn't cheating, it's doing a good job of programming a chess computer.

In the end, it's eminently clear that the director thinks that IBM cheated, and the repeated comments about IBM's stock rising 15% the day that Deep Blue won suggest the idea that IBM cheated to pump its stock price (Kasparov even compares IBM and Deep Blue to Enron). However, there is plenty of outside opinion, within both the chess and computer science communities, that Deep Blue won fair and square and that Kasparov lost because he simply couldn't get past his view of computers as "dumb machines" and got psyched out by a machine that didn't seem so dumb after all. I just wish that the director had let us see the alternative opinion.