![]() |
![]() |
|
Man vs. Machine: Deep Junior Makes The Fight Worth It j000000020030210dz2a0000q By Garry Kasparov 1024 Words 10 February 2003 The Wall Street Journal A14 English (Copyright (c) 2003, Dow Jones & Company, Inc.) |
Years before the first computer was actually built the famous British mathematician Alan Turing envisaged it playing chess and beating the human world champion. When the very first computers were delivered to the Los Alamos Scientific Laboratory in the 1950s the scientists there immediately started to program it to play chess. The quest had begun. For 50 years one of the recurring motifs of the science of artificial intelligence was a chess computer beating the world champion. It was almost an obsession in the community, with scores of teams in dozens of universities dedicating their academic careers to achieve this goal. The discussion never ceased over the decades: When would this momentous day come? |
It came on May 11, 1997, at 3:30 p.m., when IBM's multimillion-dollar machine, Deep Blue, narrowly defeated me in a six-game match. One of the greatest milestones of artificial intelligence had at long last been achieved. --- But what did IBM Research do in this situation? Instead of turning the machine over for study, instead of conducting follow-up experiments (Deep Blue could have played against other grandmasters, in tournaments, against other chess programs), instead of spawning dozens of doctoral theses, IBM dismantled the machine and terminated the experiment. It was as if they had sent a man to the moon and brought him back safely, but had done no scientific research in the process. "Take our word for it, we were there. Here's a photo," they said. No rock samples, just the PR value of having achieved the goal. Not only did they never go back, but they made it seem like there was no point in anyone ever going back. This past week in New York City, the man-machine experiment was relaunched and redefined when I played a six-game match against the world champion computer program Deep Junior. It was a tough battle and even though I was much better prepared than I was for Deep Blue in 1997, the match finished in a 3-3 tie. The main difference, however, was that thanks to the sponsors, organizers, and participants, we brought computer chess back to its scientific origins. Until now, competitions of Man vs. Machine have been viewed as something mythological because we knew so little about the machine. What the machine did was promoted heavily while how it did it was shrouded in secrecy. In 1997, when IBM terminated the Deep Blue project, the computer-chess field was left to talented enthusiasts. In the past five and a half years, new programming techniques have combined with superior chess knowledge to create a quality of machine chess that is far superior to that of Deep Blue. In game five of my match with Deep Junior it played an imaginative sacrifice of the type generally considered impossible for a computer player. It was a landmark moment for computer chess and the science and programmers behind it. Unfortunately, because Deep Blue's records were never made public it is quite useless to discuss the strength of Deep Blue vis-a-vis the strength of Deep Junior. IBM said Deep Blue crushed other programs, but what value are these statements without records? Yes, Deep Blue was 100 times faster, but so what? Sheer power means little in chess because it is a mathematically near-infinite game. The only way to measure the strength of a chess-playing computer is to analyze its moves. While putting Deep Blue's six games to the test with current top programs -- Deep Junior and Deep Fritz -- we discovered that they consistently play better than Deep Blue. The only exception is when Deep Blue showed a stroke of genius in one game (when I suspected certain interference). Today, after I play a match with Deep Junior we can reconstruct the computer's decision-making process with our own copies of the program. Moreover, due to Deep Junior's public track record of nearly a decade, I can analyze hundreds of its games. All this means we are now in scientific territory. My match with Deep Junior is the beginning of a new era in computer chess. Both sides were under strict supervision and every aspect of each game was recorded. As a result, we can see the strengths and shortcomings of the machines, and of ourselves. This marks an important shift in the history of computer chess. We are now moving away from corporate interests -- i.e., Man vs. Machine matches as PR for companies like IBM -- and toward fair and documented procedures overseen by the World Chess Federation. X3D Technologies, organizer and co-sponsor of this new match, had the courage to continue the long tradition of chess and science. They displayed the entire match live with their three-dimensional image technology, and all of the games and match details are stored at their Web site, X3Dworld.com. They were rewarded with tremendous exposure and have already talked about continuing their involvement. --- What makes this new era so exciting is that there are many programs using different techniques that produce distinct styles. Deep Junior is as different from Deep Fritz as Kasparov is from Karpov. Chess offers the unique opportunity to match human brains and machines. We cannot do this with mathematics or literature; chess is a fascinating cognitive crossroads. Despite the increasing popularity of the game, chess has long suffered from a lack of sponsorship. IBM exacerbated this situation by creating the impression that the experiment begun by Turing was over, that there was no more to be learned. I believe that this new match has convincingly refuted that argument. There was record interest in this match on its own merits, without a giant PR machine. ESPN provided live coverage, a first in the history of chess. The match achieved the sort of success that will encourage sponsorship money to flow back into this game and this great experiment. For this X3D Technologies and Deep Junior should be commended. --- Mr. Kasparov, the world's top-ranked chess player, is a Journal contributing editor. |
For assistance, contact Factiva Customer Service by visiting Factiva Customer Service
and selecting Ask A Question via e-mail or CallMe
(c) 2003 Dow Jones Reuters Business Interactive, LLC. Trading as Factiva. All Rights Reserved . |