Online NewsHour: Checkmate

CHARLAYNE HUNTER-GAULT: Yesterday was day four in a landmark six-game match at the Pennsylvania Convention Center in Philadelphia.

CHARLAYNE HUNTER-GAULT: In one corner, Garry Kasparov, reigning world champion. In the other, Deep Blue, an IBM computer and reputedly the world's strongest chess machine. At day's end, the match was dead even. This six-game match is the first of its kind pitting man against machine in standard regulation conditions, i.e., a series of games in which each player has two hours to make his or its first 40 moves.

CHARLAYNE HUNTER-GAULT: Kasparov was favored going into the match because the time of a regulation game allows a human to use his imagination and creativity, qualities Kasparov has shown in abundance over the years. The 32-year-old Russian citizen became the youngest world chess champion in history in 1985 at age 22. Kasparov has lost an occasional ten or fifty-minute speed game to a computer but never the longer regulation game, i.e. until game one on Saturday, which Big Blue won after 37 moves. Kasparov recovered and won game two. Games three and four have both ended in a draw. The final two games will be played tomorrow and Saturday.

CHARLAYNE HUNTER-GAULT: For more on the match and its meaning, we're joined now from Philadelphia by C.J. Tan, the IBM scientist who oversaw the development of the Deep Blue computer, and by Yasser Seirawan, a chess grandmaster and publisher of "Inside Chess." And, Mr. Seirawan, what is going on in the chess match today?

YASSER SEIRAWAN, Grandmaster: (Philadelphia) Well, we have, of course, a tied match, two to two, two games left, $400,000 up for the first prize, $100,000 hopefully to the loser. And we have a very tense situation, and we hope that Garry will pull it off.

CHARLAYNE HUNTER-GAULT: Just briefly describe for me how the match is being played.

MR. SEIRAWAN: Well, of course, it's a classical chess match, where the games last about six hours. The game that Garry won lasted over 73 moves, whereas the game he lost lasted 37 moves. And it's been just a very tense, dramatic--

CHARLAYNE HUNTER-GAULT: Is he sitting there, looking at his opponent, or--I mean, where is Deep Blue?

MR. SEIRAWAN: Well, that's a good question. He said, Garry said that "My opponent is invisible but not invincible," and of course, the Deep Blue is back in Yorktown in IBM labs.

CHARLAYNE HUNTER-GAULT: Mr. Tan, briefly tell me in the simplest terms, what is Deep Blue and how it's working.

C. J. TAN, IBM: (Philadelphia) Deep Blue is basically a supercomputer that we have at Yorktown Heights IBM research laboratory, and on this supercomputer, we have installed some silicon chips that will help the computer perform chess problems very efficiently and that will help us calculate millions of billions of chess positions per second. With that, we were able to achieve the results we have obtained today.

CHARLAYNE HUNTER-GAULT: But how exactly is it working? I mean, is the computer thinking as well as Mr. Kasparov is thinking, I mean in the same way?

MR. TAN: No, a computer doesn't think the way Kasparov is thinking. Kasparov is using human ingenuity, creativity, and his knowledge and experience to perform his tasks. You can, in a way, you can say he's using the best computer that there is in his head, but that computer is different from the one we have installed at Yorktown, which is basically a machine that does calculations, and based on its power and speed, we are able to look ahead as many moves as possible to extend so that we can make a very intelligent decision.

MR. TAN: Well, there are scientists behind the machine, and it's really a group of scientists using the machine as a tool to play chess against Garry with the best computer in his head.

CHARLAYNE HUNTER-GAULT: Mr. Seirawan, do you see advantages that either one of these, the computer has over Mr. Kasparov, and vice versa?

MR. SEIRAWAN: Yes, indeed. Garry, of course, just as C.J. mentioned, is using his experience, his intuition, and his judgment, also he's playing a different kind of game against the computer than he plays against for example a human being like myself. The computer has a deep search. It has tremendous calculative abilities, so in positions where there is a string or a series of forced moves, the computer has a very large advantage over the human being, however, the human being has a great deal of advantages over the computer, and I must say that prior to the match, I felt that Garry discussion

was a very big favorite. I thought he was going to win the match handily, and just a quick story to that, of course, the ACM, who is the sponsor of the match, put up $1/2 million for the match purse, and Garry said, okay, I'll play but I want a winner-take-all, so of course, before the match, he was absolutely confident of victory, and now he's really shocked that he lost the first game. I think that was a tremendous victory for IBM and its team, shocking to everybody.

CHARLAYNE HUNTER-GAULT: How is that, Mr. Tan, because Kasparov trounced Deep Blue's predecessor, Deep Thought, in 1989? What's happened since then?

MR. TAN: Well, the computer technology has advanced so much in the last few years the computer we are using today is a thousand times more powerful than the machine we used in 1989 when we played Garry Kasparov with a Deep Thought program. And basically, we are using technology we call parallel processing, and where we have 32 computer processors working together in one system to solve this complex problem, and this--

CHARLAYNE HUNTER-GAULT: And you just put any number of combinations and things--that's what I'm still trying to understand--how the computer chooses what move to make next.

MR. TAN: Well, the computer basically--in terms of chess, it, it really, what Yasser Seirawan just mentioned, the look at all possible moves as far as you can. For instance, I would say I want to make this move, and then I would guess what my opponent will be moving to counter my move, and based on his counter-move, I make another prediction how we should go about, and I do this as far as I can with all possible alternatives.

CHARLAYNE HUNTER-GAULT: Mr. Seirawan, if the computer can see the end game or however you describe it, I mean, you know, this far down the line, how could a human being possibly compete?

MR. SEIRAWAN: Well, in fact, that's the exact point. You hit the nail squarely on the head. Garry Kasparov has never beaten a computer that hasn't anticipated his every move. The difference is although the computer can anticipate or even predict Garry's moves, the computer cannot understand or value, make a correct judgment as to the strength of Garry's moves, so in other words, even today, Deep Blue sees every one of Garry's moves, and yet, Garry can still beat the computer. To make a comparison, a discussion

very terrible one, if you had a child who went on a Halloween Party and got six candies and as a parent you said to the child, "Well, you can only eat three a day," and the child quickly calculates that it'll take it two days, and the child can't eat all the candies at once. The child begins to cry. And the computer will say, okay, you have three plus three makes six, two days, that's it. So for the computer, it doesn't understand the value of what it means to the child, and the same as in chess, the computer doesn't understand the value of the strength of the moves.

CHARLAYNE HUNTER-GAULT: Mr. Tan, briefly, what is the value of all of this, the point?

MR. TAN: The point is really not for us is not chess, it's really, we are using chess as a model to understand how we solve complex problems like this with our large computers, and from that, we learn how to make better computers for the future.

MR. TAN: Will have lost, but, remember, the chess is a creative game. With this, we now have found a technology that can do the same thing just as well, and we can, therefore, use this technology to solve other complex problems which cannot be solved today.

MR. TAN: I believe we have a very good chance, and the game will be very close. It will be decided on Saturday, the last game.

Program
Support
From:

MAN VERSUS MACHINE

FEBRUARY 15, 1996

TRANSCRIPT