An interview with vladimir kramnik after his 2002 match with fritz
In 2000, Vladimir Kramnik defeated Garry Kasparov for the title of world champion. He is currently a distant #2 on the rating list behind Kasparov. In October, 2002, Kramnik played an eight-game match against Fritz in the tiny Persian Gulf nation of Bahrain. He scored two early wins but Fritz evened the score in the second half and the match finished in a 4-4 draw.
X3D Fritz is an improved version of that program and will be running on much faster hardware. Kasparov has said that he will take a careful look at the eight Fritz-Kramnik games. Below is an interview with Kramnik immediately after the conclusion of the Bahrain match.
Did the result match your expectations? What did you think the score would be?
It was clear to me that the match was going to be very tough and equal. When I had a few days to see this new program I could see it was a really big improvement over the old one. I didn't know who was the favorite, nobody knew. Maybe people were making me the favorite because they wanted me to be.
Are you happy with the quality of the games?
I showed a good level of chess, especially in the first half of the match. But even in the games I lost I cannot say that I played badly. I made one mistake in each of those games, but only one, and that was enough.
What did you think of Fritz's play?
There were not so many games where it played strangely. In many games it was simply like playing a strong human Grandmaster, it was absolutely normal, absolutely human play.
In game five Fritz played very well, better than any human. It seemed almost equal, but it managed to keeping putting on this pressure all the time, it kept finding these very precise moves, not giving me a chance to get away. I played that game really well, and I shouldn't have blundered, but the position was not so pleasant anyway. I must admit is simply played very well.
What sort of preparation did you do for this match? Did you work on specific anti-computer strategies?
In my preparation I tried to play this kind of anti-computer strategy, in some rapid training games. I could see clearly that it's not working anymore. The positional technique of this program is so much higher than years ago. It pushes pawns, builds the center, and begins activity on the flank. You cannot play like this anymore against computers. So many things I looked at in my preparations simply didn't work. I was shocked to see the level of positional improvement they had made.
I didn't play training games, just positions. Maybe 20 moves so see how the computer played certain things, if it's doing well or not. It was very general checking. I can say that most of the openings we saw in this match I didn't prepare at all. It was a normal fight, I decided to play normal chess, not anti-computer chess.
At the halfway point your second, Christopher Lutz, was critical of the Fritz team's preparation. What are your thoughts?
The Fritz team made mistakes but they learned from their mistakes. They didn't know what to expect and they wanted to start in a very solid way, but maybe this wasn't the right strategy.
What conclusions did you come to before and during the match about how you should play?
I realized that it's possible to beat the computer in some types of endgames. It's hard to explain in words, but in very strategic endgames it's possible to outplay them. I managed it in the beginning, but then they realized it too and they switched. And they were very successful in the second half of the match, they had some good teamwork. I had serious problems in the second half, I wasn't getting the openings I wanted. But I expected this, I didn't expect to get eight excellent endgames.
Is playing against computers good for your regular chess? How is it different?
Playing computers can have a positive influence. My tactical feelings are very high after this match. You really see every little thing. I was managing a level of concentration that I could never manage before. Because you know that if you lose concentration for just one moment, it's over. One mistake and it's just over.
For instance in game six, after Nxf7, I'm sure against most humans this would work very well. I couldn't say it was a mistake, but I realized afterwards that against a computer there is no way to save the game after that. You don't even realize it, but the game is already over! You are still playing, but the game is over!
When I get back to human chess it will take some time to get used to it, it' s quite different. But it shouldn't take too long. And I know very well what I learned here.
They have their good days, I have my good days. I just so happened that I had my good days at the beginning of the match. But I never thought it would be 7-1 or something like that. If you play your best you can win one, maybe two games. Maybe if you are playing great and have all the luck, one more, but not more than that.
What do you say about that incredible sixth game, in which you sacrificed a piece but went on to lose against perfect defense? And then some analysis showed that you could have drawn against Fritz in the final position where you resigned.
Objectively I think the final position of game six is losing, so I cannot say that I resigned in a drawn position. Maybe a computer won't find a way to win because it doesn't understand this fortress, but I cannot say I objectively missed a draw. But it was strange, because something happened that normally does not happen. It was a new experience.
But to take on f7, to sacrifice a knight, well I thought, "let's have some fun, I'm in the lead." I'm pretty sure that at least this game six was very interesting for the public and it will be discussed for a long time so in this sense it leads to popularization of chess and this is nice. Maybe it was too much of a hard way to popularize chess for me, but what can you do?!
Games two and three I think I played very well, there were several good moments and some very good chess played in the match. As for the last two games, I really didn't have much of a chance to do anything. The ChessBase team chose a very safe strategy in the final games. I couldn't even get a chance.
Would you really have played the super-sharp Botvinnik against Fritz in game eight?
I took a very deep breath in game eight, thinking about playing the Botvinnik against a computer. It was a moment quite similar to Nxf7 in game six, a certain challenge. But I figured, let's go and I closed my eyes and played Bg5 and prayed.
I must say they prepared incredibly well. I don't know if it was luck or really very clever preparation, but they managed to use a very strange move order to get me into a position I hadn't seen before. It's very solid for black and they managed to get me into this position.
It's a good position for white, and I would be very happy to have this position against any human. But against a computer, it doesn't care about your activity. It's a pity that I couldn't try a bit more, but it just happened like this. In the second half they were better in preparation. I wasn't getting the positions I wanted.
You are one of the few people on the planet who knows what it is like to be 2800. Is Fritz really 2800?
You can say Fritz is 2800, but you cannot measure it by numbers really. It's very strong, it's very very strong. But it depends on many things, especially the opening. In some positions, if it gets its positions you can make a draw or you can lose, two choices; you can never win. In some positions its 3000. Maybe you can suffer and make a draw. 10 Kasparovs and 20 Anands wouldn't help you in these positions.
So on the average you can say 2800 or a bit more, but it matters what you get. If you get a position like what I had in game five then no human can fight it. But if you get what I had in game two then you have a chance. It very much depends on the opening stage.
Now that it's over how do you feel about the result?
I don't know how to assess the result. I cannot say I'm extremely happy with the result because I was leading 3-1. But I cannot say I am unhappy because after the match I know how difficult it was.
Would it be different to play against a new program that is completely unknown to you?
To create a strong chess program is like to create a strong human chessplayer. You need experience, it takes time. You need to lose games to learn how to improve. It's not realistic to expect a new program to suddenly arise, one that is as strong as Fritz.