USCF Home Chess Life Online 2008 December Interview With an American Medallist: GM Larry Kaufman
|Interview With an American Medallist: GM Larry Kaufman|
|By Jennifer Shahade|
|December 16, 2008|
In July 2008, Larry Kaufman appeared on the cover of Chess Life for his win in the 2008 U.S. Senior Open. Part of the prize included expense money and conditions for the World Senior Championships in Bad Zwischenahn, Germany from October 28-November 10, 2008. Larry turned 60 on Nov.15, 2007, so it was the first year he was eligible. He'd had his eyes on the World Senior for many years, since it offered a clean shot to the Grandmaster title. Although he was ranked 17th going into the event, Larry hit his target, scoring 9 out of 11 to become a Grandmaster and World Senior Champion (on tiebreak over GM Mihai Suba.) Although he earned the GM title suddenly, it was also a long time coming: Kaufman became an IM 28 years ago. Larry is also the author of the Chess Advantage in Black and White, a key member of the Rybka team, a Shogi champion and the father of International Master Ray Kaufman. Larry talked to CLO about preparing for Senior, highs and lows of the event, and what working for the super-chess-computer Rybka taught him.
Jennifer Shahade(JS)- At what point during the World Senior did you realize you were in good form, and that you may have a chance for the title?
GM Larry Kaufman (LK)- Certainly not at the beginning... Strangely enough the first game was the only game where I had a losing position, against a 2050, Herbert Titz (see position below). The second game I got a clearly winning position, but bungled it... I had to play myself into form a bit.
JS-Obviously you achieved great form in the end. Which were your favorite games from Bad Zwischenahn?
LK- The highest quality game may have been my draw against GM Mihai Suba. The game with GM Miso Cebalo was good but I made one fairly big mistake, Qe7 (in the following position) was much worse than Qc7.
Kaufman played 24...Qe7 when White could have responded with 25.Nd6 Rc7 26.e4 and the white pawn will anchor the d6 knight. He should have played 24...Qc7 when 25.Nd6 Rcd8 26.Nb5 Rxd1 27.Qxd1 Qe7 is good for Black.
JS-What are some of your candidates for your best move of the event?
LK- Against GM Janis Klovans, I had the opening prepared well up to the point where he attacked my rook with 14.Qf3.
I had checked up to b4 with the computer in my preparation for this game. I remembered that what I'd studied was Na4 instead of Qf3. This is where my experience working with engines helped me. I knew Rybka agreed with moving the knight. If I had to play Ra7 then it's obvious White has gained something by inserting Qf3, in which case Rybka would have also suggested Qf3. So I reasoned Rykba must have been intending to sack the exchange here... I had to analyze it myself but my logic helped me focus on 14...bxc3. Later I found out that 14.Qf3 bxc3 had been played before.
Another candidate for my best move of the event would be 24.Ra1 in my game against Butnorious, which I annotated for CLO. It was a critical decision whether or not to attack directly, and Ra1 showed good judgment.
JS- In the final stretch, where you scored 5 out of 6 against 4 GMs, 1 IM and 1 FM, what set you apart from your competitors?
LK-There is a big difference in an event with multiple rounds each day and playing one-game-a-day. There’s a whole different element where you can prepare for your opponents. That’s a major reason why I was able to win the Senior, even though I went in as only a 2400 IM, (#17 seed). I know a lot more than most players about how to use computers. The other thing is I’m not a specialist. A lot of people have a very narrow opening repretoire that they know cold. I’m the opposite. I can play almost any respectable opening, so no one can predict what I'm going to play.
JS- Why have you developed such a wide repertoire?
LK- I do a lot of teaching, so I have to be able to play any opening my student wants to play. I also get bored playing the same lines. It’s more interesting for me to play new openings, and expand my horizons. I may be a chess-pro, but I’m not a chessplaying professional. Teaching, writing and software are all part of my income. So in my case, I don’t have to do things strictly based on optimum results, I can also think about what will be interesting for me.
JS- Well maybe it is optimal, that strategy makes it really hard to prepare for you....
LK- It’s true that it's almost impossible to prepare for me, I just play too many openings. I remember GM John Fedorowicz once told me at one of the U.S. Championships in Seattle, “I’ve played you ten times, and I’ve never once guessed what you’re going to play” After that, he decided not to bother preparing for me.
JS-What sets your preparation apart from another player with a big database and a strong engine?
LK- I have a feeling that a lot of players don’t understand the differences between programs and what they should have. They don’t have Rybka because they don’t know how big of a difference it is from Fritz. And then they use the free version as opposed to the new version. Beyond that, a lot of good players don’t trust programs that much. They're thinking five years ago, where engines were mostly just tactical checkers. Now you can actually trust Rybka, except in positions that are particularly blocked up. Rybka from convecta with aquarium has a feature where you can have Rybka analyze as many positions as you want for as long as you want. I have mine running overnight in the openings I’m interested in. The next day I have extremely high quality analysis.
JS- But you were not doing that during the Senior, were you?
LK-No, I do that kind of work on my more powerful home computer prior to the tournament although I still spent a few hours a day studying at the event. GM Eugene Perelstheyn should get some of the credit for my choice with Black in this event, the French. While he was here in Potomac (a suburb of D.C.) for one of the Rybka matches, he said he was interested in playing the French and thought we ought to analyze it together with Rybka. After this work, I became convinced. In all the (World Senior) games I faced 1.e4, I played 1...e6.
JS- What exactly do you do for Rybka?
LK- The program has to do two things: search through millions of positions, and then evaluate them at the end. I’ve been responsible for the evaluation function. I get a file with terms in the evaluation function and I assign numbers to terms and come up with new ones if I see something that is missing.
JS- What sort of terms do you mean?
LK- Things like, Isolated pawns are bad, bishop pairs are strong, connected passed pawns are better than disconnected ones.
JS-What’s an example of a new term that you came up with?
LK- At some point, we didn't have “pawn islands.” I also put in a new term about queen and pawn endgames where having a very advanced passed pawn is much more important than the number of pawns, while that is not true in most other endings.
JS- But how did you rank the evaluation functions? Isn't that hard to do, since all these things can be winning in some cases and losing in others?
LK- To figure out which evaluation principles are valid and how valid, I had to run big tests. It’s not a yes or no thing. You can read a book on strategy and positional chess, but many of the principles contradict each other in a regular game. How do you rate them? Which is more important? So with every proposed change, I tested to see if Rybka with the change could beat Rybka without the change. Most of these principles only show up every few hundred games, so a typical test run would be 100,000 games.
JS- Wow, 100,000 games. That sounds time-consuming.
LK- To run so many they have to be very fast games. A typical testing speed is a game in one second. (At this speed, 100,000 games would take just over a day.) Rybka can play pretty well on one second on a decent computer.
JS- What were you most surprised about when running these big tests?
LK- A lot of the principles turn out to be invalid or not very valid. For example, Rybka testing does not show that pawn islands are very significant. The weight is very tiny. On the other hand, knights on very good squares turn out to be huge, much larger than I would have guessed.
JS- So outpost knights are big, I’ll remember that! I always feel like the computer is just calculating piece values, even though it now occurs to me that even a value of a piece is a "principle" in a sense. The only thing which is really 100% concrete is analysis that leads to checkmate, and that's one thing that's impractical for both people and computers!
LK- (laughs) yes, I guess that's true.
JS- With so much time spent on the computer, do you still use and like books?
LK- If I get a new opening book, then I might put a lot of the variations in the book into the “Aquarium” and see if it agrees or disagrees. Then within a day or two, the opening has been updated by Rybka’s analysis. Generally the authors are using engines too, so the main lines will check out. When a new opening book comes out that I have reason to believe is well-done, I'll probably get it. John Cox writes a lot of good books... James Vigus did a great job with his recent book, Play the Slav ... It was obvious he worked really hard on it.
JS- Is it a valid concern that if a chessplayer uses Fritz/Rybka too much they will lose out on valuable practice? In my last "Interview with an American Medallist" with Sam Shankland, he said he rarely uses Fritz when training. Even more surprising to me was that in my interview in this month's Chess Life with Kosteniuk ...she said she doesn't use engines at all when studying.
LK- You have to try to keep your brain on working order. You don't want to overdo it. But if you really want to find out the truth, you have to use the computer. The lower level the player, the less useful computer analysis is, cause they won't be able to figure out what to do when the opponent varies. If you are a high player level, you'll at least hope to figure it out. For similar reasons, if people ask me, What section should I play in?, my normal answer is: "play up a class but not more than a class." If you play an equal player, that's not optimum, they're making mistakes that you might see... but if they're too much better, you'll just get blown away. If you're 2400 or 2500 it's reasonable to analyze with computers that are a few hundred points better than you, whereas if you're 1800 it's not as useful.
JS-Is your work with Rybka a full time gig?
LK-When Rybka 3 came out, it was close to a full time job. Right now, I'm not working so much on Rybka, I'm more into using Rybka.
JS- There is so much emphasis on junior and kid chess around the world. Do you think there should be more attention to Senior chess?
LK- There's probably some truth to the idea of chess warding off Alzheimer's. it seems there's evidence that any intensive brain use will do it and chess is as intense as you can get. I think that can be a significant promotional tool. When people retire, they can have time for chess again. In Europe, senior chess is a bigger deal (than in America.)
JS- Your son, Ray Kaufman also got his IM title approved in Dresden, so this is a great season of chess celebration for you two.
LK- Yes, it's pretty funny that we both got titles at virtually at the same time. Ray worked so hard to finally become an IM and he could finally say he caught up with his dad. It didn't work out that way cause I got my GM title a week or so before his was confirmed.
JS- Well now, Ray has extra motivation to shoot for GM! When was the first time you thought about the prospect of the Grandmaster title?
LK-For a long time now, at least five years, people would ask me Why aren't you a GM? or Are you still trying to be a GM? I'd always say, it's not realistic at my age to think of making three GM norms, given that I have a family and can't just move to Europe. But I always said, when I turn 60 I'm going to try to win the World Senior and get it that way. I definitely didn't expect to win but I went to in it trying to do everything I could. I bought a better laptop just because it's faster. I stayed in Germany five days before the event to make sure I was not jetlagged, which gave me a chance to drop by the World Championship Bonn. I stayed in a hotel with a pool so I could be sure to get exercise. I didn't hesitate to spend money if I thought it would improve my chances.
Look for an article by GM Larry Kaufman on his big win in the February 2009 issue of Chess Life Magazine. USCF Members can also read Jerry Hanken's July 08 article about Kaufman's win in the 2008 U.S. Senio r, which includes annotations from Larry. The next Interview with an American Medallist will feature WGM Rusudan Goletiani. Also see the first installment with IM Sam Shankland.