Greengard's ChessNinja.com

Kasparov Retirement Interview

| Permalink | 58 comments

Part 1 of my interview with Garry Kasparov is up here at ChessBase. All I have to say is thank god the typing is over. Seven hours of recording into twenty pages of interview, yeesh. Why didn't I just dump the entire thing into one MP3 file and get it over with? It's a lot tidier this way, since the conversations took place over a week. So I changed the order around in quite a few places.

Many of the pictures at his hotel room were taken by his 11-year-old daughter Polina and his slightly older fiancee, Dasha Tarasova. (Kidding, kidding. She's 23.) I just handed them my camera. There will be pics of them up with part two or three. It's a sort of double byline interview since much of the material comes from when I was present when Garry was being interviewed for the NY Times piece. They used maybe 3% of it and much of that focused on politics, which I've put into part three. With my own questions I wanted to concentrate on the chess questions the mainstream media aren't interested in.

In much of that I tried to pin him down on the theory of evolution, chess evolution. After his work on My Great Predecessors no one is better qualified on the topic, so I asked him a lot about whether, for example, today's top players are really better than he and Karpov were in the 80's. As for best games ever, we discussed different criteria. From Part 2:

Itís actually going to be quite a problem for me to collect my best games. Even by my judgment there are many that qualify at the highest standards. Let's see, games 16 and 24 from Moscow [1985 Karpov WC match], game 24 from Seville [1987 Karpov WC match], Korchnoi '82 in Lucerne, and the Topalov game [Corus 1999]. But then youíre missing game 22 from Leningrad with Nd7, the sealed move [1986 Karpov match]. Okay, so those would five good and memorable ones. But really game 24 from the Seville match wasn't a great game. So maybe cut that one and I'd take Anand, game ten of the [1995] New York world championship match.

What criteria are you using?

The Seville game would be just as a sort of heroic accomplishment. Korchnoi was a world recognition game. Games 16 and 24 from Moscow were great ideas and important games. Also, the decisive game of the match and a great novelty. Topalov, probably best combination ever.

By pure chess standards it would be the two with Karpov, 16 and 24. The Anand and Topalov games, and... hmm. I would add to this list the Astana game against Kramnik, the Berlin Wall with e6. Runner-ups would be the Seville and Korchnoi games. But I had a problem making a list of thirteen best games! I had modest aspirations of having thirteen ďbest of the bestĒ games.

He then talks about how the Predecessors series, now expanded to ten volumes, will include two on his own games with over 250 "best games"! That should make the selection a little easier.


Wow, the interview is really good. Thanks Mig!

Some interesting comments from Garry. He managed to show his arrogant self while justifying it with convincing arguments. For one, his take on why he did not participate in Dortmund 2002 (no guarantees of a final match against Kramnik) is good, even though I don't believe him (why did those reasons only come up now? A few months ago Mig said Garry's main problem with the format was the rapid/blitz playoffs).

Kasparov wanted a direct rematch, Kramnik wanted Garry to go through a qualifying cycle. Regarding the latter Kasparov's demands were unreasonable: guaranteed prize pool of 2 million dollars for the final match with Kramnik? He should have known that Kramnik and his association did not have enough leverage to make it happen. Kasparov might well have said "Kramnik, you have no ability to raise the huge amounts of money/guarantees to convince me to play in a qualifier, so you are obliged to play a direct rematch against me".

That being said, Kasparov is justified in stating that Kramnik has proved himself very little since 2000, and especially in the past year. The draw against player #5 (Leko) was unimpressive and Kramnik's recent tournament results are a joke. The quote attributed to Kramnik (can anyone point me to a reference for it? ) "how can you use a match played two years ago to judge the better player?" was dead on and shows how hypocritical he is of the whole situation.

Too bad Kramnik will not come up to defend himself on this board because I would like to hear both sides of the issue. Greg Koster, wanna share some of your thoughts? Maybe you can present Kramnik's point of view on the matter, and explain why he should still be regarded as the best chessplayer.

Wow! This is one of the best chess interview ever! Very informative! By the way,when did Garry split up with Julie (his second wife)?


John (aka littlepawn)

It's definitely not fair to say that Kramnik hasn't won any tournaments since 2000. He did, after all, win Linares 2004 and several others, if I'm not mistaken. I realize that this is unimpressive compared to Kasparov's victories in the two years following their math.

I definitely like his idea of a chess school in the US. Don't you guys think that maybe it was a bit rude to say that the reason that Topalov got thrashed in Amber 2005 was because he had used up his "chess luck" against Kasparov and the Linares field? It's interesting how Kasparov just HAD to throw that bit in.

Also, I'm very happy that he has just added Nakamura to his list of "the next best".

Kasparov retires at 2812.

Did someone notice that 2 + 8 + 1 + 2 = 13!?

13 is Kaspy loved and mythical number...


10 volumes of "My Great Predecessors"!
Was he thinking in humiliating George Lucas and the 6 "Star Wars" movies? ;-)

Great interview, Mig, I'm definitely looking forward to parts 2 and 3!

Murali, this interview shows a progression in Kasparov's views about the World Championship.

In the late 1990's Kasparov's conduct and comments appeared to indicate his adherence to three principles:

1) The world champion remains so until he dies, abdicates, or is beaten head-to-head. "My title cannot be taken away by decree. To be the new world champion, somebody is going to have to sit down across the table and beat me fair and square."--Kasparov. A five-year hiatus between title defenses (1995-2000) does not constitute an abdication.

2) Defeated champions should not be given a rematch; and the challenger should be determined through a candidates event. Kasparov's criticized the rematch rights exercised by past champions. And rather than annoint a challenger he initially invited the top candidates to fight it out in a series of matches that resulted in Shirov-Kramnik.

3) A potential challenger who refuses to proceed without iron-clad match terms and financial guarantees loses his shot. Kasparov invited Shirov, Anand and Kramnik to participate in title events without financial guarantees. Shirov agreed, but sponsorship fell through. Anand declined and lost his title shot. Kramnik agreed and ultimately won the match and took over the champion's responsibilities of finding an appropriate challenger.

By inviting the world's top players to the Dortmund Candidates event and asking them to play without iron-clad financial guarantees. Kramnik was acting in accordance with Kasparov's late-1990's principles. As Mig's interview demonstrates, however, Kasparov's principles of the late 1990's have undergone modification and now, in 2005, Kasparov evidently believes:

1) Two years without a title defense or outstanding results in tournaments constitutes de facto abdication.

2) The champion should not stage a candidates event but should handpick the highest rated challenger and

3) A contender who refuses to participate in a candidates event without ironclad match terms and financial guarantees should not lose his shot.

It is unclear which principles Kasparov was adopting when he accepted FIDE's title shot invitation but declined the invite from Dortmund. Kasparov signed on to the Kasparov-Ponomariov and Kasparov-Kasimdzhanov matches (and sacrificed much time and money preparing for them) without first demanding match terms and financial guarantees; he proceeded out of faith that he could ultimately work things out with his opponents and FIDE president Kirsan Ilyumzhinov and that sponsorship could be found. On the other hand, Kasparov declined his Dortmund invitation, insisting on match terms and financial guarantees before he'd put himself in line for a possible Kasparov-Kramnik II.

But neither the 13th nor the 14th champion is running from the other and Kasparov's comments to the contrary are unfortunate.

Murali, where did you come up with the idea that Kasparov had a demand of 2 million dollars to play Kramnik? He said he wanted an invitation with a confirmed match. The two million figure came from Einstein. He was saying "where is the money" not "this is what I want." Important difference.

Greg, just to point out that he didn't make the number up, the two years Garry mentioned was the span provided for title defense by the agreement for the 2000 match.

Kasparov's doubts about Dortmund were borne out entirely. It took over TWO YEARS for Kramnik-Leko to happen. Wanting a real contract with guarantees before putting yourself at the mercy of your opponent's agent isn't a bad idea. (Yes, one of Dortmund's organizers is Kramnik's agent. Also Lekos agent, so that was easy...)

Garry explains at great length why he thinks his title was valid and Kramnik's is not. Maybe you should read this new interview the rest of us are talking about. There's a link at the top.

For the record (yet again), I disagree with him on this, which is why I thought it important in the interview to pin him down on exact reasons.

As for accepting FIDE and turning down Dortmund, the Dortmund invitation was rejected half a year before Prague, which changed many things. Kasparov really didn't want to toss in with Ilyumzhinov because he knew it was hypocritical after years of bashing him. But it was the best possibility for a shot at the the title, although it certainly bit him in the ass. Of course had he rejected Prague he would have been universally condemed for blocking unification, so it wasn't a good situation.

As bad as FIDE was and is, Braingames/Einstein had already defaulted on several things and Garry had no wish to hook up with them again, especially when an "enemy" had joined their ranks (Hensel).

There's an irony here: Kasparov talks about "world recognition" from a game with Kortchnoi, which he considers for that reason to belong to the "best" category. Yet he showed such a terrible lack of sportsmanship after he lost "the most beautiful game" to a 12 year old boy. Not only did he refuse to shake Radjabov's hand, but he also strongly objected to the grounds on which the prize was made, namely, the fact that a boy beat the champion in a game featuring an offered sac. I like Kasparov, generally (I read his earlier autobiography while in high school), but it's hard to forgive him for that one.

The Predecessors series is now 10 volumes? Who gets the extra pages? I trust that isn't just case of splitting up the material into thinner books. Hmmm - wry comment coming - I suppose it's too much to ask that one volume be devoted to the computer!

On Deep Blue Garry says: "Show me the printouts of all the games, donít tell me we canít understand them or they are too complicated."

During the midst of a competitive match I don't consider this a legitimate request. After it is done, okay. (Contrary to his belief, the matches were not conducted for the sake of science; speaking as a computer scientist myself, the structure would have been completely different were that the case.)

What surprises me is his insistence that the logs were never released. They were. You can scrutinize them yourself. I don't know exactly when they were released but you'd think Garry would have found out through the grapevine pretty fast. Not that he'd care to look at them anymore, I suspect. Too many open wounds.

In case you are wondering, here is the secret location.


NathanB, Yes he showed poor sportsmanship, but he was absolutely correct in his assessment of the game. The game was crap: The kid was worse and Kasparov blundered. It wasn't even close to the best game, except for that fact about who lost. He was pissed off about the journalists who voted is a questionable manner.

One other comment in general: Kramnik is still a chickensh**, and did in fact do everything possible to avoid Kasparov after 2000. As far as Kasparov's comment about Kramnik not winning tournaments, I really think that you have to read whti Kasparov's ego taken into account: Of course Kasparov knows Kramnik won a few tournaments, but I would venture to guess (and I personally immediately felt this was what he meant) that he meant that Kramnik was not sucessful in winning any tournaments where they wer both participants.

Mig, since Kasparov is your friend, I am curious if the two of you have ever passed the time playing blitz, as friends who play chess often do. And if so, did you ever come close to winning?

Regarding the Radjabov game, Kasparov said that if the prize had been for impact or recognition he would have awarded it to Radjabov himself. Comparing it to the spectacular game against Korchnoi is bizarre. I think the argument I made at the time stands: if anyone other than Radjabov had been playing black, the game doesn't win a prize.

jkominek, I don't think Kasparov was talking about asking for the logs during the match. He was referring to now when he is making his accusations about evidence. IBM said they would release the logs and they did, but several years later. Does anyone know exactly when those logs were posted? I remember the discussion but not the date.

Mind, I don't think they cheated, but they certainly did everything they could to allow that impression to fester, which may not have been unintentional. To me, the best argument against cheating is the simple fact that today's programs have caught up and play the same moves, or better ones, than DB did. On my Athlon64 3800+ desktop, Junior 9 plays both 37.Be4! in game 2 and 11...h5!? from game 5 as its first choice after less than five minutes.

In a way, Kasparov's argument that today's programs are better than Deep Blue refutes his theory that they cheated. It's a little far-fetched to think Deep Blue wasn't at least five or six years ahead of other programs. (I.e., it would be silly to say that today's programs have reached a point that Deep Blue could only reach with human interference.)

Ooh! A response from Mig! :-D I just have to follow up just slightly: I don't remember reading what you said Kasparov said. I do believe you, and maybe I missed it, but I do remember him ranting at the journalists because of the fact that they felt the game was beautiful because a boy beat Kasparov. This tends to undermine his more gracious take that you mentioned on the game in question. And I still think he's being inconsistent about the Kortchnoi game. I also remember reading about his refusal to shake hands with the boy-GM after the game.

Anyway, that's all I'll venture to say on this subject as I've said my piece. Btw, I think it's great that there's a chess blog out there, and yours is a good one.

Thanks for the great article! Looking forward to parts 2 and 3 with photos as well. Believe it was Korchnoi who stated that all GM's were "mad" to a certain degree. When we read the writings from Garry and Susan Polgar at least, I'm happy to know this doesn't always hold true. Thanks again for your frank interviews.

I'm just not sure what the Radjabov game, award, and subsequent Kasparov tantrum have to do with the Korchnoi game. That was one of the most amazing games of the year, period. The point of Kasparov's anger about the Radjabov game was that it wasn't. Kasparov's comments about the incident and the game are here:


Of course the Radjabov game could be as important for recognition as well, but that's not what the prize was supposed to be for. But comparing the impact of the two games is reasonable, although winning on a blunder isn't quite the same as playing an evergreen.

You know, Mig, I read Chessbase.com pretty regularly, but somehow I missed that one. Thanks for the link. I feel a bit better about Kasparov now.

Ha, me playing blitz with Kasparov?! I'm far too weak for that to be interesting for either of us. My online blitz rating generally hovers between 2400-2500. I pick up the odd win against a GM, but those are guys who are lucky to win one in a dozen against Garry.

My only experience over the board against Garry was when he was training for his handicap match against Chapman in 2000. Since I was around the same level as Chapman (~2300) I got to be a crash test dummy, getting two pawns against him to test the different configurations, etc. I got to one theoretical draw, which I lost. It was informal, no clocks.

I've sat in on various analysis and preparation sessions, but knew enough to keep my mouth shut!

For those who did not know the whole series is going to be 10 books, here is the recent split list:
Vol.5. Karpov - Korchnoi
Vol.6. Karpov - Kasparov
Vol.7-8. Kasparov's best games
Vol.9. Kasparov vs. computers
Vol.10. World Champions biographies (no chess analysis).
As you see, MGP itself is still in 6 volumes. May be 7, if count the last one.
This first appeared during his interview to Sport-Express after the Linares press conference.

Mig, forgive me for asking, but I see one Michael Greengard in USCF's ratings who happens to live in NY but has an 1825 rating. FIDE also has nobody listed under Greengard. Is that an old rating?

Some statements in this interview sound like a joke:
" In 2000 my title had value because I was the best in the world. In my book, Kramnikís title expired no later than 2002. He had to defend his title and did not. More importantly, I won a few tournaments in a row in 2001 and he failed to perform at the highest level..."

Kramnik is the second player in the history to become 2800+. And he was there during July 2001 - January 2003.

What Garry really means, is "The problem is that from a wider perspective, looking at our later (sic! - Vlad Kosulin) results, it was an anomaly. So either he needed to play a rematch, and if he beats me again then thatís it, heís made the point and shown it wasnít a fluke".
For Kasparov only the best can be (not become, but be, sic!) the WC. And while during last 2 years Kramnik continues to fall down, Kasparov is not allowed to get the title back. But Leko is. Here is his point!!! This makes him nuts and this is why he makes all these repeating bullish statements like "Kramnik is afraid of playing with me", or "They all (sic! - Vlad Kosulin) were gloat when matches (with Pono and Kasim) were bog down." He is just nuts on the whole chess world.

Follow-up (this was accidentally cut from my previous post):
If Garry believes Kramnik's title expired in 2002, why and what was he trying to replay during last 3 years? Why wouldn't he play in FIDE champ, or organize a new title match with, say, Anand? We can afford one more Champion, why not? ;-)
I can understand Garry is nuts of being beaten by somebody and not allowed to get a satisfaction for 4 years, but this is very disorderly sports (thanks to Garry!), and he had to fight like a lyon to get the new match, he had to participate in every classifier (Dortmund, both FIDE champs) to get the new match right, instead of talking talks to FIDE knowing they are morons, and if he gave up, he gave up. No matter why!

I think you need to look up what "sic" means. And while I'm all sure we respect your mind-reading abilities, it seems clear that the point he's making is that Kramnik didn't perform better than HIM, not that Kramnik is a horrible player.

Jonas, my USCF rating is from when I was in high school, from around a dozen games. I didn't play chess seriously until I left the US and played in Argentina. I played there regularly for five years until around 1997, gaining a 2310 Argentine rating in club play and occasional open events (they use the same formula as FIDE). After a six-year layoff I played in the World Open in 2003 with predictably disastrous results. Actually, I played some decent chess at the Amateur Team event earlier in 2003, but that's a casual event and I was relaxed.

Sorry, Mig, about "sic", my fault :-)
How would you mark something worth a special attention, like "mind!"?


Quotation marks around a word or phrase can be used for emphasis or to indicate irony, for example:

--While I'm all sure we respect your "mind"-reading abilities...

I remember now: I should use NB! instead of sic!.

Greg, I did not mean to indicate an irony, but to point reader's attention to the word or phrase I marked.

Mig wrote: "jkominek, I don't see Kasparov asking for the logs during the match. It was after the match when he was making his accusations. IBM said they would release the logs and they did, but several years later."

Oh he most definitely made the demand during the match itself. This was his undoing. Game 6 would never have been that disaster had he kept his composure and focus. By asking for printouts after Game 2 Garry revealed that he was rattled -- and IBM exploited the weakness.

I'll quote a bit from A New Era: How Garry Kasparov Changed the World of Chess by Khodarkovsky and Shamkovich. Page 202: "all these were so inconsistent with computer play that Garry requested explanatory printouts for three moves, 35. Bxd6, 37. Be4, and 45. Ra6. We were told we could have them just after the match. But he wanted to understand why the computer refused to win material, and instead played very positionally, the way a strong grandmaster would play. Garry requested he printouts through the Appeals Board, which comprised Monty Newborn, Ken Thomson, and Mike Valvo. But IBM's CJ Tan refused to give them up. After negotiations between the Appeals Board and CJ Tan, Dr. Tan at last agreed to give the printouts of the move not to Garry but to Ken Thompson. Ken Thompson asked CJ Tan to email or fax the printout to him at home. Little did we know, the war of the printouts had begun."

And from Monty Newborn's book Deep Blue, page 162: "Kasparov asked Ken Thompson to examine Deep Blue's printouts of the two moves in question. Thompson, in turn, approached the Deep Blue team for the information, but obtained it only prior to Game 5. An examination of the printout revealed nothing of concern. Thompson's observations were conveyed to Kasparov, but the long delay in obtaining the information troubled Kasparov to the point where he seemed to find it difficult to concentrate on the primary ask at hand."

As a side note, I personally began the match cheering for Deep Blue and left sympathizing for Garry. IBM won through psychological pressure, not by over the board play itself, and though some in business view a "win at all costs" attitude as exemplary, I found it distasteful.

Plus, IBM blew a fantastic publicity opportunity. The way to derive most bang for the buck from the series is not to lose and then win the rematch (with sour feelings), but to keep it going. Not that it can be proscribed by design, but the ideal thing is to lose the first match, draw the second, then win the third. This build maximum suspense and drama.


P.S. Vlad Kosulin - thanks for the new volume breakdown of MGP. Looks like my wry comment wasn't!

Sorry, I was just referring to what he is said in the quote you gave about not saying we won't understand the logs. I'm aware of what happened during the match. Good quotes anyway.

You can see more about the GK books at http://www.chesschamps.com

As for Kasparov-Deep Blue game 2, Garry asked for the proof IBM is not cheating next day, during the press-conference immediately after game 3 finish. And he repeated the claim few times after other games.

As a matter of fact, he resigned instead of fighting for a draw because he was totally shocked. You should see his mimics during the game!

You can watch the whole match story shown in great details in a full size documentary "Game Over. Kasparov and the Machine" made in 2003. Mig should definitely know the details, because he made comments for the movie.

Again, I'm well aware of the details, having watched video of every game and press conference and gone over it recently with the people who produced and directed "Game Over". I thought the poster was talking about specific comments.

I only appeared for a few seconds in the movie (thank god; I look like a cave troll in it), but they spent around 12 hours with me getting background and going over the chess aspects of Garry's paranoia.

Actually I think you see my hands and computer screen as much as you see me. When they show clips of moves being made on the screen that's at my apartment on my laptop, using Fritz 8.

The scene where you see Garry get up all shocked is the resignation of game six, if we're thinking of the same scene. Game two was actually a drawn position, and finding that out before the next game upset him easily as much as any stuff about cheating and logs.

Oddly (or not?) something similar happened to Kramnik against Deep Fritz in Bahrain. He resigned in a position that, with best play by white, Fritz would never have won. It's hard to prove a forced draw, but it was relatively simple to show that Fritz couldn't find the win because it loved to go into a queen vs rook position that was a perfect fortress draw. The key difference is that Kramnik didn't find out until after the next game, not before!

More on the Kramnik-Fritz game here: http://www.chessbase.com/newsdetail.asp?newsid=558

It shows how your psychology is affected by playing a machine. Two of the world's best players simply assumed the game was lost because that's what the machine thought. There is no way either of them resign against any human in those positions.


Re the Linares 2003 "Beauty Prize" incident:

The many posts on this topic over the last two years all agree on the following points:
The game Kasparov-Radjubov did not display highest quality chess. Kasparov's conduct was inappropriate. And whether a 15-year-old newcomer's defeat of a legendary grandmaster renders their game "beautiful" depends on one's definition of "beauty". A more interesting perspective on the matter might be to try to ascertain what it was that motivated Kasparov to act as he did.

(For the Chessbase account of the incident see http://www.chessbase.com/newsdetail.asp?newsid=851)

Under what circumstances might an individual be expected to act as Kasparov did? Let us imagine a seasoned businessman whoís dominated his corporation for thirty years. One morning a fifteen-year-old walks into his office and they begin negotiations on a certain matter. The seasoned businessman is initially prevailing, but then suffers a temporary lapse and finds himself out-negotiated. The following week at an awards ceremony to commemorate the negotiation an elaborate program is presented, praising the 15-year-old as the leader of the future and suggesting that the old businessman is a has-been who should just go away. Under such circumstances one might expect the stunned old businessman to be unable to restrain himself and to lash out with the equivalent of: "How could you give the beauty prize to a game in which I lost a piece because of a stupid mistake? ... I consider this to be a public insult and humiliation....This is the greatest insult that you have done to me in my life! It is an insult to me and to chess.Ē

If at the Linares award ceremony Kasparov lashed out because he suddenly felt he was being treated as an unloved, blundering has-been who should just go away, the reasons are not difficult to discern. One imagines that Kasparov still carries with him a self-image formed during what was perhaps the brightest period of his life, when he was the universally beloved, invincible, teen-aged ďboy from BakuĒ standing on the path to the world title. What a shock to that boyish, invincible, beloved self-image the Linares award ceremony must have been! The Linares award was given because Kasparov was old and Radjubov young, because (for the moment, at least), Radjubov was invincible and Kasparov a blunderer, because Radjubov was beloved and Kasparov not.

One expects that Kasparov might do most anything to bring back the days when he was youthful, invincible and beloved by all. Lately heís engaged himself to a youthful woman. He works to project an image of a player head and shoulders above the rest. And heís abandoning the quarrelsome arena of professional chess and its attendant politics, expressing sincere appreciation for the many kind notes he's received on the occasion of his retirement.

Letís hope that in retirement he finds the contentment he desires.

I originally quoted: "Show me the printouts of all the games, donít tell me we canít understand them or they are too complicated."

Mig wrote: "jkominek, I don't see Kasparov asking for the logs during the match. It was after the match when he was making his accusations. IBM said they would release the logs and they did, but several years later."

Mig rewrote: "jkominek, I don't Kasparov was talking about asking for the logs during the match. He was referring to now when he is making his accusations about evidence. IBM said they would release the logs and they did, but several years later."

Mig, your edited sentence is missing the main verb. Probably you meant to insert "think" or "believe".

Either way, you say he's talking about now? How can that be? The logs are available for examination now. I'd love to see him analyze them in print. (According to www.chesschamps, volume VIII is dedicated to Man vs. Machine.)

When did IBM post the logs? Good question. Digging through computer chess newsgroups may provide the answer, or we could just ask David Levy. My guess is not until after there was no chance of a second rematch. So my take on the quote is this: Garry is talking about the year or two after the 1997 loss, when paranoia ran high. During your interview he briefly "relived" that difficult time.

Or he could have been reliving the match itself. Certainly he wanted the logs then.

Yah, fixed it, thanks. I'm not entirely clear on what he meant either. I suppose it's possible he doesn't know about the logs, or perhaps he considers them either insufficient or possibly doctored after the long delay? Conspiracy theories have a nice way of expanding to include all available evidence. I'll ask him today.

My feeling is that it's just a very generalized anger and frustration at feeling he wasn't treated fairly and got jerked around by IBM. Perhaps it's unkind to say, but if he had drawn game six this wouldn't be nearly such a sore point, I imagine. Losing sucks, as we say around here.

If I'd put the best years of my career into a project (Read _Behind Deep Blue_) and the guy across the table was denying my success and calling me a liar and a cheat, I'm not sure I'd be terribly motivated to cooperate with him.

I should also note that above I didn't include another move. While Game 2, 37.Be4 was a very "human" prophylactic move, the more contentious one was the previous move, 37.axb5 instead of grabbing material with 37.Qb6. I don't know if any current programs play that one yet, although analysis shows it was the best move. The same lines and point are there for 37.Be4, but that one other computers play now. But since 37.axb5 is always a close second choice, this doesn't seem very relevant either.

For Kasparov, on suspicion alone, to insinuate that IBM was cheating is as outrageous as if IBM's team, on suspicion alone, had accused Kasparov's team of flashing him coded hand signals. One would expect an individual aspiring to a major political role in of the world's great nations to have managed this matter more skillfully.

Kasparov's conduct during and after this match must have given pause to any potential corporate chess sponsor.

Perhaps, but there is no way that a potential backlash was close to the sheer publicity generated. As Kasparov says, his complaints got very little publicity. It's not as if there was more corporate sponsorship before, and Kasparov remained bankable. KasparovChess.com brought in millions and there was the MSN game versus the world.

Kasparov's temper and occasional tantrums shouldn't surprise anyone anymore. (Not that they should be ignored or forgiven.) Certainly his political roles are unlikely to be brinksmanship and diplomacy. Politics is a big game, there is room for many temperaments. Russia hardly has politics as usual. Putin is not going to be moved by subtlety. It's also important to have people on the front lines willing to yell and fight. (E.g. Ross Perot was a loon and a lousy politician, but put a lot of issues no one else would touch on the table.)

One also hopes that Kasparov will meet the process halfway at some point, in those cases where the yelling and fighting doesn't work. Politics has a way of doing that, although again, Russia's situation is very different. Climbing atop a tank had a lot of meaning there not too long ago.

My thoughts:

Garry Kasparov's argument for Kramnik's title being illegitimate is a good argument for Kramnik being a weak paper. It does not however take away from the title's legitimacy.

Kasparov's views on title defenses actually have never been ironclad, aside from belief in an active champion. Anand, Short and Kramnik have all ended up playing him under very different circumstances.

Prague changed everything. It seemed like all the crap could be put behind us with a series of three simple matches. One took place.

Kramnik said in October 2004 that he sees no reason to play winner of Kasparov-Kazimdzhanov, citing that he signed up to play winner of Kas-Pon and that fell through and now it's a completely different story. Two months later the match collapses. I don't doubt that had Kasparov felt he would have a shot at world title in case of victory he wouldn't have retired right away or at least not allowed the match to disappear into oblivion. In February after the match fell through Kramnik accuses Kasparov of not wanting to play a match before playing Kramnik, along the lines of "I held off Leko, why don't you play somebody too, Garry Kimovich?" This may not be ducking. But it is Kramniking.

The saddest part is when Kramnik wants to play, he plays at least as good as anybody else in the world. Crushed Garry in 2000. Put on a power-punch effort in the last games of the Leko match.

Chess is dying. Few people left are worth to watch. They play like automatons, the top are not who plays best but who makes the fewest mistakes. My hopes are pinned on Mister Topalov, out of Bulgaria.

I am just curious as to Kasparov may think the "guiding hand" on Be4 and axb5 was. Was it Joel Benjamin? Surely he is a decent grandmaster but not anywhere near the level of a superstrong grandmaster and it is doubtful he would have felt confident to make such a decision for the machine in such a critical position. But if not Joel, than who could it be? The idea is a little bizarre. In my opinion, Kasparov is better off sticking to the arguments that they prepared for him but he wasn't allowed to prepare for them, that it was a one off thing with alot of emotional and pyschological pressure(see the last game especially) and that the damn thing doesn't even exist anymore!

Maybe he'd be even better off just getting over it and moving on.

Two thoughts here...

1)I wonder if people aren't grossly exaggerating when they say Kasparov's loss to Deep Blue caused sponsors to lose interest in chess. I bet people who don't follow chess regularly don't even remember that match - if anything, people at large would love to see a rematch (especially considering that the human was defeated by a slim 2:1).

2) Even if Kasparov's suspicions are unfounded, there is no question he played under the unfairest of conditions. Deep Blue had access to all of Kasparov's games, Kasparov had no knowledge of the newer Deep Blue whatsoever.

Why did IBM dismantle the computer, other than to be able to claim forevermore that Deep Blue defeated the legend Kasparov and was thenceforth unbeatable? Furthermore, IBM's general reticence (not to mention outright untowardness) towards Kasparov's inquiries casts a dark shadow over IBM's ethical behavior during the whole process.

I agree that IBM's behavior during and after the match was very poor. However, I understand that Feng-Hsiung Hsu (Deep Blue designer) obtained commercial rights to the Deep Blue design and offered to play a rematch with Kasparov, however Kasparov declined (http://www.chesscenter.com/twic/feng.html).

I think it is unlikely that current generation programs running on non-specialized PC's are better than Deep Blue. Deep Blue was running on on dedicated VLSI chess chips that peform in a single processor cycle (e.g pawn structure, king safety) what takes an Intel chip 100,000 cycles to do. In 8 years since the match, that advantage 100,000:1 advantage has shrunk by a factor of only 10.

The problem with Hsu's offer is that he didn't have a machine and didn't have any money. His offer was "find a sponsor for the machine and the match and I'll build it and play you."

No one debates the superiority of Deep Blue's hardware. But compare the nodes per second of the top micro programs HIARCS, Shredder, and Junior. In the same position you'll see their NPS separated by roughly a factor. HIARCS looks at (say) 20,000, Shredder 200,000, Junior 2,000,000. Yet they play at an almost identical level.

And we only have six games. Analyze them with today's programs and you get similar moves and evaluations throughout. In some cases they are superior, as Kramnik and Kasparov have shown. Sure, they are cherry-picking positions and have an incentive to promote their opponents, but the analysis speaks for itself.

That's why dismantling DB was by far IBM's biggest crime. Having it in play would have provided a huge ongoing PR boom for the game. Entering tournaments, playing other computers, analyzing positions, fantastic!

Deep Blue considered 200,000,000 positions per second while Deep Junior considers only(!) 3,000,000 per second. Furthermore, Deep Blue's position evaluation was similarly hardware advantaged. (200/3)^2 > 4444.

I don't doubt that Deep Blue would have benefited from some of the positional evaluation ideas in the current generation of software chess programs, but there is no way they could overcome the overwhelming tactical firepower of a parallel VLSI chess array.

Now post your proof of this conclusion. Being a great calculator doesn't automatically make a better chessplayer. Junior's three million is greater than HIARCS' 20,000 by a larger degree and they play about even.

Also, the 200,000,000 number was for the media. It was rarely reached and far from normal. Parallel eval is certainly great, but it's not a panacea. Donninger of the Hydra project has talked about the difficulties and redundancies.

The bottom line is that unless you can demonstrate chess superiority it's all angels on pins. Saying Deep Blue must have been better because it must have been better is not helpful. Nor is saying that it was better in ways we cannot detect or prove. If you like, I'll agree that Deep Blue was stronger in theory but not in practice! As Amir Ban of Junior said, if it's true that Deep Blue was so superior and yet ended its career without leaving behind a single move to prove it, it was a horrible waste.


You are correct regarding Hsu's offer, IBM was "not interested" and Hsu was acting as a private individual. However, Kasparov did not appear to display any interest. If he had, do you doubt sponsorship would have been found? I feel that both IBM and GK handled this very poorly.

Regarding the strength of the machines, it is impossible to say based on just a few games. You are correct that nodes per second by itself is nearly meaningless. However, you must understand that Intel software based machines must make a tradeoff between quality of position evaluation and number of positions per second since they have general instruction sets capable of very limited parallism. Deep Blue's VLSI core was capable of evaluating hundreds of complex positional factors simultaneously in a single chip - and there were 256 of them in Deep Blue. There really is no other chess machine ever built that is comparable to Deep Blue. Of course, we could build a much better Deep Blue type machine now. However, I don't think one could make a commercial case for it.

Trust me, I had all this stuff beaten into my head by everyone from Frederic Friedel to Frans Morsch to Joel Benjamin. I've read epic posts by Hyatt on why Deep Blue HAS to be better. I'm happy to admit it could have been.

If there's anything we've learned about chess sponsorship over the years, it's that there's no sure thing. Without the IBM PR machine, and with the human Mount Everest already scaled, it's impossible to gauge the interest in another match. I spoke with Garry and his manager Owen Williams during these conversations. In their view Hsu just didn't have anything and it would have been a massive fundraising and PR project with all the risk on one side and no clear timetable. There was no match to be interested in because there was no opponent.

Proof is not possible w/o building another DB style machine. Hydra is trying to solve a more difficult problem. I won't minimize the difficulties of parallization, however, it is much easier to solve such a problem in a specialized domain such as chess.

I am not trying to "prove" anything. However, I am reacting to the assumption that I have seen in nearly every article I have read that Deep Blue has been surpassed. I think many just assume this must be true since 8 years have passed and processors are so much faster now. You and I know better. Fritz at the time was able to duplicate 80% of Deep Blue's moves in a minute and most of the other 20% within 8 hours, however, a handful, 1%, could not be duplicated no matter how much time or hints were given.

It would be fascinating to see a match between the modern machines and Deep Blue. I think the modern machines are better at playing human like chess - and do have some knowledge advantages over Deep Blue as it was in 1996/7. However, I don't think that is enough to match the non-human cobra-like chess Deep Blue played. I also believe that if we were to design a new deep blue like machine with modern technology and experience with the current generation of software, that it would be no contest, not even close. Sigh, unfortunately, "angels on a pinhead"!

All good, although duplication is not the goal unless you hold that Deep Blue's play was ideal and we know this isn't the case, if only because it lost game one and was lost in game four! There are places where today's engines play better. To my knowledge, the only move from the six games that is known to be definitely best that other engines don't play is 36.axb5 in game two.

Chess in practice is quite finite and there are a limited number of decent moves in any given position. With only six games, it would be hard to imagine more than one or two positions where true superiority could be displayed even if it existed. It's just rare to find actual game positions that are both very difficult and contain a clearly superior move. 36.axb5 may be one of these, at least versus 36.Qb6.

Ok, I can't resist: I think one reason GK was not interested was pride. It was barely OK to lose to IBM. However, losing to a relatively unknown engineer/scientist would be risking too much.

I do see that from GK's perspective, that if he did win, people might have claimed that his underfunded opponent was the reason and that a properly funded effort by IBM would have won. I think Hsu said at one point he could build a DB like machine for $50K using a PC as a host. My belief is that GK would have played a stronger opponent than the one that defeated him. However, I know GK could have played a stronger match himself.

Incidentally, when evaluating Deep Blue's games using today's chess programs we better make sure the programs don't already have the Deep Blue games in their databases! If Fritz had game 2 in its database it might be able to come up with axb5. Don't these chess engines do that already, copycat successful moves from games in its databases?

Mig, When are you going to post the next 2 interviews?

Waiting for the third. Thanks for your nice work.

I visited this page first time to get info on people search and found it Very Good Job of acknowledgment and a marvelous source of info......... Thanks Admin! http://www.reverse-phone-look-up.net

Twitter Updates

    Follow me on Twitter



    About this Entry

    This page contains a single entry by Mig published on April 5, 2005 4:47 PM.

    Nakamura Watch was the previous entry in this blog.

    Fischer-Benko in Serbia? is the next entry in this blog.

    Find recent content on the main index or look in the archives to find all content.