Blunders in chess – Kramnik wasn't the first

The chess duel Man vs Machine, Vladimir Kramnik vs Deep Fritz is being staged from November 25 until December 5th. It is sponsored by the RAG AG, one of Europe's largest energy companies. The venue is the National Art Gallery in Bonn, Germany. Schedule:

There is full live coverage on the Playchess server, as well as on the official site and a number of partner sites.

Game two – the missed mate

After the mate-in-one blunder by Vladimir Kramnik in game two of his match against Deep Fritz 10 in the Man vs Machine match, and our attempted explanation of the mechanics of the blunder, we have received a number of interesting letters on the subject. We bring you a selection, and then a small collection of similarly inexplicable blunders by world champions or world championship candidates. We do this with apologies to Vladimir Kramnik, who has probably heard enough about the subject.

First the readers' feedback

Yannick Roy of Montréal, Canada
Interesting but not entirely satisfactory answer. Of course it is true that the knight protecting the h7 square being on the 8th rank is a rare configuration, seen more often in composed problems than in actual games, but still... It's quite simple to see, and I'm sure it takes less than the fraction of a microsecond for a player of Kramnik's calibre to spot it.

Somehow I feel a more general explanation, linked to the type of stress a match against a computer puts on a human player, might be in order. Being on the razor's edge for a few hours, calculating variations endlessly, probably makes the pieces dance in front of one's eyes. The only other type of explanation I can think of would be of a psychoanalytical nature; maybe somehow, subconsciously, Mr. Kramnik wanted to lose the game, or even embarrass himself in front of the chess world, out of repressed guilt or something. But of course that will have to be dealt with between Mr. Kramnik himself and his analyst, if he has one (or if he seeks one after this blunder). I can't help but notice, however, that he also made a blunder (although not nearly as spectacular as this one) during his 2002 match in Bahrain, just around move 35. Is it something with that number? Is that what it's all about? Who knows. One thing at least is clear; the human brain remains, for better or for worse, much more complicated than a silicon chip.

John Rood, Holbrook, MA, USA
I liked your "cognitive psychology" explanation of Kramnik's blunder. I found it convincing. I wish to add one more factor that might have contributed to missing the mate. The knight had just arrived at the key square by a very natural, apparently innocuous, manner, namely the capture of a piece in an exchange of material. Thus Kramnik might not have had any "suspicions" of its position being crucial for a hidden mate.

Andreas, Thyrhaug, Stockholm, Sweden
How about letting people vote, hopefully honestly, if they missed 35.Qh7 mate, or not, before it was executed on the board? I will vote "No, I honestly did not see 35.Qh7 mate before it was executed on the board." Like Kramnik and many others I was also looking at the pawns on the queenside. I just froze to ice when 35.Qh7 mate was played. I thought something must have gone wrong with the input on the official web site, and that this position was not actually on the real board. I think the statistics of how many people missed it would be interesting! The result of the game is not important. The widened insight into how the human brain recognize patterns was worth more than a million dollars!

Zak Seidov, Ashkelon, Israel
I like your comment on this blunder of the millenium and the explanation given by Alexander Roshal, who told us that the mating pattern that occurred during the game, with the white queen protected by a knight on f8 is extremely rare in chess. I must say that being some 2400 (Playchess blitz Elo) player I didn't see mate during the transmission! But my explanation is not that the "mate construction" was unusual or rare in chess. The problem was that during the last (rather forced) several moves before mate White wasn't at all trying or threatening to mate, but rather to achieve a draw! The move 34.Nxf8 was not aiming at mate but was the only move to restore material balance. And this was in minds of all human beings including Vlad and me...

Tobias Nordquist, Sandviken, Sweden
I had a similar experience. In a blitz game I was white and had in a wild position taken a bishop on g4 with Ne5xg4. My move was a little bit clumsy and the piece landed a little bit betwen g4 and g5. Anyway it was a blitz game swop swop, and in my mind the Knight was on g5. At some moment I played Qd3, and then Qh7 mate (I was sure). My oponent looked at me and then at the knight betwen g4 and g5. He "corrected" it to g5, saying "I didn't see the knight". I have thought about this incident many times, and was reminded of it by Kramnik's blunder.

Kyle Morrison, Waterloo, Canada
In the article regarding how Kramnik could have missed the mate in 1, it was mentioned as an example of chess pattern blindness that GM Yasser Seirawan missed it as well. It was said in the article that Seirawan, after ...Qe3, started discussing White's possibilies, all except mate in 1. This is in fact not true. I was listening to the live broadcast, and before the move had even been relayed to the net (both the ChessBase and Spiegel online sites), Seirawan was shouting into the microphone "Oh my God, oh my God, Kramnik has blundered!" This is just a point of information, that Seirawan spotted it right away.

Andrew Pressburger, Toronto, Canada
Because in the entirely fortuiotus realm of human affairs we cannot know what happens from one minute to the next. Our existence is merely a series of hopeful yet blind guesses. So much for Aristotelian logic. Bah, humbug!

Brian Carson, Toronto
When we watch Kramnick play, we see a man confident that he can play as capably and subtly as the machine. In his own mind he is confident that he has proven his skill as equal or better. Was his slip overlooking a mate in one in Bonn, just simply proof that men get tired and machines do not. Or is this Kramnik's way of fighting to beat the machine in the next game as white and keep it all terribly dramatic, or conversely, does this just prove the global and commonly held the belief that all this is rigged to assure big sales of the ChessBase product line? I dare you to print this legitimate question.

Mark Baldock, London, UK
Your news report provided a picture of the stage including the wide screen installed in the auditorium. The screen detailed the variations Fritz 10 was considering at that particular time along with its evaluation. Please forgive me in case I am missing something, but what is to stop Kramnik just looking at the screen to see what Fritz 10 is thinking and play accordingly?

Levon Bagramian, New York, NY
I've been reading your wonderful news section for a long time now, and on some items I always felt the need to comment or add a newsworthy bit of info. This need became most apparent with the latest sensation: Kramink's overlooking the mate in one. Would it be too much to add a section under every news item for comments? Most respected news sites have it nowadays. I believe it'll add a nice level of interaction to your pages. And I believe your readership is mature and concerned enough to not allow that feature to turn into a rant-all-you-can section.

Arnel De Castro, Iloilo
The recent 2006 Man versus Machine match is very interesting. However, the computer doesn't get tired and is very consistent in its calculations, all through out the match. The blunder of Kramnik in game two was due to stress during his calculation of his previous moves. Perhaps in the future matches of Man vs Machine, humans should be given one take-back move for every game to compensate this weaknesses. It would be interesting also if you feature an article with insights from psychologists or other medical experts why humans buckle under too much pressure against a computer in a chess match. And – thank you for bringing Chess news, ChessBase!

Martin Raschti, Germany
Regarding Kramnik's ... Qe3 error I'd like to point you to the funny coincidence that the second game (!) of the Kasparov vs. Deep Blue match in 1997 also famously featured a human mistake based on the move ...Qe3 – which would've saved Kasparov through perpetual check, when he instead resigned.

Stig Martinsen, Bergen, Norway
Alexander Roshal's explanation in terms of the missing pattern (rare with the knight on f8) sounds right. However, if the mate had been White's only idea in the position, I think Kramnik would have seen it immediately. When Kramnik saw the perpetual check idea several moves before, he must have prematurely and subconsciously have decided that this was White's only idea (a very common cause of blunders, in fact). Perhaps he was also influenced by the perception that White is fighting for a draw, not for a win. It is with some Schadenfreude an amateur observes that even World Champions could benefit from following Blumenfeld's Rule: before moving, stop and look at the board afresh, "through the eyes of a patzer".

Julio Mendoza-Medina, Kentucky, USA
I have to disagree with Mr. Alexander Roshal's explanation. Although it sounds convincing, I think that GMs (and Vladimir Kramnik is a very strong one) are used to see patterns one way or the other way around. That story of the knight on an unusual position is just funny. What sometimes happens – please remember we are humans – is one of these (or both):

1. Kramnik simply overlooked the mate threat because he was very busy with his plans on the queenside. Period. He did not even consider in his mind what was White's threat.

2. Kramnik forgot to make the move to take care of the mate threat first. Often times you consider or see the threat, and you know you have to make that very move that stops it, but unexpectedly you make the move you had in your mind to do next.

Janis, Nisii, Rome, Italy
Alexander Roshal's theory on the reasons why Kramnik overlooked a mate in one during his game two against Fritz 10 is very interesting, and also somehow convincing. However, it seems to me that Mr. Roshal got a bit carried away by his own reasoning. I don't reckon that the mate pattern that occurred during the game could be considered as 'unusual' or 'weird', let alone the last two screen shots that shows two patterns that, according to Roshal, would have recognised as mating patterns. I do believe that Kramink wouldn't never have played 34...Qe3 in those positions, but not on account of the mating pattern, rather because of the hanging knight!

IM John-Paul Wallace, London England
I read with interest the opinions of how Kramnik could allow a mate in one. In my opinion it is most likely a combination of factors. As Roshal says a Nf8 and Qh7 mate is rare, but I think it is more than that. Firstly I would say that for sure if Fritz did not take something on f8, if he just moved his knight there, then Kramnik would see the mate instantly. He would ask the purpose of the move. As it was he simply thought the purpose was recapturing a piece. Also, psychologically, Kramnik had been pressing in this game, and this clouds one's human mind, one loses some objectivity. Well done on a great website!

Louis Morin, Montreal, Canada
If the computer had played Nf8 (a quite unusual move for a white knight), a bell would have rang in Kramnik's head and the mating threat would have been very obvious. But since the computer played N takes f8 (a very obvious and predictable move, capturing the unprotected rook to maintain material equality), no bell rang in Kramnik's head and the mating threat went completely unnoticed.

Jim Thompson, Helton, USA
Monday's one-mover by Fritz over Kramnik made me wonder how many world champions have been mated during their reigns. Counting the 14 from Steinitz to Kramnik, here is what I found, using Fritz 9: Steinitz, Lasker, Capablanca, Alekhine, Botvinnik, Smyslov, Tal, Petrosian, Spassky and Karpov were never mated in a serious game, while they were world champions. Neither was Fischer, but then he never played a game while officially world champion!

Max Euwe is the only world champion to have lost a serious tournament game by mate. That was in 1937, when he blundered in a lost position in the fifth round of the Bad Naumheim tournament against Bogoljubow. Kasparov was mated in a blitz game by Kiril Georgiev in the World Championship Blitz at Saint John in 1988. He was never mated in a game at classic time controls. In December 2001, in the Botvinnik Memorial 5 minute blitz tournament, Kramnik was mated -- by Kasparov! And that is the list, as far as I could determine.

Lola Ice, Boulder, Colorado
I would like to know how many times in his chess career has Kramnik been mated. In general how often do Grandmasters fall prey to mate? If it is possible please mention these stats in a future article concerning this most unusual loss.

GM John Nunn, London, GB
I don't know if anyone has pointed it out, but the missed mate in 1 bears some resemblance to the following game:

Wells,Peter K (2480) - Areshchenko,Alexander (2640) [E06]
Monarch Assurance Isle of Man (4), 2006
1.d4 d5 2.Nf3 Nf6 3.c4 c6 4.Nc3 e6 5.g3 Be7 6.Bg2 0-0 7.0-0 b6 8.Ne5 Bb7 9.e4 dxc4 10.Nxc4 Ba6 11.b3 b5 12.Ne3 b4 13.Ne2 Nbd7 14.Bb2 Nb6 15.Re1 Rc8 16.Nf4 Re8 17.h4 Bb7 18.Ng4 Nbd7 19.d5 cxd5 20.exd5 exd5 21.Bxd5 Bxd5 22.Nxd5 Nxd5 23.Qxd5 Nc5 24.Qf5 Qd3 25.Qf4 h5 26.Nh6+ gxh6 27.Rxe7 Rxe7 28.Qf6 Qh7 29.Qxe7 Nd3 30.Qd7 Ra8 31.Bd4 Qe4 32.Qd6

32...Ne1 33.Bb2 Qg2# 0-1.

Famous blunders

The following examples were collected by Johannes Fischer, chess historian and writer.

Staunton,Howard - Anderssen,Adolf [C54]
London knockout London (3.4), 1851
1.e4 e5 2.Nf3 Nc6 3.Bc4 Bc5 4.c3 Nf6 5.d4 exd4 6.e5 d5 7.Bb5 Ne4 8.cxd4 Bb4+ 9.Nbd2 0-0 10.0-0 Bg4 11.Bxc6 bxc6 12.Qc2 Bxf3 13.Nxf3 Rb8 14.Qxc6 Rb6 15.Qc2 f5 16.a3 Be7 17.b4 f4 18.Ne1 Rh6 19.f3 Ng5 20.Nd3 Ne6 21.Bb2 Qe8 22.Rac1 Qh5 23.h3 Rg6 24.Nf2 Rg3 25.Kh2 Rf5 26.Qc6 Qg6 27.Rg1 Rfg5 28.Ng4 h5 29.Nf6+

After the last move Staunton was actually lost, since after 29...Bxf6 30.Qxe6+ Kh7 White cannot defend the g2 square. But Adolf Anderssen played 29...Kf7?? One of the greatest masters of 19th century chess, renown for his sparkling, tactical play, had overlooked 30.Qe8#.

Steinitz,Wilhelm - HG Voight [B29]
Philadelphia, 01.12.1885
1.e4 c5 2.Nc3 Nc6 3.Nf3 Nf6 4.e5 Ng8 5.d4 cxd4 6.Nxd4 e6 7.Ne4 d5 8.exd6 Bxd6 9.Bb5 Qd7 10.c3 Bb8 11.Nc5 Qc7

One year before becoming the first official world champion in the history of the game Wilhelm Steinitz, in the above position, decided to bring his king into "safety": 12.0-0?? Unfortunately he had overlooked the long diagonal threat of Black's previous moves. The result: 12...Qxh2# 0-1.

Petrosian,Tigran V - Bronstein,David I [E65]
Candidates Tournament Amsterdam (2), 1956
1.c4 Nf6 2.Nc3 g6 3.g3 Bg7 4.Bg2 0-0 5.Nf3 c5 6.0-0 Nc6 7.d4 d6 8.dxc5 dxc5 9.Be3 Nd7 10.Qc1 Nd4 11.Rd1 e5 12.Bh6 Qa5 13.Bxg7 Kxg7 14.Kh1 Rb8 15.Nd2 a6 16.e3 Ne6 17.a4 h5 18.h4 f5 19.Nd5 Kh7 20.b3 Rf7 21.Nf3 Qd8 22.Qc3 Qh8 23.e4 fxe4 24.Nd2 Qg7 25.Nxe4 Kh8 26.Rd2 Rf8 27.a5 Nd4 28.b4 cxb4 29.Qxb4 Nf5 30.Rad1 Nd4 31.Re1 Nc6 32.Qa3 Nd4 33.Rb2 Nc6 34.Reb1 Nd4 35.Qd6 Nf5

Tigran Petrosian, world champion from 1963 to 1969, was one of the most solid players the game has known. In this game he was not in time trouble, but calmly played 36.Ng5?? losing his queen in one move: 36...Nxd6 0-1.

Huebner,Robert (2600) - Korchnoi,Viktor (2695) [C09]
Candidates final Merano (7), 1980
1.e4 e6 2.d4 d5 3.Nd2 c5 4.exd5 exd5 5.Bb5+ Nc6 6.Ngf3 cxd4 7.Qe2+ Qe7 8.Nxd4 Qxe2+ 9.Kxe2 Bd7 10.N2f3 Nf6 11.Re1 Nxd4+ 12.Nxd4 0-0-0 13.Bd3 Bc5 14.c3 Rde8+ 15.Kf1 Rxe1+ 16.Kxe1 Ng4 17.Bf4 f6 18.Rd1 Ne5 19.Bc2 Bxd4 20.Rxd4 Be6 21.f3 h6 22.Kf2 Rd8 23.Rd1 Kd7 24.Re1 Nc4 25.h4 a6 26.Bg6 Rc8 27.g4 Rc6 28.Re2 b5 29.h5 a5 30.Bd3 Rb6 31.b3 Nd6 32.Rc2 Nf7 33.Be2 Ne5 34.Rd2 Kc6 35.Kg3 Rb7 36.Be3 Nf7 37.Bd4 Rb8 38.Bf1 Kc7 39.Kf4 Nd8 40.Bd3 Bd7 41.Kg3 Ne6 42.Bf2 a4 43.Bc2 axb3 44.axb3 Kd6 45.Be4 Bc6 46.Be3 Ra8 47.c4 bxc4 48.bxc4 d4 49.Bxd4 Bxe4 50.Bxf6+ Kc5 51.Be7+ Kxc4 52.fxe4 Ra7 53.Bd6 Ng5 54.Kf4 Rf7+ 55.Ke3 Rf3+ 56.Ke2 Rf7 57.Be5 Nf3 58.Rc2+ Kb5 59.Ba1 Nh2 60.Kd3 Nxg4 61.Rg2 Rf3+ 62.Kd4 Rf4

In the above position Germany's top grandmaster and four times world championship candidate thought for a long time and then played 63.Kd5?? Hübner, who at the time was number three in the world (after Karpov and Korchnoi) overlooked the simple fork 63...Ne3+ 0-1. It was the seventh game of sixteen to be playind the Candidates Match in Mirano, and Hübner had been leading 2:1. He lost game eight as well, and retired from the match with the next two games adjourned (both were awarded to Korchnoi). Shortly thereafter the above position was the subject of a very cruel title picture of the German TV guide Hör Zu. However this is probably lost to posterity.

Christiansen,Larry Mark (2620) - Karpov,Anatoly (2725) [E12]
Hoogovens Wijk aan Zee (2), 01.1993
1.d4 Nf6 2.c4 e6 3.Nf3 b6 4.a3 Ba6 5.Qc2 Bb7 6.Nc3 c5 7.e4 cxd4 8.Nxd4 Nc6 9.Nxc6 Bxc6 10.Bf4 Nh5 11.Be3 Bd6

The twelfth world champion and one of the greatest tournament players of all times overlooked a simple (but long-distant) fork: 12.Qd1 attacking the bishop and the knight on h5. Karpov resigned immediately.

Links

• Official match site
• All ChessBase reports on Kramnik-Deep Fritz