Did Keres Take A Dive?
Thursday, October 3, 2013 at 2:21AM A few days ago I reviewed Max Euwe's book on the 1948 Match-Tournament for the world championship title that had been vacated by the death of the previous champion, Alexander Alekhine, in 1946. In the new English-language edition of this book Dutch grandmaster Hans Ree writes a foreword, and therein he touches on the question of whether Paul Keres was forced or at least pressured to throw games to Mikhail Botvinnik, who won the tournament and thus the championship title. After noting that the book is silent on the topic, Ree has this to say:
In my opinion, there were signs that something was amiss, especially as may be seen in game 30 [DM: Their game from the end of the third of the five cycles]. There, Keres, normally a fine endgame player, after adjournment, reaches a rook endgame that could be drawn by basically doing nothing, but instead with 50.a4, and the subsequent 53.Rd3 and 54.Ra3, he maneuvers his rook to the most passive position on the board.
In fairness, I must point out that highly qualified observers have considered this atrocity to be just one of those things that can happen to even the greatest players, but I don't agree. Overlooking a mate in one, yes, but not this.
Here is the question before 50.a4
and here is the position after 54.Ra3.
If I recall correctly, the late Larry Parr used to bang on this particular conspiracy theory drum pretty regularly, also appealing to Keres' horrible play in that endgame as evidence as well. Even though I was always far more a Keres fan than a Botvinnik fan I always rejected it out of hand, perhaps due to a general allergy to conspiracy theories.
Now, one reason I've generally rejected conspiracy theories is that they confuse what Richard Swinburne has labeled C-inductive arguments with P-inductive arguments. To elaborate: a C-inductive argument is one where the new evidence confirms a hypothesis, which is to say that it makes it more probable than it was before. A P-inductive argument also does that, and also makes it more probable than its denial. By way of example, suppose Siobhan is a fan of all things Irish, and believes that a leprechaun lives in her attic. One day she hears some noise coming from her attic, and takes this as proof. Is it proof? No. The leprechaun hypothesis isn't rendered more probable than its denial just because there is some noise up in the attic. The hypothesis has received some confirmation, however. If one supposes that there is a leprechaun up there, then one would expect noise sooner or later - perhaps if the little guy got so excited thinking about gold that he didn't pay attention to where he was walking. Unfortunately for Siobhan, mere confirmation isn't proof or even probability, and given the low prior probability of there being a leprechaun in her attic (or anywhere), this only barely budges the probability from its starting point arbitrarily close to zero.
This more or less explains why I've rejected the "fix" hypothesis. Keres's losing the first four games to Botvinnik - especially the horrid ending mentioned above - is consistent with the theory that the games were fixed or at the very least that Keres felt pressured. But is the overall evidence good enough to prove a fix? Not really. For instance: no firsthand or even secondhand evidence has come out to confirm it. Keres had a lousy score against Botvinnik even aside from this tournament, even losing blowout games to him on other occasions (and with both colors). Botvinnik himself fought vehemently against Communist Party officials who tried to have other players throw games to him - and did so at some personal risk. Likewise, Keres has never been accused of throwing games at any other point in his career.
Of course it isn't impossible that Keres took a dive, but aside from the really badly played rook ending and the fact that almost everyone (myself included) likes Keres better than Botvinnik, there really isn't much going for the fix theory. And what about that ending? I think it would make an interesting research project to see how many strong GMs of the time and in the period between the World Wars made horrible errors in rook endings that contemporary players of the same general class would avoid. But let's stick to this ending for now.
I cover this endgame in detail here, first looking at analysis by Euwe, Botvinnik and Keres and then critiquing that analysis with the help of chess engines and tablebases. The executive summary is this:
- 50.a4 is not a bad move
- 54.Ra3 is forced
- 53.Rd3 is losing, but
- only one move (53.Rd5) drew, and that it drew wasn't just obvious,
- that 53.Rd3 (which is an understandable move in its own right) doesn't draw isn't self-evidently obvious,
- and Keres was most likely in time trouble by this stage
- All three analysts - Euwe, Botvinnik and Keres - made significant errors in their print analysis of the games
Given all of the above, we should conclude that it was a complicated ending in which Keres's losing move is entirely explicable, not part of some overall bad plan starting with 50.a4 or worse, some nefarious plot to make sure that Botvinnik won the title. The conspiracy theorists should offer evidence, and absent that evidence should avoid defamatory assertions.
Reader Comments (8)
Forensic, enjoyable and well argued - great post. But conspiracy theories are fun - can you tell us about one (chess related) you agree with?
[DM: There are conspiracies, but when they have sufficient evidence supporting them (e.g. the drawing pact between Petrosian, Geller and Keres in the 1962 Candidates) are they really conspiracy "theories"? Anyhow, I'm not a fan of conspiracy theories. Maybe it's because the people I knew who embraced them when I was a kid seemed a few fries short of a happy meal, or because in many cases they defame a potentially innocent person and ruin his reputation, or because the reasoning supporting the theory was fallacious in the extreme. Of course, not everyone is innocent, but for "golden rule" reasons if nothing else one should be very cautious about making such accusations.]
I think I would have played 50.Ra4 as it's really the only active move that defends the a pawn. Provided that I had enough time to think about it, I could easily make a move like 50.a4 as well if I'm short on time. So 50.a4 doesn't strike me as a suspicious move. Rook endgames are pretty tough and it's easy to make mistakes, this one seems within the realm of possibility even for a GM when considering nerves and time pressure, etc. And since 50.a4 isn't even a bad move objectively, that makes the theory even less believable.
Ha, this old chestnut.
I would like to make a couple of points I have never seen argued anywhere else, but firstly:
"Keres had a lousy score against Botvinnik even aside from this tournament, even losing blowout games to him on other occasions (and with both colors)".
To me, not quite the right impression is given by this statement. Keres' lifetime score against Botvinnik aside from this tournament was +2=9-4. Is this that "lousy"? Not to the extent any rate that is comparable to the +1=0-4 in the WCh tournament. So you might have had in mind Keres' score against Botvinnik at the time the tournament started. This was +0=6-2. Even that is not that "lousy" (for example, Reshevsky was no better, and Smyslov was much worse); and in this case both loses were with white, so "and with both colors" would not apply to this.
But both those loses were indeed "blowouts". The first was the disastrous quick theoretical loss in the first cycle of the 1941 USSR Absolute Championship. The magazine 'CHESS' opined (at the time of Keres death, I think) that this may have sowed the seeds of an inferiority to Botvinnik in Keres' mind. Though it would seem from an opinion piece Keres wrote about 1940 about the top players of his day that he already had a considerable fear of Botvinnik.
Botvinnik does not mention any "psychology" about this game, but does say (in 'Achieving the Aim', I think) that he especially targeted Keres in their Moscow 1947 game for psychological reasons (a long Dutch game he won).
So presumably Keres would have entered the 1948 WCh Tournament in some trepidation of Botvinnik.
Now:
The extreme conspiracy argument goes that Keres threw his first 4 games, then (when it no longer mattered to Botvinnik), Botvinnik threw one back in "compensation" (or perhaps to keep the Soviet Keres abreast of the American Reshevsky - they both finished on the same score).
The first point I would like to make here is that in annotating his win over Botvinnik in 'The Middle Games of Paul Keres', Keres states that just out of the opening "Botvinnik offered me a draw" (which Keres says he declined because the position was interesting). Now, Keres may have been constrained in the truth he could tell about this period, but I don't think he would tell an outright lie. So Botvinnik's offer must have been genuine, which would put the lie to the idea he threw the game. Also, if the game was thrown, why would Keres include it as one of his best games to be annotated?
The second point I would like to make concerns the 4th Botvinnik-Keres game.
This was played in Round 20, and the scores before this round (with only 5 games left for each player, apart from Smyslov who had only 4 left) were:
Botvinnik 11
Smyslov 8.5 (and with the bye in round 20, so his score would not increase)
Keres 7.5
Reshevsky 7.5
Euwe 3.5
So with a scoreline like this, why on earth would the Soviets need Keres to throw his game to Botvinnik (and look suspicious with 4 in a row)? A draw would be perfectly fine; draws would get Botvinnik there [or if there was a problem, the Soviets could order Keres (or Smyslov for that matter) to throw his final game to Botvinnik]. The Soviets may well have been embarrassed that Keres lost this game.
So I think it is extremely unlikely that the 4th or 5th Botvinnik-Keres games were thrown. One cannot be quite so sure about the first 3 games, but as you say there is no evidence. I am inclined to agree with Taylor Kingston that Keres may have been under some sort of vague pressure "not to hinder Botvinnik". This combined with a very possible psychological inferiority could have caused him to play very badly against Botvinnik.
It could possibly be argued that the Soviets had no need for Keres to throw games to Botvinnik as Botvinnik was leading virtually the whole tournament including at the first time he played Keres and every round thereafter. But what if Reshevsky had been leading? Given that it was at the height of the Cold War, would the Soviets have had a 'Plan B' ready? This can only be speculation of course, but one of the varying reasons that Fine gave for not participating in this event was fear of Soviet collusion.
My reaction is that although Keres' moves damn his rook to a passive position, it's reasonable that even top-level players of those days could make such mistakes, especially if in time-trouble. As you point out, they didn't have tablebases, but also they didn't have easy access to the collective endgame wisdom or "intuition" we have today: based on endgames I've seen 50. Ra4 feels right as it maximizes the rook's vertical and horizontal influence, and 50. a4 feels like it fixes the rook on the fourth rank and prevents Ra4.
Just to head something off at the pass: let's not use this post as an opportunity to dredge up other conspiracy theories. I wrote this post to defend Botvinnik and Keres against an accusation besmirching their reputations and accomplishments, not to create a free-for-all to repeat other slanders. Thanks for your cooperation!
If you look at the 59 candidates you will see Keres lost out ot Tal because of an awful loss with White against Fischer, I think Keres perhaps did not cope as well with pressure as others perhaps?
[DM: Keres has been criticized for poor nerves before, but that wasn't really a factor in the 1959 event. The loss of which you speak occurred in round 15 of 28. Here's how he finished the event, starting with round 19:
vs. Benko: Won with Black
vs. Gligoric: Won W
vs. Olafsson: Won B
vs. Fischer: Won B
vs. Smyslov: Lost W
vs. Tal: Won B
vs. Petrosian: Drew B
vs. Benko: Won W
vs. Gligoric: Won B
vs. Olafsson: Lost W - but Tal had already sewn up clear first with a quick draw with a charity draw in a dead won position against Benko straight out of the opening. And if you look at the Keres-Olafsson game, the Icelandic GM came up with a big surprise in the opening and played really well.]
I don't really agree with this last post.
It wasn't just this loss to Fischer that cost Keres the tournament. In fact, he had a mimi-collapse in rounds 15-18: Loss to Fischer, draw with Smyslov. loss to Tal (with white), draw with Petrosian. Just one point out of 4. Meanwhile Tal got 3.5, jumping to a 2 point lead from a half point deficit. This was too big a gap to erase no matter how well Keres subsequently did. It was the mini-collapse that cost Keres the tournament.
It seems to me that Keres was generally unable to cope with the pressure of leading an important tournament. Here at the halfway stage (after round 14), he worried that a half point lead over Tal "was not nearly enough".
It was a recurring pattern.
In the 1962 Candidates, he was leading at the three-quarters stage. Then he couldn't win a single game, finally losing to Benko.
Same in 1956. A lucky win over Bronstein just after halfway put him in the lead. Then not a single further victory, and again a loss in the second last round (this time to Filip).
In the 1948 World Championship tournament, he started well with 2 wins. Then losses to Reshevsky and Botvinnik.
A lucky win over Spassky in the first game of their 1965 Candidates Match was followed by a draw, then 3 losses.
Even before the war this pattern was partially there.
He was winning Semmering-Baden 1937 hands down. Then spoiled it with two losses right near the end and was obliged to not lose his last round game as well in order to be sure of winning the tournament.
At AVRO 1938, he got to within half a point of Fine at the halfway stage. Then didn't win a single further game. Two losses by Fine in the next four rounds put Keres into the lead by default. But still no Keres wins, allowing Fine to catch him. Although Keres won the tournament by tiebreak, surely an outright victory would have strengthened his claim to an Alekhine match - so damage was done.
He was not really secure in his golden period (1950-2) either. In the second last round of the 1950 USSR Ch, he lost to Petrosian, seriously jeopardising victory in that event.
Same in 1951, except this time, Petrosian blew the win.
Keres seemed to lack a sharp cutting edge over whole events. Even his best victories were edgy, nervy affairs, only decided in the last round - he never dominated. With the exception of AVRO, where he scored as many as 11 draws in 14 rounds, he lost two games in them (Semmering-Baden 1937, USSR Ch. 1950, USSR Ch. 1951, Budapest 1952).
It was almost like Keres was afraid of the responsibility of the ultimate title. Every time he got close, he faltered.
Had he got a match with Botvinnik, I don't believe he would have won. Pity, really - he was my favourite chess person.
[DM: It looks like you're going to find a way to claim that Keres always cracks under pressure, no matter when he loses a game, or even fails to win a game(!). Getting involved in such an argument isn't worthwhile. Here are some quick replies, and then I'm going to close this thread down.
First, my point wasn't that he lost the 1959 Candidates because of a loss to Fischer in round 15. I was replying to someone who made it sound as if that loss to Fischer was evidence of a choke, and my point was that a loss in round 15 is hardly evidence of a psychological collapse when it's a 28-round tournament.
Second, I already noted there that there were other occasions when he did seem to crack under pressure - Curacao 1962 being the costliest example of all. Finally, in the Spassky match he was a heavy underdog, but made it quite close, also winning a late game and making the final game very interesting before succumbing. No choke at all; if anything, his nerves were at least as good as Spassky's.]
Hi Dennis, I won't bang on about this forever.
I think you have misinterpreted me to some extent.
When I said "I don't necessarily agree with this last post", I wasn't referring to just you, but also to 'True Chess Fan'. My "mini-collapse" point was in disagreement to what he said, not what you said.
[DM: It's good practice to say "I could have been clearer" than to blame the reader for failing to grasp what wasn't clear. :)]
In general my point was that Keres was not a good LEADER in an important event. He could often be a good CHASER - the 1959 Candidates and the Spassky match being very good examples. Other examples are the 1953 Candidates, and partially the 1948 World Championship Tournament itself where before that fatal Round 15 game with White against Botvinnik, he was 1.5 points behind, and he said had he won (and he claimed he obtained promising chances in that game), he would be but half a point behind "with every chance of ultimate victory".
But it does seem to me that when he was the LEADER, Keres seemed to choke, or at least partially so; or else play so without confidence or verve (ie many uninspiring draws, which allowed others to catch him) that it was virtually an equivalent.
[DM: It's not the same.]
So, yes, I am more or less saying that at the time Keres had the LEAD in the 1959 Candidates, he choked in the next four rounds; and his being in the LEAD (afterwards he was not in the lead, but was CHASING) was the reason.