Friday, April 12, 2013

Decisive moment

One of my favourite old chess books is Luděk Pachman's Decisive Games In Chess History.

Dover, 1975

It's not a masterpiece and nor are the games that it discusses, but that's part of its appeal. All Pachman does is take a historical walk through a lot of matches and tournament, ending with the Spassky-Fischer match - the book was originally published in German in 1972 - and beginning with Baden-Baden 1870, although in his introduction he goes back a little further, to London 1862 and a game between Anderssen and Paulsen.
There is a great difference (writes Pachman in his introduction) between a game played at the beginning of a tournament, when everything is open, and one played in the last two rounds, where a special effort often has to be made to obtain a particular result.
Not all of the games and positions in Pachman's book come from the last two rounds of tournaments, but most do: games played under conditions of great tension where the result is all-important and the prizes may go to whoever blunders least. As Pachman says:
None of the games is technically or aesthetically outstanding; in fact many of them contain bad mistakes.
Which is not to say that Pachman merely produced a collection of blunders, since there is more to the book and the games than that. But it does have its share, including some very famous examples from world championship matches. Chigorin allowing mate in two is the second diagram in the book, and the third comes from the 1960 Tal-Botvinnik match

where Black, instead of blundering into a mate, could have won easily with 39...Ka8.

For our purposes today, though, after the intense and remarkable drama of the last three rounds in London, what's most interesting in Pachman's book is the various games he examines from late stages of Candidates tournaments. A number of these involve Paul Keres, as one might expect given that he individual came second in four different Candidates tournaments and was therefore involved in a number of final-round and penultimate-round dramas himself.

He lost, for instance, to Benkö in the penultimate round in 1962 when a draw would have put him into the last round level-pegging with Petrosian. A win would have given him the last-round lead and maybe he played too enthusiastically for that win knowing that he had to play Fischer in the last round. (As it turned out he could still have tied for first by winning that game, but it was drawn.)

In 1959 the two-horse race between Keres and Tal might have had a different winner had Fischer beaten Tal, as he might well have done, instead of losing their penultimate-round game and thus putting Tal a point ahead of Keres when they might have been level instead. But in 1956 he was, as he would be in 1962, in a position to take the last-round lead, and unlike 1962, had a real, on-the-board opportunity to do so in his game against Filip.

Position after Black's 37th move

Kotov wrote in Think Like A Grandmaster:
White has various ways to win. He could retreat the bishop to b2, while an immediate finish comes from the line 38.Qf6 Nxe5 39.Qxe5. Keres' second said that Keres examined no less* than five ways to win, but chose a sixth possibility 38.Kh2? and after 38...Rc4! 39.Qf6 Nxe5 was a piece down as 40.Qxe5 allows the exchange of queens by Qf4+ when the ending is hopeless for White.**
Batsford, 1978

Pachman omits 1953, where Smyslov was already two points up with two rounds to play, not leaving much scope for late dramas. That was the first of Keres' four second places, but he was also involved in the fight for first place in 1950. Boleslavsky led Bronstein by half a point at the beginning of the round and drew quickly with Ståhlberg, meaning Bronstein had to win to catch him up and bring about a play-off. He had the White pieces against Keres.

Position after 26.Bd2-h6

At this point, Pachman reckons, it should have been all over for Bronstein's winning chances, and hence his chances of playing a match for the world championship:
this hasty move should have deprived Bronstein of all chances of winning, for Black could have replied with the exchange sacrifice...26...Nxb2! 27.Bxf8 Rxf8
He continues:
Black has two pawns for the exchange, which in this case is more than sufficient compensation. White's only hope of holding the game would be 28.Bb3 c4 29.Rxb2! Bxb2 30.Bxc4, after which he would still have to fight hard, though a draw is probable in view of the unlike bishops.
You may find that your chess program doesn't agree with all of this. My PC, coincidentally, blew up while I was analysing Bronstein v Keres and had to spend last Friday night in computer hospital, but I'm told that Stockfish doesn't think Pachman's preferred alternative, 26.Bc1, is actually any better than the 26.Bh6 he criticises (it reckons best play leads to a repetition with either move) though it does prefer 26...Nxb2 to 26...Bg7, which latter move Keres actually played.

Whether White would have retained any winning chances after 26...Nxb2, or for that matter whether he was winning after 26...Bg7 (maybe, maybe not) isn't so important, except when we're evaluating Pachman's analysis. The point was the drama, and the tension and the mistakes. And the outcome, which is that Bronstein won.

But having slipped back in time to 1950 let us scurry forward to the present once again, or at least as far as last week, to London 2013, one of the greatest and most dramatic chess tournaments in all history and one which would have given Pachman more to write about than he could have found space for.

Carlsen losing to Ivanchuk after Kramnik beat Aronian in a position everybody else had given up as drawn. Carlsen getting back what he had lost by beating Radjabov very late in the day after Kramnik might, or might not, have had a chance to beat Gelfand just before the time control. Then that barmy last round, with both the leaders seeming to spend as much times watching their rivals play as they did sitting at their own board: both having to play for a win, neither able to take any draw on offer unless and until they were sure of the other's result.

Pachman would have loved it. I confess that, as a Kramnik supporter, I thought his chance had probably gone the previous night, and though I was half-watching the games, the state of Kramnik's position didn't encourage me to pay close attention until Carlsen's clock position and his kingside both appeared to collapse. Even then, as Kramnik seemed already to be losing, it didn't seem quite as dramatic to me as it really was.

The analysis engine on the official site was struggling to keep up with play and my PC, even when fit, is hopelessly slow. So, although I had my program running upstairs, it was telling me nothing about the state of Ivanchuk-Kramnik that I didn't already think, while downstairs, the official site was still telling me what the state of play had been several moves previously. I gathered that it had since gone badly wrong for Carlsen, but by that stage Kramnik's position looked at least as bad as Carlsen's, and of course, he resigned not very long after the time control. Exciting, but not, I thought, as exciting as it might have been if Kramnik had ever had a real chance.

Or so I thought, right up until about noon the next day, when my copy of Chess Today's email newsletter arrived, claiming that Kramnik had, in fact, had a draw available to him.

According to Chess Today #4528:
Kramnik could have forced a draw: 35...Rxa6 36.Rxa6 Nf4+!! 37.Bxf4 exf4 38.Bxf7+ Kh8! 39.Qd3! (39.Bxe8?? Qd2+ is even winning for Black: 40.Kf1 Bd4) 39...Qxh4! 40.Bxe8 Qg3+ 41.Kf1 Qh3+ with a perpetual check.
Well well. There was plenty of comment during and after the game that Kramnik's 35...Rc8? was an error, but nothing - not that I saw - suggesting it was an error of that magnitude, that rather than being able to keep a losing game temporarily alive, he had actually missing a forced draw. A forced draw that, given that Carlsen lost, would have won him the tournament and a match for the world title.

But was that the situation? Even had he seen it, Kramnik couldn't have taken a draw, could he? Not as far as he knew. He had to keep playing for a win unless and until Carlsen didn't have a probable draw open to him.

But that's the question. At the moment that Ivanchuk played 35.Rxa6? - or rather, between that moment and the moment Kramnik replied - how did Carlsen stand?

Kramnik-Ivanchuk was running a few moves ahead of Carlsen-Svidler, in which game the turning point came after Svidler's 30th move.

This, commentators seem to agree, is where Carlsen lost it, with his 31.f3? losing, to all intents and purposes, to 31..Nf4! as Svidler played. Instead, Carlsen should have played 31.Bd5! and after 31...Bxd5 32.Qxc5+ Kg7 33.Qxd5 Kxh6 34.Qxf7 with the threat of 35.Qe7 if Black plays 34...Rc8 (Chessbase). So best seems to be 34...Ba5 after which 35.Qxe8 Bxe1 36.Qxe5 Bxf2+ 37.Kxf2 Qxh2+.

I don't know if Carlsen is winning that, but I do know that only he can win. He would obviously have taken this variation over the game line: I assume he didn't see it.

At any rate, it seems that prior to playing his 31st move, he had a draw available to him. After playing it, however, he did not.

So, goes my train of thought:
  • prior to Carlsen's 31.f3, Kramnik's 35....Rxa6 and 36...Nf4+, drawing, was a line he maybe couldn't have risked playing, even had he seen it;
  • after Carlsen's 31.f3, Kramnik, had he seen it, could have played the drawing line. It would have won him the tournament.

But did he actually have that opportunity? Which came first, Carlsen's blunder, or Kramnik's? I don't know. I'm not sure anybody knows.

I wasn't really following the commentary at the time (until a bit later than this, I was mostly listening to Bach on Radio Three) but I've watched the crucial section since.

We can't, from this video anyway, tell which move was played first. If there's a way of finding out, I'd like to know it.

Maybe it doesn't matter: I don't think Carlsen saw 31.Bd5! anyway. Maybe - more than maybe - Kramnik, had he seen his opportunity, should have taken the draw. We're probably not talking about moves players saw and passed up, we're talking about moves they didn't seem to see.

But still, what intrigues me is whether or not the moment was there. After Kramnik and Carlsen had both played nearly four hours under the impression that only a win would do. After Kramnik in particular had seen nothing in any of his visits to Carlsen's board to tell him that a draw might be enough for him. After both found out that everything had changed only when it was already too late - had there, in fact, been an in-between moment? A moment when everything had changed, when it wasn't too late for Kramnik, when it would have been too late for Carlsen?

Maybe there wasn't. Maybe we'll never know. But maybe there was that decisive moment. If only a moment. And besides, whatever was happening on Carlsen's board, Kramnik seems to have had that draw in his hand. For a moment. Then it was gone.

What I'm saying is that now I know it might have been there, I'd like to have known at the time. This is the mystery, the fascination, but also the curse of chess, that you do not really know what is going on, even who is winning or who is ahead, unless somebody tells you. And sometimes you don't want to know, you want to be in the same position as the players, looking at a board which hides nothing from you but which reveals only what you yourself can see. But sometimes, you just have to know.

Perhaps it's a tradeoff between drama and mystery, but sometimes you need the drama of knowing that something's changed, that's something's gone terrribly wrong. Like the crucial moment in Aronian-Kramnik.

Moment of madness

After showing 0.00 for about an hour, suddenly the online analysis engine went mad after 50.g6?? - and so, as a result, did the viewer.

So what I'm saying is, in a way, that I wish the online engine had been quicker off the mark, or that mine had, because that way, I'd have known. I understand, at the same time, that not knowing is part of chess, part of its very nature, just as the goal that shouldn't have stood is an intrinsic part of football. Pachman didn't know for sure what had happened in Bronstein-Keres, and that was twenty-two years down the track. That's the mystery. The mystery is worth something. We don't have to know what's happening right now. Even if it means we sometimes miss the moment.

But in this instance - what a moment it would have been. I'd have skipped the mystery to catch the moment. That moment would have been every bit as dramatic than anything I read in Pachman. Maybe more so. Maybe as dramatic as anything I can think of that has ever happened in chess.

[* This should be "fewer", a rare lapse from the translator, Bernard Cafferty.]
[** Talking of translation, I've translated into algebraic from the descriptive original, as I have also done with Pachman. By "original" I mean my edition in both cases - I assume the actual originals were indeed in algebraic.]
[Thanks to Angus French and Mark Crowther.]


Anonymous said...

Hard work to figure it out in numbers if you had to go through move by move, but the new presentation of chesscasting includes the time spent on each move, so adding these up gives the real time sequence.


Jonathan B said...

I've seen an interview with Kramnik where Vlad says he saw ... Rxa6 but couldn't play it so gambled knowing what he was playing was rubbish.

I certainly remember he paused for a while before playing ... Rc8.

I doubt Carlsen saw too much when he played f2-f3. If the clocks on the transmission feed were accurate he was down to seconds for a dozen or so moves.

But, yes ... what a moment.

Jonathan B said...

I remember watching some panel game show or other where Brian Sewell picked up Chris Addison for his inappropriate use of "less".

"I'm terribly sorry", came the reply, "In future I shall try to do it fewer"

AngusF said...

RdC wrote: "Hard work to figure it out in numbers if you had to go through move by move, but the new presentation of chesscasting includes the time spent on each move, so adding these up gives the real time sequence. these up gives the real time sequence."

You'd think so but does it? Jonathan and I were watching the games one day, assisted by a laptop and a tablet, both running the chesscasting software. Curiously different timings were shown for the same game.

ejh said...

It's Friday evening in a little village above Oviedo.

I can add up the timings, or I can go for a drink.

Blue pill or red pill?

Niall said...


Drink every time.

You have bars where you live?!!

Anonymous said...

Keres-Filip has a black pawn on a6. Otherwise the ending is by no means hopeless (except in a tournament sense).

ejh said...

Thanks, I've corrected the diagram.

You have bars where you live?

In the village where I live, no. But where I am staying while working in Oviedo, yes....

DRS said...

Excellent piece, Justin.

Even though it's only two weeks ago, my memory for the details has already faded.

However, for what it's worth, I don't remember Kramnik checking Carlsen's position between the latter playing 31. f3 and the games reaching the time control.

The videos may prove me completely wrong, of course

David Sedgwick
Deputy Chief Arbiter
Candidates Tournament 2013

DRS said...

Excellent piece, Justin.

Even though it's only two weeks ago, my memory for the details has already faded.

However, for what it's worth, I don't remember Kramnik checking Carlsen's position between the latter playing 31. f3 and the games reaching the time control.

The videos may prove me completely wrong, of course

David Sedgwick
Deputy Chief Arbiter
Candidates Tournament 2013

Unknown said...

It is possible, though a bit laborious, to work out the relative timings of the two moves by Carlsen (31.f3) and Kramnik (35...Rc8).

If you go to the video footage archived here:

Select part 3 from those clips from Round 14, and then start viewing at around 1 hour 30 minutes into that clip. Both moves are played within about a minute of each other.

Kramnik plays 35...Rc8 at 1:32:42, when the main screen is showing him playing Ivanchuk. Ivanchuk, a few seconds later, plays 36.Rh1. The Chess Casting displays claim that Kramnik had just under 10 minutes at this point, and Ivanchuk had around 2 minutes 40 seconds.

The point where Carlsen plays 31.f3 is less than a minute later. The main screen still shows Kramnik and Ivanchuk. To see Carlsen and Svidler, you have to look at the inset in the upper right-hand corner of the screen that shows the whole playing arena.

Carlsen and Svidler exchange two moves at around 1:33:30. Svidler plays a long move with his queen (30...Qh3) and Carlsen instantly replies 31.f3. At this point, Svidler had over 10 minutes and Carlsen 1 minute 18 seconds (according to the Chess Casting display that appears a minute or so later on the main display).

So it seems that Kramnik played 35...Rc8 around 47 seconds before Carlsen played 31.f3. If Kramnik had waited a bit longer, he might have then played 35...Rxa6! safe in the knowledge that a draw was all he needed. But who knows what he was thinking by that stage. Their nerves must have been shredded!

ejh said...

Christopher - thanks ever so much for that.