LinkBack Thread Tools Display Modes
  #1   Report Post  
Old October 22nd 04, 04:52 PM
Max Murray
Posts: n/a
Default Kramnik-Leko. Game 14. Annotation in Plain Speak.

Domitian: Questions or comments on the annotation? Please email:

World Chess Championship, 2004, Game 14.

Caro-Kann Defense

Kramnik must win to tie the match and retain his World Champion title…

1 e4 …

As is well known and generally accepted by the Free Chess World, the
king's pawn opening gives White the best practical chance of winning
the game (because the KP typically leads to more active positions for
White). Kramnik needs an opening for this critical game that gives him
such active chances, and this is the best first move for achieving
that goal.

1 … c6

Poor Leko will be criticized mercilessly for choosing the Caro-Kann,
and for many, many reasons. Two of the more important knocks: (1) Leko
doesn't habitually play the Caro-Kann, so despite whatever book
knowledge he has on the Caro-Kann – and we can be sure he was booked
up to the eyeballs for this game – he lacks actual experience for the
types of positions that arise. (2) Leko unvailed the Caro-Kann when he
surprised Kramnik with it for Game 3 on September 28 (a 34-move draw).
This final game was played on October 18. That means Kramnik and his
crack seconds had about three weeks to analyze and prepare a line
against the CK.

2 d4 …

Establishing the classic central pawn duo. Kramnik has no reason to
stray from this classical approach and risk an irregular line like,
say, 2 c4, or 2 Nc3, as might might have contemplated trying against a
true-blue Caro-Kann practitioner (to steer the game away from standard
Caro-Kann positions). See criticism of Leko number (1) above.

2 … d5

Black must challenge the classic center immediately. Everything else
is known to be bad. Leko hopes the ultra-solid positions that
typically arise from the CK will compliment his genius for making

3 e5 …

The advance variation reputedly is less drawish because it locks the
central pawn chain (so that fewer pieces are in danger of being
exchanged early, and therefore more pieces are on the board to
complicate things during the middle-game). Almost certainly, Kramnik
was reserving the Advance Variation for deployment exactly for this
type of match situation. See criticism of Leko number (2) above.

3 … Bf5

All us woodshifters know – just like all the proverbial Russian
schoolboys know – that the Caro-Kann allows Black to play a solid
French Defense setup [1 e4 e6 2 d4 d5] without locking in the
dark-squared bishop. The whole point of the CK is to deploy the
dark-squared bishop outside the pawn chain, and that is exactly what
Leko does with the text move.

4 h4 …

4 Nf3 or 4 Nc3 are more commonly seen. The text is not a patzer move,
however, despite damning appearances to the contrary. Rather it is
based on a logical necessity of the advanced variation setup: the pawn
structure d4-e5 virtually compels White to attack on the kingside (in
the direction those two pawns "point"). That is a drawback to the
Advance Variation. White has declared an intention to attack on the
kingside before he has sufficient justification, so now Black has the
opportunity to arrange his defensive setup to counter the looming
attack. In other words, the advance variation tips White's hand. The
text move grabs more space on the kingside to facilitate that attack.

4 … h6?!

This is Leko's first mistake (albeit a tiny-weeny one). The creeping
text move opens an escape hatch on the h7-square for the dark-squared
bishop, for whom Black has structured his whole opening to deploy
outside the pawn chain, and that is good. But the text move doesn't
deal with the problem of space on the Kingside. Even worse, if Black
thematically plays h4-h5, the White kingside pawns become bound.
Better than the text move is 4 … h5, because advancing the h-pawn two
squares addresses all these issues (and also not incidentally because
4 … h5 stops White's next move)...

(I'm braced for arguments about 4 … h6 versus 4 … h5. Let me just say
this little bit more apart from my comments above, and apart from my
own experience as a regular Caro-Kann player telling me intuitively to
push the h-pawn two squares in this position: I wonder if Leko got his
h-pawn play mixed up in his head with the Main Line [1 e4 c6 2 d4 d5 3
Nc3 dxe4 4 Nxd4 Bf5 5 Ng3 Bg6 6 h4] where the one-square push 6 … h6
is now correct and good. See Leko criticism number (1) above. And one
more thing, before anyone thinks that h-pawn affairs aren't really
critical in the Caro-Kann, be advised that Kasparov lost his match
with Deep Blue because he moved the lowly h-pawn at the wrong moment
during the Caro-Kann opening.)

5 g4 …

Move five, and still Kramnik has touched only his pawns! Despite what
horrified chess coaches everywhere might lecture about this sin,
Kramnik's "Pawns First" policy is sound in this position, as I
intimated earlier, because White needs kingside space for the attack
he is obligated to make on that flank. Besides, the text move doesn't
lose time because it "biffs" the bishop, as the Australian chess
writer Cecil Purdy liked to call pawns "tickling" bishops (another
catchy expression).

5 … Bd7

5 … Bh7 is playable, but not advisable: On the h7-square the bishop
will become a target for White's advancing kingside pawns. 5 … Be4 6
f3 Bh7 has the same drawbacks as 5 ... Bd7, without accomplishing
anything significant. So Leko withdraws his light-squared bishop back
behind the pawn chain. What was once a glorious Caro-Kann bishop is
transformed into an ugly French Defense bishop!? (Like a butterfly
somehow transformed back into a caterpillar)

6 Nd2(N) …

Black's forth move was offbeat. This sixth move is entirely new to
master praxis. That is to say, a novelty, as in Never Before Played
(NBP). The most celebrated game from the immediately previous position
is Tal-Pachman, Bled, 1961. In that game, Tal played the thematic 6 h5
and won in 41 moves. Kramnik's move is less thematic; yet that is its
very strength (said with a Nimzowitschian-like twist of logic). This
novelty develops solidly, and so can not be bad, and gets Leko
entirely out of any book knowledge in the very first moves of the
opening. Now Leko, who doesn't play the Caro-Kann, must make good
opening moves entirely on his own wits. (Playing such innocuous but
solid moves against computers early in the opening is also good, for
the same reasons.)

6 … c5

Probably best, at least from the practical standpoint. 6 … Qb6 trying
to hinder the development of White's dark-squared bishop by menacing
the b2-pawn and attacking the undefended e4-pawn leads to nothing:
White can defend with 7 c3, which he was bound to anyway in this type
of position, and then continue developing with an eventual Nd2-Nb3,
etc. 6 … Qc8 has been suggested by some, but that looks and feels
artificial: White can respond 7 Be2, or even 7 f3, and is just fine.
The text-move is a time-honored method for attaining counterplay
against Cara-Kann and French Defense advanced pawn centers.

7 dxc5 …

The reflex is to defend the pawn with 7 Nb3 or 7 c3. The text move has
stragetic advantages over that reflex: the semi-open c-file will make
the queenside unsafe for Black king to castle. And he can't go
kingside either, because that's where White's attack is building. The
position also remains fluid after the pawn exchange.

7 …. e6

7 … Qc7 is playable. Probably, Leko did not want his queen cavorting
after pawns while his kingside pieces ramain at home. For example: 8
f4 Qxc5 9 Bg2 e6 10 Nb3 Qb6 11 Ne2, and Black's development lags
significantly. The text move seeks to recover the pawn and develop the
f8-bishop at the same time.

8 Nb3 …

Kramnik could have used the time Black is expending on rounding up the
c5-pawn to develop with 8 Ngf3, which is reasonable and good on
principle. However, the text move is better than 8 Ngf3 for a specific
reason, as we shall see, and moves that are good for a specific reason
are better than moves that are good on general principle only…

8 … Bxc5
9 Nxc5 …

Yielding the two bishops, which is the specific reason mentioned in
the previous note. 8 … Qc7 might be a slightly better choice than the
text move, although Black still gets his two bishops. For example: 8 …
Qc7 9 f4 Bxc5 10 Nxc5 Qxc5, etc.

9 … Qxa5+
10 c3 …

Black needs to play c2-c3 anyway, so he might as well do so now. For
example: 10 Bd2?! Qxc5 and then the threat of 11 … Qd4, attacking the
e5-pawn and the b2-pawn, forces c2-c3.

10 … Qxc5
11 Nf3 …

White boasts a spacial advantage on the kingside, the two bishops, and
a wonderful d4-square for posting the knight (or bishop as it turns
out!). Black is clearly worse. 11 Be3 is playable, but it drives the
Black Queen to a good square (Qc5-Qc7) where she eyes the potentially
over-extended e5-pawn (note that the e3-bishop also blocks defense of
the e5-pawn down the e-file.) Kramnik prefers to wait before deploying
his dark-squared bishop.

11 … Ne7

More flexible than 11 … Qc7, which tips Black's hand. White then
responds 12 Bd3 Ne7 [or 12 … Nc6] 13 Qe2, and the e5-pawn is safe and

12 Bd3 …

Again, Kramnik prefers waiting to deploying the dark-squared bishop.

12 … Nbc6

The point of all Kramnik's waiting: Black had to develop the king's
knight, and by doing so deprived the Black queen access to the good
c7-square. True, earlier Leko could have played 12 … Qc7. He has
another square planned for her majesty…

13 Be3 Qa5

The threat is 14 … d4, and then 15 Bxd4 Nxd4 16 Nxd4 Qxe5+, or 15 Nxd4
Nxe5 16 Be4 Nd5, is at least equal for Black. Kramnik blocks that
threat with the following expedient…

14 Qd2 …

14 Qe2 was the alternative. It may have been better. The text move
allows Black to try and free himself…

14 … Ng6?!

… But that isn't the freeing move, mister Leko. Black needed to play
14 … d4. The central idea is 15 cxd4 Nb4 [threat NxB] 16 0-0 Qd5
[threat QxN]17 Be2 Bb5, with active play. Leko reportedly rejected 14
… d4 as "unclear," which is certainly true. After all, the game is for
the World Championship, so why risk an unclear position with its
lottery-like outcome (Leko's thinking?).

15 Bd4 …

Perhaps Leko failed to fully appreciate this move when he played Ng6.
True, 15 Bxg6 fxg6 doubles Black's pawns, It also gives active chances
for Black down the f-file after 0-0. Is that where Leko expected to
get his activity, down the f-file after BxN?

15 … Nxd4
16 cxd4 ...

My goodness, Kramnik has given up his two bishops "advantage" to enter
an endgame! Does he really think he can win this position?

16 ... Qxd2+
17 Kxd2 …

White is only slightly better for the endgame (space advantage on the
kingside). Leko must have been confident that his position is more
defendable than it actually turns out to be. Such differences in
evaluation are the essense of chess played between Grandmasters
capable of seeing most everything equally well. That said, what is
most remarkable about this game is not that Kramnik proceeded to win
it, but that Kramnik made winning it look easy against one of the best
defenders in the world!

17 … Nf4

Moving the knight NOT to avoid the doubling of his g-pawns after Bxg6.
No, the exchange of minor pieces leaving a bad bishop against knight
is what Black must avoid. For example: 17 … Rc8 18 Rac1 Bc6 19 Bxg6
fxg6 20 Rc3 0-0 21 Ke3, and Black's bishop is bad: 21 Ke3 Rf7 22 Nd2
Rcf8 23 f3 Rf4 doesn't help Black change the bad bishop dynamic,
despite appearances that Black is making progress.

18 Rac1 …

White now has a lead in development.

18 … h5

Again, 18 … Nxd3 19 Kxd3 Bb5+ 20 Ke3 Kd7 appears to be making progress
for Black but doesn't address White's underlying advantages in this
endgame position: namely, White's kingside space advantage and play
against a bad bishop. So Leko seeks to redress the space advantage
issue with the text move. Ironically, Leko he tries accomplishing this
by putting the h-pawn on the h5-square – the square he should have
played to on move four!?

19 Rhg1 …

A standard maneuver to open a file advantageously for the rook (after
gxh5). That begs a question: Why did Leko invite this obvious manuever
upon himself by playing 18 … h5? Obviously, Leko understands that
opening lines for the side behind in development is bad, and behind in
development Black is. Leko must have felt that White's kingside space
advantage (as defined by the advanced pawns) was the worse of the two
evils. He probably also wanted to develop his h8-rook down the h-file,
since connecting the rooks more conventionally with 0-0 or 0-0-0 is
not feasible. His rook activity on the h-file should, at least
partially, offset the White rook on the g-file. Leko has to seek
activity somewhere.

19 … Bc6

White threatened decisive penetration with his rook: 20 gxh5 Nxd3 Kxd3
Bb5+ 22 Ke3 Kf8 23 Rc7.

20 gxh5 Nxh5
21 b4 …

"Kramnik plays the position with the energy it requires to exploit his
lead in development and the awkward black pieces. Leko never
recovers." –

21 … a6

To stop White expanding further with b4-b5. The alternative, 21 … Kd8
22 b5 Be8, allows the pawn push but limits the queenside weakening of
a7-a6. Still, the text move is better because White can induce the
a7-a6 weakening anyway. For example: 23 Rc3 Rc8 24 Ra3 Ra8 25 b6 a6,

22 a4 Kd8

22 … Bxa4 23 Rc7 is too scary for Black to contemplate. He has to lose
time moving his king to cover the c7-square.

23 Ng5 …

Threating the fork: 24 Nxf7+.

23 … Be8

Leko is like a battered heavyweight boxer backpeddling "on his
bicycle", as the pugilists say.

24 b5 …

The threat is 25 b6. Black has no defense to this threat. At first
blush, 24 … axb5 25 Bxb5 Rf8 appears plausible for Black, but then
simply 26 Rc2 (or 26 Rc3) threating to double rooks on the c-file is
decisive. Black can't afford to tie a whole rook down to defense of
the f7-pawn.

24 … Nf4

Leko wants to play 25 … Nxd3 26 Kxd3 axb5 27 axb5 Bxb5+ except that
it's one move too late for that. So the best he can do is uncover the
rook attack against White's h-pawn.

25 b6 Nxd3

25 … Rc8 26 Rxc8+ Kxc8 27 Rc1+ Bc6 28 Rxc6+! Bxc6 29 Bxa6+ and 30 Nxf7
gives White a winning exchange-sacrifice. Leko must exchange off
White's light-squared bishop first, before trying to close the c-file
(incidentally leaving himself with that bad bishop against a good

26 Kxd3 Rc8

Under no circumstances must White be allowed to plop a rook down on
the c7-square.

27 Rxc8+ Kxc8
28 Rc1+ Bc6

Black has no choice but to abandon the defense of the f7-pawn to block
the rook from the c7-square.

29 Nxf7 Rxh4
30 Nd6+ …

An "octupus" knight. Leko's goose is cooked here, although the winning
procedure for White is not yet entirely clear.

30 …. Kd8
31 Rg1 …

Better than trying to finesse the position with 31 f3. For example: 31
… g5 32 Rg1 Rh5 33 a5 Kd7 Ke3, and progress for White will be slower
than the text move.

31 … Rh3+

31 … Rh7 would ultimately bring Black close to helpless zugswang. Here
is a long variation, not necessarily with the absolute best moves, to
illustrate what I mean: 32 a5 Kd7 33 f3 Kd8 34 Rg6 Kd7 35 Ke3 Ke7 36
Kf4 Kd7 37 Kg5 Ke7 38 f4 Kd7 39 f5 exf5 40 Nxf5, etc. Unplayable is 31
… Bxa4 because 32 Nxb7+ Kc8 30 Rxg7, and Black's position quickly

32 Ke2 Ra3
33 Rxg7 Rxa4
34 f4!! …

The woodshifter move 34 Nxb7+ gives Black good drawing chances. For
example: 34 … Bxb7 35 Rxb7 Rxd4 36 Ra7 Kc8 37 b7+ Kb8 38 Rxa6 Re4+ 39
Kd3 Rxe5=

(The woodshifters watching the game on the Internet (I was one of
them) swooned in admiration and awe after Kramnik's text move appear
on their monitors, and its brilliance began to sink in.)

The exquisite point is that Black is PLAYING FOR MATE.

34 … Ra2+

Kramnik has given us an endgame for the anthologies. Surely,
comparisons will be made to Capablanca-Tartakower, New York, 1924. In
that game, Capablanca similarly maneuvers his rook to the seventh
rank, his king to the sixth, and delivers the denouement, despite
losing pawn after pawn in the meantime.

34 … Rxd4 35 f5 exf5 36 e6 [threating 37 e7+ Kd7 38 e8(Q)+ Kxd6 39
Qe7#] Re4+ 37 Nxe4 fxe4 38 Rc7, and Black can't stop one of White's
two remaining pawns from queening.

35 Kf3 Ra3+

35 … Ra1, threatening a series of checks, is more stubborn. (Perhaps
Leko should have opted for the checking stall to get to the control
and gain more clock time to look for another Houdini-like escape, as
he did in an earlier match game??

36 Kg4 …

f4-f5 is unstopable.

36 … Rd3
37 f5 Rxd4

37 … exf5+ 38 Kxf5 Rxd4 39 e6, and it's mate in seven [!]

38 Kg5 exf5
39 Kf6 …

39 Kxf5 also wins.

39 … Rg4
40 Rc7 Rh4
41 Nf7+

It's mate in two: 41 … Ke8 42 Rc8+ Kd7 43 Rd8#.

Leko resigned.


Questions or comments on the annotation? Please email:

Thread Tools
Display Modes

Posting Rules

Smilies are On
[IMG] code is Off
HTML code is Off
Trackbacks are On
Pingbacks are On
Refbacks are On

Similar Threads
Thread Thread Starter Forum Replies Last Post
Kramnik-Leko. Game 14. Annotation in Plain Speaking. Max Murray (Chess General) 0 October 22nd 04 04:50 PM
Staunton's rich and varied past John Townsend (Chess General) 15 May 24th 04 03:51 AM
Kramnik: Boring chess or better chess? (super GM-chess after Linares 2004) Gregory Topov (Chess Analysis) 28 March 29th 04 11:13 AM
Kramnik: Boring chess or better chess? (super GM-chess after Linares 2004) Gregory Topov (Chess General) 16 March 26th 04 06:22 PM
RGCA: Akopian - Kramnik analysis Andrew Templeton (Chess Analysis) 5 January 17th 04 03:15 AM

All times are GMT +1. The time now is 09:30 PM.

Powered by vBulletin® Copyright ©2000 - 2019, Jelsoft Enterprises Ltd.
Copyright ©2004-2019 ChessBanter.
The comments are property of their posters.

About Us

"It's about Chess"


Copyright © 2017