littlefermat: 1 Nf3 d5 2 d4 Nf6 3 c4 e6 4 Nc3 dxc4 5 e4 Bb4 6 Bg5 c5 7 Bxc4
cxd4 8 Nxd4 Bxc3+ 9 bxc3 Qa5 10 Bb5+ Nbd7
<10...Bd7 leads to a rather unpleasant position for Black. After 11
Bxf6 gxf6 12 Qb3 a6 13 Be2 Nc6 14 0-0 Qc7 15 Rab1 (15 Qa3!?;
15 Rfd1) 15...Na5 16 Qa3 I had already experienced this for
myself ( Khenkin vs Kramnik, 1989 )>
11 Bxf6 Qxc3+ 12 Kf1 gxf6 13 h4 a6
<Very dangerous is 13...Ke7?! 14 Rh3 Qa5 15 Rb1 Rd8 16 Qc1!,
when there is literally nothing that Black can move (Adorjan vs A Chernin, 1990).>
14 Rh3 Qa5 15 Be2 Nc5?!
<15...Ke7 is stronger, although after 16 Rc1 Rd8 17 Qc2 White, in
my opinion, has more than sufficient compensation for the pawn.>
16 Nb3!
<Exchanging Black's only active piece is even more advantageous,
for the reason that the white knight at d4 is not too well placed...>
16...Nxb3 17 Qxb3 e5
<At that time this was a new move - and an unsuccessful one. It is
true that, in order to refute it, at the board I had to find several
precise and difficult moves. Earlier 17...Qc7 18 Rd1 Bd7 was
tried, but after 19 Qb2! Black had serious difficulties.>
18 Rf3
<The only way of maintaining the initiative and of forestalling the
opponent's clear plan: ...Qc7(d8)-e7 and ...Be6. Instead 18 Rc3
looked tempting, but then 18...Be6! would have fully equalised,
e.g. 19 Qxb7 Qxc3 20 Qxa8+ Ke7 21 Qb7+ Kf8 22 Rd1 Bxa2 23
Qxa6 Be6.>
18...Qd8?!
<18...Ke7 was dangerous on account of 19 Rc1 followed by Rfc3,
but in the light of subsequent events perhaps Black should have
gone into an ending: 18...Be6 19 Qxb7 0-0 20 Qe7 Qd8 21 Qxf6
(21 Qc5!?) 21...Qxf6 22 Rxf6 Rfd8, where he has certain drawing
chances.>
19 Rc1!
<This move was underestimated by my opponent. Now it is not
apparent how he can complete the development of his pieces. I will
give a few variations: (a) 19...Bg4?! 20 Rd3 Bxe2+ 21 Kxe2 Qe7
22 Qb6 Rd8 (22...Qe6 23 Qxe6+ fxe6 24 Rc7) 23 Rc7 and wins;
(b) 19...Be6 20 Qxb7 0-0 21 Rd1 Qe8 22 Rxf6 and White has a
decisive advantage; (c) 19...b5 20 Rc6! Be6 21 Qa3 (21 Rxe6+?
fxe6 22 Qxe6+ Qe7) 21...Qe7 22 Bxb5! Qxa3 23 Rxa6+ Ke7 24
Rfxa3 with a positional and material advantage; (d) 19...0-0 (the
most critical, but also insufficient) 20 Qe3 Kh8 21 Qh6 Rg8
(21...Be6 22 Rd3 Qe7 23 Rc7!) 22 Rd3 Qe7 (better is 22...Qf8 23
Qxf6+ Qg7, when Black faces a difficult defence in the endgame)
23 Rc7! Qe6 24 Rd6! Qxd6 25 Rxf7.
Black chooses another alternative, which is little better.>
19...Qe7 20 Qb6 Qd8?!
<Reluctantly played. Kaidanov's suggestion of 20...Bg4 21 Rc7 Qe6
would have lost to 22 Qxb7 Rd8 23 Bc4! Bxf3 24 gxf3 Qh3+
(24...Qd6 25 Bxf7+ Kf8 26 Bd5) 25 Ke2 0-0 26 Rxf7, but the
lesser evil was 20...Qe6 21 Rxf6 Qxb6 22 Rxb6 Be6, with some
chances of saving the game.>
21 Rc7!
<More energetic than 21 Rxf6.>
21...Qd4 22 Re7+!
<Simpler than 22 Qxd4 exd4 23 Bc4, which, however, was also
quite strong.>
22...Kxe7 23 Qxf6+ Kd7 24 Rd3
<This is somewhat more accurate than 24 Qxf7+ Kd8 (24...Kc6? 25
Rf6+ Kc5 26 Qe7+) 25 Rd3 (25 Qf6+ Kc7) 25...Bd7 26 Rxd4 exd4
27 Qf6+ Kc7 28 Qe5+ Kc6, where Black connects his rooks
(which, it is true, also should not save him).>
24...Qxd3 25 Bxd3 Re8
26 Bc4!
<The main thing is not to allow the opponent to coordinate his
pieces.>
26...Re7 27 Bxf7
<White has a small material advantage and a big positional one. The
outcome of the game is decided.>
27...Kd8
<Or 27...b5 28 Bd5 Ra7 29 Bc6+ Kd8 30 Qf8+, winning the ill-
fated rook.>
28 Qb6+ Kd7 29 Bb3 Ke8 30 Ba4+ Kf7
<After 30...Bd7 Black finally 'unravels' himself, but on the way he
loses a rook: 31 Bxd7 Rxd7 (31...Kxd7 32 Qxb7+) 32 Qe6+ and 33
Qg8+.>
31 Qd8
<Black resigns. He did not in fact manage to complete his
development.>
Kramnik's <My Life and Games>
Kibitzing
sac: 22.Re7+
(6.07): 31...Kf6 32.Bb3 Be6 33.Qxa8 Bxb3 34.Qh8+ Ke6 35.Qg8+ Kd7 36.axb3 Kd6 37.Qd8+ Rd7 38.Qf6+ 
