1.d4
I had an idea in the Db3 Grunfeld that I had wanted to try. In Wijk aan Zee, Kasparov entered the same system against Svidler, but he had another idea in mind. Now it was my chance.
1...Sf6
2.c4
g6
3.Sc3
d5
4.Sf3
Lg7
5.Db3
dxc4
6.Dxc4
0-0
7.e4
a6
Svidler repeats the Hungarian system.
8.e5
b5
9.Db3
Sfd7
10.h4
Kasparov played e6, but I assumed that Svidler had improved on that.
10...c5
11.e6
c4
I had briefly acquainted myself with the theory of this move, but had concentrated on Black's alternatives.
12.Dd1
Sb6
13.exf7+
Txf7
14.h5
Sc6
15.hxg6
hxg6
16.Le3
This is all theory. The main move is 16...Lg4, but Svidler had something else in mind.
16...Lf5!
Svidler said that he was surprised that this natural move had hardly ever been played.
17.Sg5
Obviously, the difference between Lf5 and Lg4. I thought it would be better not to follow Lputian-Sokolov blindly, so when I saw this plan, I decided to go for it.This is the sort of thing one shouldn't really do this without developing all his pieces, but if you already played h4 on move 10, then it doesn't matter.
However, this move isn't totally illogical. White wants to play g4 and Lg2 instead of placing the bishop on e2.
Still, Black seems to be OK after this move. [17.Le2
Sb4
18.Tc1
I remembered that Lputian-Sokolov had reached this position. 18...e5!
Later I discovered that this was played in Kroeze-Van Haastert Wijk aan Zee 1998. (No, not Hoogovens, but Sonnevanck!). I havent' checked everything, but it looks good for Black.]
17...Tf6
[17...Sxd4
Black seems to have enough compensation here. For instance 18.Sxf7
Kxf7
19.Tc1
e5
20.g4
Le6
21.Lg2
Ta7
]
18.g4
Le6!
The bishop goes for the d5 square. [18...Ld7
19.Lg2
]
19.Sce4
[19.Lg2
Ld5
20.Sxd5
Sxd5=/+
]
19...Ld5
20.Dd2
[20.Sxf6+
exf6
21.Sf3
Dd7
Gives Black too much ~/= for the exchange!]
20...Td6?
This wins the d-pawn (White can hardly save it now, but White's attack on the h-file more than compensates. [20...Dd7!
Svidler. The idea is that Black can now meet f3 with Txf3.
It is difficult to analyse a position like this. White has many options - he can play the immediate f4 or interpolate Td1 and Td8 first or Lg2 and so on. The usual cop out "unclear" seems called for!]
21.f3!
Now the rook will be sorely missed on the f-file.
21...Lxe4
[21...Sxd4?
22.Lxd4
Lxe4
23.Dh2+-
(23.Lxg7
Txd2
24.Le5+-
) ]
22.fxe4
Sd7
[22...Sxd4
23.Lxd4!
This seems to work and is much simpler. (23.Dh2
I considered this during the game. White plays Lxc4+ and follows up with Dh7+ and castles check, but afterwards a simpler win presented itself. 23...e5
) 23...Txd4
(23...Lxd4
24.Dh2
Lg7
(24...Kf8
25.e5
) 25.e5!
Tc6
26.Le2+-
) 24.Df2!
(24.Dh2
Dd6!
25.Lxc4+
Sxc4
26.Dh7+
Kf8
) 24...Df8
(24...Lf6
25.Th8+!+-
) 25.Dh4
Df6
26.Le2!
Black seems to have no defence against Dh7+ and Tf1.]
23.Dh2
Sf8
If White gets Dh7 in, then it's usually over. [23...Db6
24.Lxc4+
bxc4
25.Dh7+
Kf8
26.0-0+
Sf6
27.Dxg6
Kg8
28.Df7+
Kh8
29.Tf3
Sxg4
30.Th3+
Th6
(30...Sh6
31.Df5
Kg8
32.Dh7+
Kf8
33.Tf1+
Ke8
(33...Tf6
34.Se6++-
) 34.Dxg7+-
) 31.Txh6+
Sxh6
32.Dg6
Kg8
33.Tf1+-
]
24.e5!
[24.d5
Se5
Allows Black to keep fighting.]
24...Td7
[24...Txd4
Fails to... 25.Df2
Da5+
26.Ke2+-
]
25.Se6!
Da5+
[25...Sxe6
26.Lxc4!
A recurring theme. 26...Da5+
(26...bxc4
27.Dh7+
Kf7
28.0-0++-
) 27.Ld2
Dxd2+
28.Kxd2
Txd4+
29.Ke3
bxc4
30.Dh7+
Kf7
31.Taf1++-
]
26.Ld2
Sxe5!
[26...Db6
27.Sxg7+-
]
27.Le2!
c3
[27...Db6
28.Sxf8+-
]
28.Lxc3?!
[28.bxc3!
Sf7
Gives White a better version of the next variation.]
28...b4
29.Sxg7
bxc3?
Svidler thought he was lost already and didn't find his last chance. [29...Sf7!
30.Ld2
Kxg7+/-
The open h-file and his two bishops promise White a massive edge, but Black is still fighting. Also, White has to watch the d4 pawn.]
30.Dh8+
Kf7
31.0-0+
The reason why Peter played on so long - he wanted to finish a game where White plays 0-0+ winning. Unfortunately, it was published in the net as 31.Tf1+ resigns and it took a while for this to be corrected. Still, for the record, 31.0-0+ was the last move. 1-0