Learn From The Greats (part four)

Spanton (2147) - Stefan Djurić (2504)
Seeburg (Switzerland) 1999
English Symmetrical, Botvinnik Formation
1.e4 g6 2.c4 Bg7 3.Ne2 c5 4.Nbc3 Nc6 5.d3 d6 6.g3 Nf6 7.Bg2 0-0 8.0-0
By a slightly unusual move-order, the game has reached a tabiya in the Symmetrical Variation of the English Opening when White uses a Botvinnik set-up, ie pawns on c4 and e4, king's knight at e2 and a fianchettoed king's bishop.
8...Bd7
Other moves are more popular, including preparing queenside play with 8...a6 and starting to reroute the king's knight to d4 by 8...Ne8. But the text has been played by many grandmasters.
9.h3
Preparing Be3 without being annoyed by ...Ng4.
9...Ne8 10.Be3 Nd4
Preventing 11.d4, which would be the answer to 10...Nc7.
11.Qd2 Nc7 12.Bh6
Previous games had seen 12.f4, but Stockfish10 and Komodo10 prefer the text.
12...Qc8!?
The engines strongly dislike this, preferring 12...e5, or 12...Bxh6 13.Qxh6 e5. The idea is that with dark-square bishops being exchanged, it makes sense for Black to put pawns on dark squares.
13.Bxg7 Kxg7

How should White proceed?

*****
*****
*****
*****
*****
14.Nxd4!?
Djurić raised his eyebrows at this, but it is Komodo10's choice. Stockfish10 gives 14.Kh2 e5 15.f4 with a slight edge.
14...cxd4 15.Ne2
The engines slightly prefer the more-active 15.Nd5!?, one point being that 15...Nxd5 16.exd5 Bxh3 does not win a pawn as White has 17.Qf3.
Edit: as has been pointed out in the comments, this note makes no sense as the queen cannot go to f3. I should have written 17.Qf4, the point being White will win the d4 pawn.
15...e5 16.Qb4 Ne8 17.Kh2 a5 18.Qd2 Qc5 19.b3 Qb4 20.Rfd1
The engines reckon White should avoid an exchange of queens, for example by playing 20.Qc1!?, although there is then no doubt who has the better queen.
20...Nc7 21.a3?
This creates a weakness at b3.
21...Qxd2 22.Rxd2 Ne6 23.Rda2?!
Presumably hoping to play b4, but that was never likely. The engines reckon Black is better whatever White plays, but their fluctuating suggestions include Rb1, f4, h4 and Nc1.
23...Nc5 24.Nc1 b5 25.Rb1 Rfb8 26.Rab2 Rc8
The engines reckon Black should be doubling on the b file, but the text is fine too.
27.Bf3 Rab8 28.Bd1
If 28.Bg4, to try to swop off White's bad bishop, Black can simple reply 28...f5 with tempo.
28...f5 29.f3 fxe4 30.fxe4 Rf8 31.Kg2 Bc6 32.Bf3 Rb7 33.b4 Na4 34.cxb5 Bxb5 35.Rc2 Nc3
A fabulous outpost for the knight.
36.Rbb2 axb4 37.axb4
If 37.Rxb4, then 37...Rbf7.
37...Ra7
This is better than 37...Rbf7, which can now be met by 38.Rf2, whereas after 37.Rxb4 Rbf7, White cannot play 38.Rf2? because of 38...Nd1.
38.Rd2 Ra1 39.Ne2 Rf7
Not 39...Nd1? 40.Rxd1 Rxd1 41.Nxd4!
40.Rbc2?
After many of moves of just about holding on, a material-losing mistake comes. The engines give 40.h4, but reckon White has a positionally-won game.
40...Nb1 0-1
If 41.Rd1, then 41...Ba4.
LESSONS FROM THIS GAME
White was fine out of the opening, but several imprecise moves in the middlegame allowed Black to build a strong initiative on the queenside. Black never went for a quick kill, but gradually increased the pressure until a tactical mistake came.
(Next: two games against Kasparov)