Monday, April 04, 2005

Dragon Games 3: Brilliancy or swindle?

Computers sometimes tell you things about your games that you would rather not find out. This has always been one of my favorite Dragon games, a nice crushing win over a higher-rated opponent. Annotators (but not computers!) tend to suffer from result bias, they tell you that whoever won was winning all along, that the other guy never had a chance, etc. So I played this game out on Fritz one evening, and was given a major shock.

White: Ilan Brand (X) Black: Victor Reppert (A)
Intra-club League, 1972
1. e4 c5 2.Nf3 Nc6 3. d4 cxd4 4. Nxd4 g6 5. Nc3 Bg7 6.
Be3 Nf6 7. Bc4 d6

Entering via the Accelerated move order avoids the 9.
O-O-O system, which according to some commentators,
such as Mikhail Golubev, is thought to be more
dangerous than the 9 Bc4 system. But it does allow
White to play the dull and boring Maroczy, uh, er Swiderski Bind with 5.c4, which for a long time was thought to provide White
with an advantage. More recently, under the influence
of Tiviakov, it has acquired a drawish reputation, but
Tiviakov has suffered a couple of losses in this line.
I don’t remember why I didn’t play 7 …O-O 8. Bb3 d6 9.
f3 Bd7, though this would only yield an independent
variation if I were to play 10. Qd2 Nxd4 11. Bxd4 b5,
a line that Dennis Monokroussos has discussed a couple
of times on his chess blog.
Which move order should a player use who intends to
enter the Dragon? Perhaps it is a question of which
lines your opponent is more likely to be happy in.
Some players are happy Maroczy Bind players, others
are not.

8. f3 O-O 9. Qd2 Bd7 10. O-O-O Qa5 11. Bb3 Rfc8 12. h4

Here 12. Kb1 Ne5 13.Bg5 is the now hotly debated Moles variation.

12...Ne5 13. Kb1

But this variation poses a set of problems as well. My theory is that it will eventually prove to be the most dangerous line of all.
Historically, this variation was thought to give White
an advantage, though this assessment has been
challenged by Ward. Here the game diverges from the
Polchinski and Rowley games.

13...b5

This is the move originally played by Korchnoi in
1967, and recommended in Ward’s Winning with the
Dragon. In WWTD2 Ward doesn’t retract the
recommendation, but presents 13…Nc4, which used to be
the main line, as a viable system for Black. However,
a testing line has arisen. After 14.Bxc4 Rxc4 15. Nb3
Qc7 (Ward’s recommendation) 16. g4 (instead of the
usual 16. Bd4) Rc8 17. e5, and now I think best for
Black is Nxg4 18. fxg4 Bxe5, with a very unclear
position that may be just a little better for White.
Here 15…Qd8 may be an underrated alternative; the
standard refutation is 16. e5 Ne8 17. h5 Bxe5 18. hxg6
hxg6 19. Bd4 but now Bxd4! 20 Nxd4 e5! 21. Qh6 Qf6 22.
Qh7+ Kf8 23. Nd5 Qg7 is only a draw. However, 17. exd6
may be a way White can play for the advantage. 15…Qa6
and 15…Qe5 are also alternatives, but I don’t have
much confidence in either of them.

14. h5

I consider this to be the weakest of a set of three
alternatives. 14. Ncxb5 is the most popular variation,
and Rowley talked me out of 13…b5 shortly after this
game by convincing me that Black doesn’t have enough
for the pawn in the endgame after 14…Qxd2. Ward and
others had some successes in this variation, but in a
game against Reeh, Ward faced 15. Bxd2 Nc4 16. Bxc4
Rxc4 17. b3 Rc5 18. c4 a6 19. Nc3 Rb8 20. Kc3 Ng4 21.
fg4 Bxd4 (Reeh-Ward Bern 1993) and after 23. Nd5,
White could have secured and advantage, according to
Golubev. However, Golubev also suggests 17…Rcc8, which
deserves some more careful attention. Also, 15…Rab8 is
another important line, after which one line is 16.
Nc3 Nc4 17. Bg5 Kf8 18. Rhe1 a5 19. Bxc4 Rxc4 20. Nde2
Nxe4 21. fxe4 Rxc3 22. Nxc3 Bxc3 23. Re3 Rxb2+ 24. Kc1
be5 25. Rb3 Rxb3 36. cxb3 with a slight edge for White
according to Velickovic. But there are lots of
alternatives to explore here.
14. Bh6 is considered superior to the text here. Now
14….Bxh6 15. Qxh6 Rxc3 16. bxc3 Qxc3 17. Ne2 Qc5 18.
h5 Nxh5 19 Rxh5 is bad for Black, and if 18…Nc4 19. hg
fg 20. Qd2 Be6 21. Qd4 was better for White in
Grabarcyzk-Jedryczka, Poland 1994. But I wonder if
17…Qb4, which has appeared in a couple of games, might
be an improvement for Black here. Uh, no. after 18. h5 Nc4 19. hg fg 20 e5! White looks good. However, 16...Nc4! has won 2 1/2 out of three for Black in this position, in spite of being neglected by the theory books.
The other major line is 14…Nc4 15. Bxc4 Bxh6 16. Qxh6
bxc4 17. h5 Rab8 18. Nd5 Nxd5 19. exd5 Qa3 (Qc3 and
Qb4 are unexplored alternatives) 20. Nb3 cxb3 21. bxa3
Bxc3++ 22. Ka1 cxd1(Q)+ 23. Rxd1 Bf5 24. g4 was played
in the complex game Tolnai-Jovicic Leibnitz 1990.

14...Nc4

Black has the option of playing 14…Nxh5, which leads
to familiar positions with Kb1 and b5 included. Or
perhaps 14…Rxc3, which Ward played in the position
with these two moves excluded, should be considered
here.

15. Bxc4 bxc4 16. hxg6

If 16. Bh6 Black has Rab8 17. Bxg7 Qb4 (Ward).

fxg6 17. Bh6 Bh8

A well-known stratagem, preserving the Dragon bishop.

18. Nde2

24 years later, in the game Bisby-Summerscale British
Championship 1996, White played 18. Nf5 gxf5 19. e5
and here dxe5 looks like it would have given Black a
decisive advantage.

Rab8 19. Qe1 Be6 20. Qh4 Qb4 21. Bc1 a5 22. e5 dxe5
23. Ne4

This is the critical position of the game. I thought
for many years my move was brilliant, but actually
White should have won. The idea is that after White
does his worst attacking h7, he ends up with nothing
and Black crashes through. That’s what happened, but
objectively Black was better off defending with 23…h5,
which leads to a perfectly good game for Black.

c3?! 24. Nxf6+ Bxf6 25. Qxh7+ Kf8 26. b3 a4 29. Qxg6??

White misses the boat. 29. Bh6+! Ke8 30. Bg7!! was a
winner. Score one for the silicon monster.

Bf7 28. Bh6+ Ke8 29. Qf5

White gets to threaten mate in one. But it does him no
good.

Rc7 30. Bc1 axb3 31. axb3 Bxb3 0-1

1 comment:

Victor Reppert said...

An addendum; after 14. Bh6 I was looking for a way to save Bxh6 15. Qxh6 Rxc3 16. bxc3 with Nc4, which had gotten 2 1/2 out of 3 on the database, but after 17. Bxc4 bxc4 18. Qd2 Rab8+ 19. Ka1 White looks good.