How Fearsome is the Keres Attack?

I have always wondered how to best meet the primitive move of Paul Keres, the audacious 6. g4!? in the Sicilian Scheveningen. This is a very interesting question; it greatly depends on the temperment of the two players. For example, if white is one-dimensional (attack-only) the answer might be quite different than a white player who is more positionally sophisticated. Likewise, is black yearning to counter-attack or is he content, à la Ulf Andersson, to slowly accumulate positional advantages?  Before we start, I refer interested readers to a postage stamp featuring Mr. Keres!

Let’s examine these issues in a crazy game I played as black in the Swiss “A” Team League vs. GM Lothar Vogt.

GM L Vogt (Biel) – IM Mark Ginsburg (Riehen)    2000

1. e4 c5 2. Nf3 d6 3. d4 cxd4 4. Nxd4 Nf6 5. Nc3 e6 6. g4!?

keres1.png

Is it just me, or does this move look weakening?  Maybe I am being too Soviet here (“pawns do not move backward”).  One of the key squares weakened is f4.   Maybe black can get a knight there later?  Maybe black can interfere with white castling long (make him castle short) and then exploit the overextended kingside pawns? Strangely, as a junior playing white, I saw no objection to 6. g4 and played it a lot – scoring both a win and a loss versus NM John Meyer when very young. White’s agenda is clear – g4-g5 and further space gaining.  If 6…h6, black has slowed white down a little bit but he’s given white a “lever” that might help the future pawn storm. Let’s focus on the less usual methods that avoid 6…h6.

6…Nc6!?  An interesting suggestion in the old book by Kasparov and Nikitin on the Scheveningen. Black plans Nc6xd4 at some point followed, usually, by a timely …e6-e5.  This is particularly effective if white has played f2-f4.

 7. g5 Nd7 8. Be3  Logical.  A very natural alternative (but one that does not develop!) is 8. h4.  Black can then respond with 8…Nxd4 or 8…Be7 or even the provocative 8…Nde5!? aiming to exchange off a pair of knights to lessen the defensive burden.  The first one appears rather weak but the other two are more playable.  Let’s take a look.

A. Not equalizing is 8. h4 Nxd4?! 9. Qxd4 Ne5 10. Be2 Nc6 11. Qd3 Ne5 12. Qg3 Bd7 13. f4 Nc6 14. Be3 Qa5 15. O-O-O Rc8 16. a3 and white is better.

B.  8. h4 Be7 9. Be3 Qc7 10. Qd2 a6 11. O-O-O with a small edge.

C. The most principled try is 8. h4 Nde5!? 9. Be2 a6 10. Be3 Na5 11. f4 Nec4 12. Bc1 Qb6 with murky play.  White can also play 9. Nb3, avoiding the knight exchange. Then we have this example game:  8….Nde5 9. Nb3 Be7!  10. f4 Ng6 11. Qd2 h6 12. gxh6 Bh4+ 13. Bf2 Rxh6  Black is fine here. 14. O-O-O Bxf2 15. Qxf2 Qh4 16. Qxh4 Rxh4 17. Nb5 Kf8 18. f5 (18. Nxd6 Nxf4 is about level) 18… exf5 19. exf5 Bxf5 20. Nxd6 Bg4 21. Rd2 Bf3 22. Rg1 Nce5 23. Bg2 (Relatively best is 23. Nd4 Rf4 24. c3 Rd8 25. N6f5 Nc6 26. Bb5 Rd5 27. Rxg6 fxg6 28. Ne6+ Kf7 29. Nxf4 Rxd2 30. Kxd2 gxf5) 23… b6 24. Nd4 Bxg2 25. Rgxg2 Rf4 and black drew the game eventually,  1/2-1/2 Willumsen,H-Larsen,O/Aarhus 1989/EXT 1998

In this line, rather dubious is 8….Nde5 9. Nb3 Na5?! (as above, 9…Be7! is correct and not the time-wasting text) 10. Nxa5 Qxa5 11. f4 Nc6 12. Qd2 a6 13. O-O-O Be7 14. Qf2 Qc7 15. Bb6 Qb8 (white has a huge edge; I just provide the rest for tragi-comedy) 16. h4 O-O 17. f5 Bd8 18. Be3 b5 19. f6 Ne5 20. fxg7 Re8 21. Bd4 Qc7 22. Ne2 Bb7 23. Ng3 Rc8 24. Kb1 d5 25. Bd3 Nc6 26. Bc3 d4 27. Bd2 Ne5 28. Bf4 Nxd3 29. cxd3 Qd7 30. Rdf1 b4 31. Bd2 a5 32. g6 hxg6 33. h5 g5 34. h6 f5 35. Rh5 Ba6 36. Qe2 f4 37. Bxf4 gxf4 38. Rxf4 (Poor GM Kindemann!  Time trouble? He misses a forced mate with the elegant 38. h7+ Kxg7 39. h8=Q+! Rxh8 40. Qg4+ Kf7 41. Qxf4+ Ke7 42. Rh7+!! Rxh7 43. Qf8 mate) 38… Rc1+!!  A stunning black defense that must have shocked the Grandmaster. 39. Kxc1 Qc7+ 40. Kb1 Qxf4 41. Rh3 Kh7 42. Nh5 Qf7 43. Rf3?  (And now white even loses). 43…Qxh5 44. Rf8 Qxe2 45. g8=Q+ Kxh6 46. Qh8+ Kg5 47. Qg7+ Kh4 48.Qh6+ Kg3 49. Qg6+ Kh3 0-1 and black pulled off an improbable and very lucky upset, Kindermann,S (2500)-Dietze,W (2290)/Germany 1991/GER-chT.  I could not find this fiasco on chessgames.com.

 8…Be7?!  Very interesting is 8…Nde5!? here, carrying out the principal idea without delay. 

keres3.png 

Position after 8…Nde5!? (Analysis) 

For example, 9. Be2 Na5!? 10. f4 Nec4 11. Bc1 (11. Bf2!?) 11…d5! with a very unclear game where black has sufficient chances.  

If white decentralizes tries the interesting 9. Nb3 avoiding exchanges, 

keresnb3.png 

Position After 9. Nb3 – This might be pretty good 

black has the simple 9…Be7 10. f4 Ng6! with a reasonable game. The knight is well placed on g6 observing the weak squares that white’s 6th move created.  For example, 11. Qd2 h6! 12. gxh6 Bh4+ and black is fine, as was shown in Willumsen-O. Larsen 1/2, Aarhus 1989.  But 11. Qd2 looks weak; see the next paragraph for an improvement; namely 11. h4!  – in retrospect, 9. Nb3 might be good. This needs further work.

December 2007 note: recently I had the opportunity to test this line versus GM BOOrrj on ICC in a 5-minute game.

GM BOOrrj – Aries2  ICC 5-Minute, 12/28/07

9.  Nb3 Be7 10. f4 Ng6 11. h4! and this indeed is critical. White wants to swamp black and punish the unusual N on g6.  The game went: 11….h6 12. Qf3 and I could find nothing better than 12…hxg5 13. hxg5 Rxh1 14. Qxh1 and white obviously has an edge. This line needs revisiting.

 9. Rg1!? is another logical try.  9. Rg1 Nxd4 10. Bxd4 Nc6 11. Be3 Qa5! is a good move.  There might follow 12. Qd2 Be7 13. h4 O-O 14. O-O-O Rb8! Another good move to get the rook behind the battering b-pawn ram. 15. Kb1 b5 16. h5 b4 17. Ne2 Bb7 18. g6 b3!! A tactical motif well worth remembering.  Black has dynamic equality after 19. Qxa5 bxc2+.

9. h4 O-O 10. Qe2!?  White also has 10. Qd2 here but after 10…Nde5 black has no particular problems.  For example, 10. Qd2 Nde5! 11. Be2 (white has to watch the N fork on f3) 11…Na5! 12. b3 (what else?) Nac6 and black forced white into a rather ugly concession. 

The eagle-eyed reader would have noticed by now another path. Completely different is the surprisingly strong gambit line 10. g6! and white gets a strong attack after 10…hxg6 11. h5 g5 12. h6!.  Black can also defend with 10. g6!? Nde5, but after 11. gxh7+ Kh8 12. Rg1 it is looking very good for white.  Conclusion: the game move 8…Be7 is inaccurate and black would be better off with 8…Nde5 which needs practical tests.

10…Nxd4 11. Bxd4 e5   This is very commital, of course. Black gives away d5 to commence active operations.

There is the interesting (a little crazy, but interesting) gambit 11… b5!? here to scare white away from the natural plan of castling queenside. There might follow 12. Nxb5 Ba6 (Also possible is 12… Qa5+ 13. Nc3 or 13. Bc3) 13. a4 and white looks to be better, but this is a good blitz try. Black will enjoy some open lines.

The slower 11… a6 12. O-O-O b5 13. g6 fxg6 14. Bh3 Nb6 15. e5 d5 16. Qg4 Nc4 17. h5 b4 18. Ne2 g5 19. h6 g6 20. Qg3 a5 21. c3 gives white some edge.

12. Be3 Nb6 Following Nikitin and Kasparov to prepare Be6 and Rc8 and temporarily guard d5.

13. Bh3  White can play 13. O-O-O; but there are tricks. Here is an example of a pitfall line, 13…Be6 14. Rg1 Rc8 (Obvious intentions) 15. h5? Rxc3! 16. bxc3 Nc4! and black is much better after the forced and sad 17. Qf3 Qa5 18. Bxc4 Bxc4 with a huge attack.  And note that 15. Nd5? Nxd5 16. exd5 Bf5! is also very good for black. Thus black is not too afraid of 13. O-O-O.

13…Be6?!  A very cool tactical break-out here and a great blitz try is 13… d5!!?

 keres2.png

Position after 13….d5!!?  (Analysis) 

14. Bxb6! (Note that 14. Bxc8? Qxc8 15. exd5 Ba3!! is very good for black!) 14…Qxb6 15. O-O-O! (The best reaction, other moves give nothing) 15…d4 16. Nd5 Qc5 17. Rdg1 Be6?! (nobody likes to play the more passive 17… Kh8 18. Bxc8 Raxc8 19. Rh3 f5 20. gxf6 Bxf6 21. h5 Rc6 and white is a bit better) 18. Nf6+! Kh8 19. Qh5 gxf6 20. Bxe6 d3 21. Bb3 Qxf2 and black is hanging on but it’s not a lot of fun.

14. g6! Our familiar gambit line is still good. Black has to play and pretend that everything is in order, but in reality white has a huge edge.

14…hxg6 15. Bxe6 fxe6 16. h5 Most accurate is 16. O-O-O! Qe8 17. Qg4 Rf6 18. Bg5 Rxf2 19. Qxe6+ Kf8 20. Qg4 and white has a big initiative.

16… gxh5 17. Rxh5?! Here, 17. Qxh5! Nc4 18. O-O-O Nxe3 19. Rdg1! Bf6 20. fxe3 Qe8 21. Rg6 is very strong and really too much for black to handle.

 17… Rc8 18. O-O-O  A good alternative here is 18. Qg4! Qd7 19. O-O-O Nc4 20. Rh6 Bf6 21. Nd5 Nxe3 22. fxe3 with a big white plus.

18… Bf6 19. Qg4 Kf7 20. Bg5  The obvious 20. Rh7 Nc4 21. Bh6 Rg8 22. Rd3 with a big attack is too strong.  However, white hasn’t done anything wrong yet – see the note to his next move.

20… Rh8 21. Rd3? Now white misses a really crushing shot. 21. Nb5! is very hard to handle. 21… Nc4 22. b3 a6 23. Na7 Ra8 24. bxc4 Rxa7 25. Bxf6 gxf6 26. Rxd6! and white wins.

21… Rxh5 22. Qxh5+ Kg8 23. Bxf6 Qxf6 24. Qe2?  Another bad miscue. 24. Rxd6! Rf8 25. Kb1 Qxf2 26. Rd1 Nc4 27. Rh1 Ne3 28. Ne2 Nf1 29. a3 Nd2+ 30. Ka2 Nf3 31. Nc3 Qxc2 32. Qg6 and white is better.

24… Qg5+ 25. Kb1  Now black is completely OK.

25…Rxc3?!!  This move was not necessary.  Black is fine after the safe and sound 25… Rc6 26. Nb5 d5 27. Nxa7 Rc4 28.exd5 exd5 29. Rg3 Rh4 30. a3 Qf4 31. Qb5 Rh6 and even has winning chances due to his superior coordination. White’s knight is sadly offside. 

26. bxc3 Na4  Yes, this looks optically great for black.  But there is no KO.

27. Qd1 d5  Black gains nothing from 27…Qh4 28. Ka1 Qxf2 29. Rxd6.

28. Ka1 Nb6   The attacking 28… Qe7 29. exd5 (white also has the simple 29. Qh5! and the black queen cannot move off of the defense of e8) 29…Qa3!? can be met by 30. Qb1 e4 31. Rh3 e3 32. Rxe3 Nxc3 33. d6! Nxb1 34. Rxa3 Nxa3 35. d7 and wins.

29. Rg3 Qf4 30. Qg1 Qf7  Playable is 30… Qf6 31. Rg6 Qf7 and black holds.

31. Rh3 dxe4  Natural and good was 31… Qf4 with a solid game.

32. Qg4 Qf5  Perfectly good is 32… Nd5 33. Qh5 Qxh5 34. Rxh5 Nxc3 35. Rxe5 Kf7 36. Kb2 Nd1+ 37. Kc1 Nxf2 38. Kd2 b6 and black has enough pawns to bother white – the chances are balanced.   The weird computer-y move 32…Nd7 is also OK: 33. Qxe4 Qxf2 is equal.

33. Rg3 Qxg4??  Awful.  Mild time trouble was no excuse.  Black holds on after 33… Qf6 or 33… Qf7 34. Rg1 Nd5 35. Qxe4 Qxf2 36. Re1 Kf7 37. Qxe5 Qf5. The exchange of queens is clearly suicide.

34. Rxg4 Now white wins easily.

34…Kf7 35. Rxe4 Kf6 36. Rb4 Kf5 37. c4 g5 38. c5 Nd7 39. Rxb7
Nxc5 40. Rxa7 g4 41. Rc7 Ne4 42. a4 Kf6 43. a5 Nxf2 44. a6 g3 45. Rc3 1-0

Very sad.  But let the investigation of defending the Keres Attack without ….h6 begin!

Tags: , ,

2 Responses to “How Fearsome is the Keres Attack?”

  1. The Fabulous 00s: More Defending vs. The Keres Attack « IM Mark Ginsburg Presents A Personal Chess History Says:

    […] There’s something very logical looking about this move. 6…h6 gives white a lever for a later g4-g5. And the older 6…a6, once the most popular, has been convincingly shown to be too slow. So that leaves 6…Be7 (similar to the text) and the very risky 6…e5?! which we will cover in another installment.  For more on 6…Nc6, see my first article (the GM Vogt game). […]

  2. Eric Jordan Says:

    I wonder why nobody seems to take Nfd7 more seriously – I played this several times in internet tournaments and had very promising results with this moves which takes away the “strength” of a further g4-g5 and now white has to cope with the “weak” h4 square, for starters. In all variations when I checked this with my computer this was at least not any worse than the usual other way like Nc6 and Be7 or the once “best” a6 because in all of these variations Black has to retreat his knight to d7 anyway but WITHOUT being FORCED to do so…

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s


%d bloggers like this: