I’ve Been Working on the Caro …

by admin on November 10, 2009

… All the live-long day! (Sorry about the reference to a nineteenth-century folk song that is probably best forgotten.)

Anyway, Michael Goeller posted a comment a week ago that set me off on another serious binge of Caro-Kann study. He pointed out this post in Mark Ginsburg’s blog, which re-opens the debate over what I called the Homo Erectus Variation of the Caro-Kann, where White plays the ungainly lunge: 1. e4 c6 2. d4 d5 3. e5 Bf5 4. g4!?

Ginsburg calls this the “Fugly Advance Variation,” a name that I’m not sure is acceptable for a family audience. Another possibility would be the “Mackenzie-Ginsburg Variation,” which would have the ironic advantage of naming the variation after an adamant opponent of it! But maybe the best name of all would be “Bronstein’s Folly.” That’s because Ginsburg dug up an old game Bronstein-Petrosian from 1959 where Bronstein (who was never afraid of a sharp battle) played this saucy move.

Ginsburg wrote about this variation because it appeared in a game from Week 9 of the U.S. Chess League between Pascal Charbonneau and Tenghsuren Enkhbat. Before we get to that game, there is still an old matter that I need to follow up on from the original debate between Ginsburg and me.

Ginsburg claims, in his notes to Charbonneau-Enkhbat, that Bronstein’s Folly is refuted by 4. … Be4! And he gives a very persuasive argument in his original post here. By the way, you should definitely read his post, because it’s a very nice example of using logical reasoning to fight against computer-inspired analysis. However, there is a catch, which is that he makes it appear too easy. In nine tournament games after 4. … Be4 I have only faced one opponent who came up with the key eighth move in Ginsburg’s refutation. That was an opponent who prepared for me at home, and even so he managed to mess up later and lose. The story is recounted in this post.

I have two arguments to make against Ginsburg. First is that in his refutation, 4. … Be4 5. f3 Bg6 6. h4 h5 7. Ne2 hg 8. Nf4 Bh7! 9. e6 fe 10. fg Qd6! he does not consider White’s answer 11. Qe2. If he was thinking of 11. … Be4 12. Rh3 e5? then he would run into 13. de Qxe5 14. Ng6. However, I am not going to press this argument too hard because I think that Black can play more patiently with 11. … Nd7 and still be doing quite well for the basic reasons that Ginsburg outlines.

My second and more important argument is White can avoid Ginsburg’s line by playing a different move order! I did not realize this when I wrote my original series of posts on Bronstein’s Folly (formerly the Homo Erectus Variation). The move order I would now like to recommend is 4. … Be4 5. f3 Bg6 (see diagram below) 6. Ne2.

There are two points to this move. First, White has a momentary pang of conscience and actually plays a developing move instead of 6. h4. I should not have to justify that decision! Second, this move is more flexible. 6. … h5 would now be a mistake because after 7. Nf4 White is a tempo up on the line Ginsburg recommended. I think that 6. … c5 is also too early, because now White plays 7. h4 h5 8. Nf4 followed either by 9. Nxg6 or (if 8. … Bh7) 9. e6! Again, if we play a la Ginsburg for Black and try 9. … fe 10. Nxe6 Qd6, it just doesn’t work because of 11. Qe2. Black is a tempo behind Ginsburg’s line, and now 12. Bf4 is coming.

So I think the main variation after 6. Ne2 must be 6. … e6, a solid developing move for Black that ends all of the e5-e6 possibilities for White. Now we play 7. h4 h5 8. Nf4, reaching the diagrammed position.

Comparing this with Ginsburg’s refutation, we see that Black has played … e6 instead of … hg. If Black now tries to play 8. … hg 9. Nxg6 fg 10. fg! Rxh4 11. Rg1! we get to a position I discussed in my original post, which is surprisingly good for White.

An alternative for Black, which I have faced several times in speed chess but not yet in a tournament, is 8. … Ne7. Here Black is trying to turn the move order to his advantage; because he has played … e6 he now has the option of defending his bishop. The jury is still out on this, but I have had good luck in my speed games with the provocative 9. Nc3 Nd7 10. b4!? Black’s problem is that all of his pieces are stepping on each others’ toes, and another problem is that if White gets his knight to b5 (e.g. after 10. … a5 11. b5 cb 12. Nxb5) the threat of Nd6+ is pretty big. And if 12. … Nc6 to defend that threat, now 13. Nxg6 takes the wrecking ball to Black’s kingside. If White doesn’t want to tempt fate with 10. b4, another option would be 10. Be3 with the plan of Qd2 and O-O-O.

Finally, there is a move that I’m sure Ginsburg would approve of: 8. … Bh7! Black sticks to his policy of creating no weaknesses, and he realizes that he is not really sacrificing a pawn because after 9. Nxh5 Bg6! White cannot hold the knight on h5 and the pawn on h4. I see two possible treatments for White. First, White can simply return the pawn with 10. Nf4 Rxh4 and then play the computer’s recommendation 11. Bh3. The computer considers this position equal. Black once again has some pieces that step on each others’ toes; for instance, if he plays 11. … Bh7? to save his bishop, it blocks the rook’s retreat and White wins the exchange with 12. Ng2.

But the alternative, and really exciting possibility after 9. … Bg6 is to go for broke with 10. Bg5?! Qb6 11. Nf4. White asks Black: Do you have the guts to take on b2? Because Black is a Caro-Kann player, he probably doesn’t. If he does play 11. … Qxb2, then we can offer a whole rook with 12. Nxg6 fg 13. c3!?

Again, because Black is a Caro-Kann player, he probably doesn’t have the guts to take the rook. If he does, then after 13. … Qxa1 14. Qb3 we have just the kind of wild, obscure position White was hoping for.

Is this sacrifice sound?

Probably not. The computer gives Black an advantage of between 1 and 2 pawns. But I cannot emphasize too strongly, your opponent is not a computer! Against a human opponent, what I would expect to happen is that Black will give up his queen for a rook, giving him the nominal advantage of two rooks against a queen. But it’s not going to be such a great position for the two rooks, because there are no open files and because there are nice diagonals for White’s queen and bishops, and weak pawns on e6 and g6 for them to target. I would be happy to play White here.

Once again I feel as if I have probably fallen into the trap of giving you lots of lines, rather than reasoning from general principles as Mark Ginsburg did in his post. So let me back up and try to explain from general principles why Bronstein’s Folly is playable for White and how you should play it.

  1. First, you should recognize that psychology is a big part of this variation. You smack your opponent right in the mouth on move 4 and you don’t stop. By playing 1. … c6, your opponent has already advertised that he is not in the mood for a fight, so you take the fight to him.
  2. In the event that your opponent plays 4. … Be4, I now think that 5. f3 Bg6 6. Ne2 is a somewhat more precise move order than 6. h4 right away.
  3. Yes, White has created big weaknesses with 4. g4, and Ginsburg correctly points out that pawns cannot move backward. In order to get compensation, White must (a) create equally serious weaknesses in Black’s pawn formation, and/or (b) prevent Black from castling queenside. The key moves to do this are e5-e6 and Nf4xg6 (after Black has played … h5). The pawn thrust is especially effective if Black’s bishop has retreated to h7, because e6xf7+ will flush Black’s king out into the open. Also, g4-g5-g6 may come into play. If Black allows the capture on g6, then of course Bd3 or Qd3 targeting the g-pawn is next on White’s agenda.
  4. In lines where Black plays … Ne7, he may have some problems because all of his pieces are in each others’ way. The bishop on g6 is in the way of the knight, the knight on e7 is in the way of the bishop on f8, and the pawn on c6 is in the way of both knights. I like the idea of trying to prevent Black’s freeing move … c6-c5 with Nc3, Be3 and possibly even b2-b4.
  5. Don’t be afraid to sacrifice some material!

In my next post I will discuss the 4. … Bd7 variation again. Ginsburg, in his post, criticizes this move: “Black submits to white’s bully-boy ploy.” Well, yes, that’s right — and it goes perfectly with point (1) above, that 4. g4 is partly a psychological move. Curiously, in my experience and in spite of Ginsburg’s comments, masters almost always play 4. … Bd7. The reason is not that  masters are more easily intimidated than amateurs. I think the reason is that masters are more willing to make a temporary concession, in the expectation that over the long term White’s weaknesses will come back to haunt him (and also expecting that the bishop will find greener pastures on the queenside after … c5). The amateur, on the other hand, probably will not even consider playing 4. … Bd7 because it looks like a loss of time and it puts the bishop back behind the pawn chain.

Anyway, if you are going to play Bronstein’s Folly against masters, experience shows that you will often face 4. … Bd7. For that reason it behooves you to be prepared for that move and take it seriously. More on that next time.

Print Friendly, PDF & Email

{ 8 comments… read them below or add one }

Michael Goeller November 11, 2009 at 9:26 am

I knew that if I egged you on you would respond to Ginsburg…. I think Ne2 is a critical improvement. And I like your Queen-trap line (though it is not critical to your analysis). It reminds me of the “Mammoth Trap” in the Caveman Caro-Kann. But are you sure the Queen cannot escape by pushing the pawn b7-b5-b4?

BTW: Playing the Caveman Caro-Kann, I find that about 4 out of 5 Caro players “do not have the guts to take on b2.” So you are onto something. But the ones who don’t take tend to be the strongest players and go on to beat me most of the time…. 🙂


admin November 11, 2009 at 6:03 pm

A couple of followups: For those who are too lazy to follow Mark Ginsburg’s link, his basic argument is that after 6. Ne2 Black should play 6. … e6 7. h4 h6 (rather than … h5). He says, “… as any Caro player knows, if you have a hiding square for your bishop and have e7-e6 in as well, you’re happy.”

This is a good nugget of information for Caro players as well as people playing against the Caro. Still, I would be interested in seeing what results other players get with this line. I think that players under 2200 will not have nearly the positional subtlety of Ginsburg, and ironically players over 2200 seem to avoid the whole 4. … Be4 debate altogether.

In response to Michael Goeller’s comment, I think I was too optimistic about the rook sacrifice after all. After the most logical sequence, 14. … b5 15. Bd3 a5 (… b4 is also good) 16. O-O a4 17. Bxg6+ Kd7 18. Qc2 b4, White’s problem is that he can never spring the trap with Nd2 because Black takes the c-pawn. I didn’t really appreciate this fact when I was taking my first superficial look at this line. White can try to stir up trouble with Bf7 and f4-f5 and so on, but Black can defend with … Ne7 and then just push right on with a4, b3, etc. The only slight trick is that after ab, Black should continue with a3 (rather than recapturing on b3) because this forces White eventually to sacrifice his knight.

The rook sac is probably fine for blitz or skittles, but for a tournament game White had probably better play the first line I mentioned, or not even take on h5 in the first place.


SL September 17, 2015 at 7:19 am

Why not play 5.Nf3 rather than 5.f3? It seems logical and I can’t find a problem with it.


Roman Parparov September 18, 2015 at 5:58 pm

5. … h7-h5!


SL April 16, 2017 at 11:32 am

It seems to me and to the computer that 6.e6 is better than fine here.

I assume 5.Nf3 doesn’t work because it’s never been essayed but the computer thinks it’s comparable to 5.f3 in strength.


admin April 16, 2017 at 2:20 pm

Hi SL, tbanks for your comment!

I have two problems with 5. Nf3. First is that even if Black totally plays into my hands with 5. Nf3 h5 6. e6 f6, I still have trouble seeing any kind of concrete advantage for White. Second is that 5. Nf3 allows Black the option of simple, calm development with 5. … e6 6. Nbd2 (I assume this is your idea) 6. … Bg6.

In the opening I’m not just looking for moves, I’m looking for IDEAS. The idea behind Bronstein’s Folly, or whatever we end up calling it, is that White is determined to do at least one of two things: disrupt Black’s kingside pawn structure and prevent Black from castling. The two key moves White has to accomplish these goals are e5-e6 and Ne2-f4. By playing 5. Nf3 e6, we abandon both of those goals. Then we have no tangible justification for the self-inflicted weakness of the pawn on g4.

So even if the move 5. Nf3 is evaluated by the computer as being as good as 5. f3, it is the absence of ideas behind it that troubles me.


SL April 17, 2017 at 6:54 am

Thanks! That broad perspective is incredibly helpful. “In the opening I’m not just looking for moves, I’m looking for IDEAS” is a statement I expect I will return to over and over agan.

SL September 18, 2015 at 8:26 pm

Ok so 6.e6


Leave a Comment

{ 2 trackbacks }

Previous post:

Next post: