In the game I played as white, we followed theory to the 16th move! It later turned out we'd studied the same opening text. Unfortunately for me, I hadn't looked at it within the weeks before the tournament...

The game continued:
10.Nh3 c4 11.Be3 Qb4 (11...axb5 12.0-0-0 Qb4 is more usual) 12.0-0-0 axb5 13.Nf4 Ne7 14.Qf2!?

<<This avoids the deeply discussed queen sacrifice line: 14.Bb6 h5 15.Rxd5 Qxc3+!! 16.bxc3 Nxd5 with a very unclear game. White is probably objectively better, but it's easy for black to generate a dangerous attack on the black king. The classic example is B Lalic vs Khalifman, 1997;>

14...Qa5 15.Bb6 Qa6 16.Rxd5 (the rook can't be taken, otherwise a white knight will make it to c7) 16...Bc6

and to here we have been following Lim Chuin Hoong, Ronnie vs Wong Zi Jing, 2000 which continued 17.a3. I certainly considered the move, but I had to admit to myself that I couldn't remember any further, so I tried to look at the position completely afresh. This led to me playing the rather bizarre novelty 17.Rxb5?!?

click for larger view

The main idea being to free the d5 square for the knights. White gets some okay compensation, but not enough. Play continued 17...Bxb5 18.Nfd5 (Perhaps stronger would have been 18.Ne4) 18...Nxd5 19.Nxd5 Qxa2! (now I realised I was in some real trouble) 20.Nc7+ Kd8 21.Nxb5+ Kc8 22.Na7+ Kb7 23.Qd4 reaching this bizarre position:

click for larger view

Both sides are well behind in kingside development and have exposed kings, but it turns out white's is in more danger, and I should have been beaten fairly quickly from here. In the end I was lucky to struggle into an ending down a piece, but with an extra pawn and an initiative, and the game was eventually drawn.

1-0 9373 =39% total:52%
1/2-1/2 6450 =27%
0-1 8281 =34% total:48%

<BENKO GAMBIT, WHY YOU SHOULD PLAY …g6 BEFORE …Bxa6

Mr. Ortner asks:

I usually don’t care to update the openings I play. That’s why my reference book for the Benko gambit is from 1991 (by Suetin).

Recently, I came across a review of Neil McDonald’s new book THE BENKO GAMBIT REVEALED, where I read the following:

“...he explains why you have to play g6 before taking on a6 with the bishop.”

I don’t know McDonald’s book and wondered what’s wrong with taking on a6 with the bishop immediately. Moreover, my book doesn’t see anything wrong with the line 1.d4 Nf6 2.c4 c5 3.d5 b5 4.cxb5 a6 5.bxa6 Bxa6 either.

Is there really something like a refutation of this line? Should I think about buying a new opening book for the Benko?

Silman replies:

A lot of things change in theory in a very short time, but the need to get new books and keep up with those changes depends on the level of competition you face. If you’re a serious tournament competitor, you really should upgrade. If you just play for fun on-line or with friends, then there’s no reason to do so.

The view that an immediate 5…Bxa6 is inaccurate has been generally accepted for a couple decades. However, this is based on the assessment of the following line: 1.d4 Nf6 2.c4 c5 3.d5 b5 4.cxb5 a6 5.bxa6 Bxa6 6.g3 d6 7.Bg2 g6 8.b3! Bg7 9.Bb2 0-0 10.Nh3 Nbd7 11.0-0 and though Black has tried just about every possible move and idea, White seems to come out on top.

Because of this, 5…g6 became Black’s main choice. Now Black can meet 6.b3 Bg7 7.Bb2 0-0 8.g3 with 8…Nxa6 9.Bg2 Bb7 when Black is doing well due to the possibilities of …Nb4 and/or …e6.

As white I never accept that a6 pawn, then you're almost always stuck with hand castling, and what seems to be a lost endgame.

At the interactive chessok.com is a rybka engine, on the hardest level it will allow the benko gambit, and sometimes play the handcastles version as white. It's fast, and good practice if you can't afford the real rybka/fritz set-up, or can't get to anything else otherwise.

If you still have your prem. membership, you'd know the answer. =)

4...e6 2 games, all 0-1

4...bxc4 Christian Manley vs J Rosenberg, 2001 0-1

4...g6 C Aarnes vs J Lunnan Hjort, 2001 0-1