1. d4 Nf6 2. c4! c5 3. d5 b5 4. cxb5 a6 5. bxa6 Bxa6 The usual move is 5... g6 and if 6. b3 Bg7 7. Bb2 Nxa6! is a better way to capture. The point is that ...Nb4, hitting d5 and a2, can not be well dealt with by Nc3 because of the latent pin down the long diagonal, which often allows ...Nfxd5 at some point. Nevertheless, this is quite a technical point and Pedersen writes that the double-fianchetto is "...probably nothing to be afraid of" anyway. It is not worth memorizing such details until you have some playing experience. 6. Nc3 g6 7. g3 d6 8. Bg2 Bg7 9. Nf3 O-O The move order 9... Nbd7 is designed to restrict White's options. The cutting edge move is 10. Rb1! after 10.O-O Nb6! forces 11.Re1,which is not a priority move, to meet 10... Nb6 with 11. b3 reaching the desired structure without wasting even a single move. 10. O-O Nbd7 11. Rb1 Qc7 I think both of us started to think for ourselves here. I have since read that Pedersen has "never understood this positioning of the queen in the Benko." He doesn't like the way it blocks the path of the king's knight to the queen-side via e8-c7-b5.My next move happened to hit that nail on the head. 12. h3! This shuts the door on the knight's other good route, g4-e5.Annotating his game against van der Sterren, Michael Adams praises this idea even with the black queen on a5 - it must be even better here. 12... Rfb8! This is such a normal move that I failed to notice that it neutralised my last move. 13. Re1? Negligence and incompetence - I miss Black's threat and prepare to play a lemon. 13. Qc2! 13... Ra7 Objectively this is wrong but it is such a psychological master stroke that it deserves praise. Mark is so sure that I am going to play the following error that he makes a useful waiting move, since the execution of his threat will then be even stronger. What a cheeky sod! 14. e4? In his book, Robert Bellin says that playing the fianchetto system "...implies giving up the natural plan of central expansion as this would weaken the white squares. Having given the a6-f1 diagonal to Black, the d3-square would be particularly weak and easily exploited by a knight on e5." Adams points out that Black sometimes plays 14... Bc8 in this line. "The bishop comes to f5 with the aim of encouraging White to respond e4." He may then return the bishop to a6 so you could say that I just lost 4 tempi! However, since the knight can't get to e5 as I have played h3, I left the table for a stroll, full of confidence. 14... Ng4! 15. Qc2 15. hxg4 Bxc3 16. Re3 Bg7 is great for Black as White has real trouble developing his queen-side. The removal of the Nf6 and Nc3 has amplified the power of the Bg7 considerably. Mark and I dismissed this line but it may be no worse for White than the game continuation, which was bad enough!; 15. Bd2? Bd3 picks up the pawn on b2 and the a2-pawn won't be long. 15... Nge5 16. Rd1 Although Qc2 and Rd1 turn out to be normal moves in this variation, they are usually to prevent threats, not to answer them. It felt really bad to be dancing to Black's tune like this. And the rook would not usually take two moves to get from f1 to d1. 16... Rab7?! Mark made a draw offer here but I declined as he had missed what I considered to be a winning move. Had he offered a draw a move earlier or a move later, it would have been accepted without hesitation! In the post-mortem I suggested that 16... Nxf3+! would be very strong. 17. Bxf3 Ne5 18. Be2 (18. Bg2 Qa5 is like the game but without White's knight on e1, a major loss.) 18... Bxe2 19. Qxe2 c4 looks awesome for Black - just think of that knight on d3. 17. Ne1! Not played with any hope of trapping the knight with b3 and f4 but giving defensive cover to d3.The fact that my queen is defended even becomes important and makes this a star move, emphasising the importance of Black missing 16...Nxf3+! However... 17... Qa5! My hope that Ne1 was my salvation faded after this move and I regretted declining the draw offer. I searched in vain for a continuation that avoided defeat. In the chapter "Fortune favours the lucky" Simon Webb points out how wrong it is to sit there analysing all your opponent's winning combinations - that is his job! I forced myself to play something and found the path of highest resistance. 18. b3 This obliges Black to calculate accurately and it is the only move to give me a chance if Black does make a mistake. 18... c4? Mark bashed this out instantly but he may still be regretting it. It is a thematic move but it is not effective here. In fact, it is quite harmful. I knew that 18... Nc4! was winning, perhaps subconsciously recalling this shot from van der Sterren-Adams. Once I showed it to him, Mark immediately recognised it as winning. We decided that White's best was 19. Ne2 (19. bxc4 Qxc3 20. Qxc3 Bxc3 21. Rxb7 Rxb7) is a typical Benko ending. The pawns on a2 and c4 are dead, leaving Black with a protected passed c-pawn - winning. 19... Na3! 20. Bxa3 Qxa3 although it is a Benko nightmare for White. Mark highlighted the unchallengeable Bg7 as the key. The extra white pawn is irrelevant - if it survives it is still not going anywhere. I have had so much experience at recovering from lost positions that I don't need more than one invitation back into the game. 19. Bd2! Last move, this would have been met by 19...Nc4 but this has been ruled out by Black's hasty pawn push. 19... Nc5? Alarm bells should have been ringing! White always wants to meet ...c4 with b4, gaining two mobile passed pawns. Does Bd2 make b4 possible? If so, ...Nc5 makes it even more powerful. Incidentally, I showed this game to my German friend Fritz and he made his first comment here. The game was obviously error-free so far... [In the post-mortem we agreed that 19... cxb3 20. axb3 is about equal. However, I recently read GM Jonathan Tisdall say "I always feel that if Black's c-pawn is not on c5...Black's compensation is well on the way to vanishing." I now believe that White's passed pawn is a clear advantage. In addition, Black has to beware of his knight being trapped by f4.] 20. b4! Rxb4 21. Ne2? [ 21. Rxb4 Rxb4 (21... Qxb4 22. Rb1) 22. Nb1! is the route to victory - Fritz.] 21... c3? [Fritz suggests 21... Rxb1! which I planned to answer with the obvious 22. Bxa5 (22. Rxb1 Qc7= Fritz - yeah right!) but 22... R1b2 (22... R8b2? 23. Rxb1!) 23. Qc1 Rxe2 gives Black excellent play and is probably better for him. Look at those knights. I have since learnt that such queen sacrifices are common in the Benko.] 22. Bxc3 Bxe2? [ 22... Rxb1 isn't so attractive now that the passed c-pawn has gone but it is still the best try.] 23. Qxe2? [My German friend sees that 23. Bxb4! wins the exchange, since after my assumed reply 23... Bxd1 24. Bxa5 wins on the spot.] 23... Qb5 The move that Mark had been banking on - he saves the exchange as White's queen is en prise. 24. Rxb4! This had the same effect on Mark as 14...Ng4 had on me. 24... Qxe2 25. Rxb8+ Bf8 We both felt that White was winning here, with two rooks against a queen, a passed extra pawn and Black's pinned bishop. However, I now believe that Black's queen and knights can generate enough counter-play. The transformation from winning to apparently losing had a devastating effect on Mark, one with which I am thoroughly familiar. One day I will show you my defeat to IM Harriet Hunt that followed a similar course to this one. I believe he had given up any real hope of saving the game. 26. Ra1! Fritz prefers 26. Rc1 which indirectly protects both the a- and the e-pawn. 26... Nxe4? (26... Qxa2? 27. Bxe5) 27. Rc2 However, the priority is to push the a-pawn and losing the e-pawn is irrelevant. Furthermore, Black has ...Nd3 ideas so I don't want my rook on c1. 26... f6?! It will take White five moves to queen his a-pawn so Black must try to make every move really count towards distracting him. Playing 26... f6 27... Kf7 and 28... Bh6 takes three moves to achieve nothing particularly useful. 27. a4 Ned3! Mark realises the urgency. 28. Nxd3 Qxd3! I had not even considered this excellent move, which makes my bishop feel very uncomfortable. I gradually became more and more panicked as I realised that I could not extinguish Black's play. 29. Bb4? I had to play something and this attacks his knight... 29... Qd4! Keeping up the pressure , threatening ...Qxa1 and ...Na6; it looks like we both missed the latter! I nearly died when I noticed 29... Nxa4 but 30. Bf1! is the refutation.(30. Rxa4 Qd1+) 30... Qc2 31. Bb5 and the best Black has is 31... Qc7 32. Rxf8+ Kxf8 33. Rxa4 and White has a winning material advantage. 30. Ra3?? By now I had fallen to pieces and variations were racing around my head. Not the ones that mattered. [I was too worried about my a-pawn to play 30. Rb1 but in fact(30... Nd3! would be the best response, using that classic queen and knight combination. if(30... Nxa4 then(31. Rc1! is good for White. The threat is Rcc8. 31... Nc5 32. Bxc5 dxc5 33. Rc8) ) regains the pawn and leaves White playing for the win.White's best chance to play for the win is probably 31.a5 although it looks very risky. 31. Be1 Nxe1 32. Rxe1 Qxa4) loses White's trump card.] 30... Nxe4?? I consider this a symptom of having already given up. [I nearly died again when I saw 30... Nd3 This time it was my South African friend that helped me to see the light -Thabo, not a computer program- 31. Rxd3! Qxd3 32. Bf1 and try as we might we could not stop the a-pawn queening.; My German friend showed me the simple fork 30... Na6! Although I can still fight after 31. Rxf8+ Kxf8 I am fighting for survival, not victory. What a difference it makes for Black to be left with the knight rather than the bishop!] 31. Bxe4 Losing that knight is a major blow. 31... Qxe4 32. a5 White must be winning but Black tries for a perpetual check. 32... Qb1+ 33. Kh2 Qf5 [Try 33... Qf1 to stop the coming fortress.] 34. Be1! Qe4 35. Ra1 Qe5 36. Rbb1! Fortress complete. 36... f5 37. a6 f4 38. a7 fxg3+ 39. fxg3 Qe2+ 40. Kg1 Bg7 41. a8=Q+ Kf7 42. Bf2! Black resigned. 1-0