1. d4 Nf6 2. c4 g6 3. Nc3 Bg7 4. e4 O-O 5. f4 d6 6. Nf3 a6 7. a3

I wasn't really satisfied with the position I got with 7.a4 against Yermolinsky, so I tried this suggestion of Vaisser's. However, I don't think Black is particularly troubled by this move.

The point of a3 is that after 7... c5 8. dxc5 dxc5 (8... Qa5 9. b4 shows the utility of a3.) 9. Qxd8 Rxd8 10. e5 White should be able to claim a small endgame edge, but Black even managed to win after this continuation in the game Filipovic-Goric Croatia 2000.

9. e5 trying to gain some space looks more to the point.

This seems much more logical than 10...Bb7, which was played in the game Dagaeva-Chuvatkina, St. Petersburg 1999.

Better is 11. exd5 cxd5 with approximate equality. I wasn't spending a lot of time on my moves here (time control was G/70), and superficially thought that my weakness on d4 would be balanced by his on c6. However, Black can quickly bring a lot of fire power against d4 (Qb6, Rd8, Bg4, Bg7) and it isn't clear how White can generate similar play versus c6.

11... dxe4 12. Nxe4 Bg4

now I had a good long think.

I looked at various ways to sac the d-pawn for the 2 bishops and leave Black with the c6 weakness. The problem is that Black can maintain pressure on b2 making it troublesome to develop the queenside, for example 13. Nf2 Bxf3 14. Qxf3 Qxd4 15. Rd1 Qb6 with slight advantage to Black since the move White would like to play 16. b4 is met by 16... Nd5 and 20...Nc3

[16. Qxf3? Qb6 and White can't guard both b3 and d4.]

16... a5 17. Be4?!

After this move, I start getting in deeper trouble. White could maintain close to equality with 17.a4. While this move weakens the b4 square, White is finally ready to counterattack the weak Black pawns with Bd2 and Rc1. A move ...Nb4 by Black at some point can be met by Bxb4 causing opposite colored bishops, White if Black responds ...axb4 White will have a passed a-pawn, while if Black recaptures with a piece, Be4 and Rfc3 will further pressure c6. The White knight can find a home on c5 in many lines as well.

The engines prefer 20... Ra5 looking to win the c5 pawn outright and maintaining all of Black's other advantages (passed d-pawn, a4 and Bg7 clamping down White's queenside).

21. Qxd5 Rfd8 22. Qc4 Rd1+ 23. Kh2 Qe1 24. Re3 Qg1+ 25. Kg3 Bd4

26. Re2

not 26. Rxe7?? Qf2+ 27. Kh2 (27. Kg4 Qxg2+ 28. Kh4 Bf2#) 27... Rh1+ 28. Kxh1 Qg1# or 26.Rf3?? Qe1+ 27. Kg4 (27. Kh2 Bg1+ 28. Kh1 Bf2+ 29. Kh2 Qh1#) h5+ 28. Kg5 Kg7 with mate shortly.

26... Bxc5?

Regaining the pawn, but allowing White a tactical escape. Black keeps a big edge with 26... Qf1 when it seems that White must sacrifice the exchange with 27. Be3 Rxa1 28. Bxd4 since 27. Qc2 Rb8 -+ looks deadly

27. Be3!

This blow equalizes the game.

27...Bxe3 28. Rxd1 Bf2+ 29. Rxf2 Qxd1 30. Qc2 Qd6 31. Rd2 Qf6 32. Qe4 Ra5 33. Rd4 e5 34. Rxa4 Rxa4 ½:½

35. Qxa4 exf4+ 36. Qxf4 Qxb2 37. a4 the outside passed pawn means little since White can hardly prevent perpetual check. Compare this to the queen ending I played against Alexander Ivanov, where he was even willing to give an outside passed pawn in an attempt to keep the game going.