Here is the second part of my report from Mainz. A PGN file containing Glaurung's games with search depths and evaluations can be found on my home page.
In the first round of day two, I was paired against two of my friends from last years tournament: Fritz and Viktoria of the Loop team. Viktoria instantly fell in love with my MacBook. I hope she will succeed in convincing Fritz that the Loop team needs a MacBook, because this would increase the chances of a Mac version of Loop.
Round 6: Loop-Glaurung
Initial position: rkqrbnnb/pppppppp/8/8/8/8/PPPPPPPP/RKQRBNNB w DAda - 0 1
By far the most exotic-looking starting position of the tournament so far.
1. d4 d6
[diag]rkqrbnnb/ppp1pppp/3p4/8/3P4/8/PPP1PPPP/RKQRBNNB[/diag]
2. a4?
A very serious mistake. It is very difficult for white to castle kingside, and white's second move renders queenside castling undesirable. This, in turn, makes it difficult to activate the a1 rook.
Is 2. a4 already the decisive mistake? It seems hard to believe, but I can't find many obvious mistakes in the rest of the game, and Loop never recovers after the unforunate opening. One of my biggest surprises in this tournament was how often the game is decided in one of the first few moves. Against AICE and Sjeng, Glaurung was probably winning before move 10, and it is quite possible that this was the case against Loop as well.
2... e5! 3. dxe5?
Another mistake. After the opening of the d file, kingside castling for white is not only difficult, but virtually impossible.
3... dxe5 4. Nf3 g6 5. Bc3 f6 6. Ne3 Ne6 7. Rxd8 Qxd8 8. g3 Bc6 9. Nd2 Ne7 10. Ra3 Qe8 11. Ndc4
[diag]rk2q2b/ppp1n2p/2b1npp1/4p3/P1N5/R1B1N1P1/1PP1PP1P/1KQ4B[/diag]
The position would have been roughly equal, if not for the fact that black can still castle.
11... Nd4! 12. Na5 Nxe2 13. Qf1 Bxh1 14. Qxh1 Nxc3+ 15. Rxc3 c6 16. Rd3 O-O-O
[diag]2krq2b/pp2n2p/2p2pp1/N3p3/P7/3RN1P1/1PP2P1P/1K5Q[/diag]
The rest is technique.
17. Qd1 b6 18. Nc4 Rxd3 19. Qxd3 Qd7 20. Qc3 Bg7 21. f4 Qd4 22. Qb4 Kd7 23. fxe5 fx5 24. c3 Qd3+ 25. Ka2 Ke6 26. Qb3 Nd5 27. a5 b5 28. Nxd5 Qxd5 29. Ne3 Qxb3+ 30. Kxb3 Bh6 31. Ng4 Bg5 32. Kc2 Kf5 33. Nf2 Be3 34. Nd3 c5 35. g4+ Ke4 36. b3 c4 37. bxc4 bxc4 38. Nb2 Bf4,
and Ingo finally allowed white to resign 10 moves later.
Round 7: Glaurung-Ikarus
Glaurung seemed to get a tiny advantage in the opening, but gave it away after a dubious-looking gambit. There followed some strange tactical complications, which ended in a materially unbalanced position where Glaurung had a rook for a bishop and pawn, but had serious difficulties in completing the development. The position was difficult to evaluate, but Ikarus played very well in this phase of the game and soon proved that black had a clear advantage. After 27 moves, we had the following comical position:
[diag]8/kp3p2/6p1/2p4p/p2b4/Pr1q2P1/RPR2P1P/1KQ5[/diag]
Ikarus won easily.
Round 8: Naum-Glaurung
Alex (who is a good friend) and I had spent the whole tournament hoping that we would be paired against each other, and in round 8 we finally got the chance.
Initial position: nrqbbkrn/pppppppp/8/8/8/8/PPPPPPPP/NRQBBKRN w GBgb - 0 1
1. O-O e5 2. Nb3 Ng6 3. d4 exd4?!
To me, 3... d6 looked better, but of course I am just a patzer.
4. Nxd4 d5 5. c4 dxc4 6. Qxc4 c5?
A really awful move. Black neglects the development, and the pawn on c5 becomes a weakness. It also doesn't help that the d5 square gets weak. What was wrong with the simple 6... Nb6?
7. Rc1 b6 8. b4 Be7 9. bxc5 bxc5 10. Qc2 Bd7 11. Nb3 c4 12. Qxc4 Qxc4 13. Rxc4 O-O
[diag]nr3rk1/p2bbppp/6n1/8/2R5/1N6/P3PPPP/3BBRKN[/diag]
After a rather simple sequence of moves, white has succeeded in capturing black's weak c pawn. Black's only compensation is a lead in development. For me, it is hard to believe that this is enough. Glaurung seems to agree, giving +0.65 for white here. As it turns out, however, white quickly gets into trouble. Whether this is because black's activity is worth more than it looks like, or because Naum plays badly, is difficult for me to say.
14. Ng3 Rfc8 15. Rxc8+ Rxc8 16. Nd4 Bf6 17. Ndf5 Be6 18. a4 Nb6 19. e4
Nc4 20. Nh5 Bb2 21. Bg4 Rd8 22. Be2 h6 23. Nfg3 Rd4 24. f4 Ne3
[diag]6k1/p4pp1/4b1np/7N/P2rPP2/4n1N1/1b2B1PP/4BRK1[/diag]
The position is beginning to look promising for black. Naum decides to exchange down to an endgame with an interesting material imbalance:
25. f5!? Nxf1 26. Kxf1 Rxa4 27. fxg6 Ra1 28. Nf4 Bc3 29. Nd3 fxg6
[diag]6k1/p5p1/4b1pp/8/4P3/2bN2N1/4B1PP/r3BK2[/diag]
The dust has settled, and it looks like black is better. White has two knights against a rook and a pawn, but black's passed pawn looks very dangerous.
30. Kf2 Bd4+ 31. Kf3 a5 32. e5 Bd5+
With a big jump in score: At move 31, Glaurung had +0.84 for black, which increased to +1.75 at move 32. It turns out, however, that Glaurung has chosen an over-optimistic and too greedy plan. It plans to grab white's two remaining pawns on the kingside, but by doing so he allows white to activate his king and create really serious threats with the passed e pawn.
33. Kf4 g5+ 34. Kf5 Bxg2 35. e6 Bh3 36. Ke4 Bg1 37. Kd5 Bxh2 38. e7 Bd7 39. Bg4 Be8
[diag]4b1k1/4P1p1/7p/p2K2p1/6B1/3N2N1/7b/r3B3 w - - 3 40[/diag]
A very interesting position. Glaurung still thinks it is winning (+1.92 here), but the score quickly drops in the next few moves. Naum, if I recall correctly, thought white was better. I don't know what to think. Black has a material advantage, but his pieces are poorly coordinated, and the black king is difficult to activate. White's minor pieces are beautifully placed, his king is active, and the passed pawn on e7 is very dangerous.
40. Bc3 Rb1 41. Be6+ Kh7 42. Nf5 Rb5+ 43. Nc5 Bg1 44. Bd4 Bxd4 45. Kxd4 Rb6 46. Bd7 Bg6 47. Nd3 Bh5 48. Kc5 Rb8 49. Nd6 a4 50. Bxa4 Ra8
[diag]r7/4P1pk/3N3p/2K3pb/B7/3N4/8/8[/diag]
It is beginning to become clear that white is clearly better, and as the game continuation proves, the advantage is sufficient to win. There is little doubt that black was better around move 30. Exactly where was the turning point? I have no idea.
51. Bc6 Ra7 52. Bb7 Rxb7 53. Nxb7
[diag]8/1N2P1pk/7p/2K3pb/8/3N4/8/8[/diag]
Both engines had a winning score for white. Glaurung had -3.00, I don't remember exactly what Naum showed. The two authors were less convinced. How can white promote the e7 pawn without allowing black to sacrifice the bishop and reach a drawn endgame?
53... g4 54. Nd6 g3 55. Nf4 g2 56. Nxg2 g5
With a score of -12.62. Naum displayed something even bigger, if I recall correctly. With such scores, there is of course no doubt that the engines have seen a forced win. Because the stupid authors still didn't see how white could win, we played on for a few more moves:
57. Ne3 Kg7 58. Ng4 Bg6 59. Kc6 Bh5 60. Nf5+ Kg6 61. e8=Q+
At last, it is obvious even to us lowly patzers that white is winning, and Ingo kindly allowed black to resign.
Despite the numerous mistakes by both sides, this was the most fascinating and entertaining game I played in the tournament. Losing games like this is far more fun than winning games like the one against Baron in the 5th round!
Round 9: Glaurung-Xinix
Xinix made a dubious pawn push in the early opening, and was under heavy pressure before move 10. It managed to relieve some of the pressure by exchanging queens, but only by accepting a really ugly pawn structure. This is the position after black's 20th move:
[diag]3k4/1r1p1r1p/1b3pn1/p1p2p2/R4P2/1P2P3/1BPP3P/2K1NR2[/diag]
From this point, however, Xinix started defending exceptionally well. Glaurung finally managed to win a pawn, but was forced to allow so many pawn exchanges that we emerged with a theoretically drawn KRPP vs KRP endgame. I am fairly sure Glaurung must have been winning at some stage of the game, but it is hard to say exactly where it let the half point slip. At any rate, Xinix displayed outstanding defensive technique, and fully deserved the draw.
Glaurung ended at place 6 with 5/9, after wins against AICE, Sjeng, The Baron and Loop, losses against Shredder, Ikarus and Naum, and draws against Jonny and Xinix. The result itself is decent, but I was somewhat disappointed with the games. Glaurung did not succeed in winning any really great games. The best game was probably the win against Sjeng, but that game was marred by Sjeng's unfortunate opening mistakes.
At any rate, the tournament was tremendous fun, just like last year. I already look forward to coming back in 2007. Thanks a lot to the organizers, the other programmers, and to Ryan (whose machine I used) for making this tournament such a pleasant experience!
Tord
