Glaurung in Mainz: Part 1/2

[Tord Romstad]

Hi all,

Here is the first part of my report from Mainz. Please note that I am a weak chess player and that I did not use a computer program to analyse the games, my comments to the games may therefore contain serious mistakes. They should be seen as a summary of my observations and feelings during the games, and not as high-quality analysis.

In case some of you actually want to look at the games, you can download a PGN file containing Glaurung's search depths and evaluations (always from white's point of view) here.

Round 1: Glaurung-AICE

Ray Banks, who kindly allowed me to use his quad Opteron in this tournament, had problems with his Internet connection this morning. This forced me to run Glaurung on my MacBook in the first round. This is a smaller handicap than it sounds like: The MacBook (powered by an Intel Core Duo 2 GHz) is a nice computer, and more than half as fast as the quad Opteron.

AICE made a dubious bishop vs knight exchange at g3 at move 4, surrendering the bishop pair and opening white's h file. Just two moves later, AICE made it worse by castling kingside, right into white's open file. Glaurung quickly developed a winning attack.

Round 2: Sjeng-Glaurung

By round 2, Ray's Internet connection was back up again. He later sent me an SMS saying that his connection was once again not working, but fortunately this happened only after the game against Sjeng was finished.

Starting position: bnrbkrqn/pppppppp/8/8/8/8/PPPPPPPP/BNRBKRQN w FCfc - 0 1

1. f4 b6 2. Qf2 e6 3. e3 Ng6 4. O-O?

[diag]bnrbkrq1/p1pp1ppp/1p2p1n1/8/5P2/4P3/PPPP1QPP/BNRB1RKN[/diag]

Just like AICE in the first round, Sjeng makes the mistake of castling too early. Black's pieces are excellently placed for an attack. Glaurung doesn't hesitate:

4... Nh4!? 5. g3 g5! 6. Bh5?

This looks pointless. White's bishop isn't doing anything useful on h5.

6... Be7 7. b3 O-O-O 8. Nc3 f5 9. Bb2?

A waste of time. Many programs play tend to play moves like this in starting positions where a bishop is placed in the corner. The purpose is almost certainly to "develop" the bishop. Of course, with the nice, open diagonal, the bishop is already excellently developed on a1. Moving it to b2 serves no purpose. Glaurung also used to make this particular mistake until a couple of weeks before the tournament, when I finally added some new evaluation code for this pattern.

After white's 9th move, we have the following position:

[diag]bnkr1rq1/p1ppb2p/1p2p3/5ppB/5P1n/1PN1P1P1/PBPP1Q1P/2R2RKN[/diag]

Black is probably close to winning already. White's bishop on h5 and knight on h1 are poorly placed, and black has a very strong attack. Glaurung's score at this stage of the game was already +1.15, which rapidly increases to +2.35 by move 13 and +4.32 by move 16.

9... gxf4 10. Qxf4 a6

This move looks very strange, but perhaps it isn't so bad. White was threatening Nb5, which forks the a7 and c7 squares.

11. Be2 Qg6 12. b4 Bg5 13. Qc4 d5 14. Qb3 d4

Now there is no longer any doubt that black is winning.

15. Nd1 dxe3 16. dxe3 Rd2 17. Rf2

[diag]bnk2r2/2p4p/pp2p1q1/5pb1/1P5n/1Q2P1P1/PBPrBR1P/2RN2KN b - - 2 17[/diag]

17... Rxe2!

A pretty combination, but of course a very easy one to find for a computer. The rest is simple.

18. Rxe2 Qh5 19. Qxe6+ Nd7 20. Kf1 Qf3+ 21. Rf2 Qxh1+ 22. Ke2 Ng2 23. Rxf5 Qe1+ 24. Kd3 Rxf5 25. Qxf5 Qxb4 26. Bd4 Ne1+ 27. Ke2 Bf3+ 28. Qxf3 Nxf3,

and black won after 41 moves.


Round 3: Glaurung-Jonny

This time, Jonny had problems with the remote hardware. After some consultation with Ingo (the chief arbiter), we decided to both use our local machines to make it a fair game.

Starting position: rnqnbkrb/pppppppp/8/8/8/8/PPPPPPPP/RNQNBKRB w GAga - 0 1

1. d3 O-O 2. O-O d6 3. Bc3 f5 4. g3 Bh5 5. Qd2

[diag]rnqn1rkb/ppp1p1pp/3p4/5p1b/8/2BP2P1/PPPQPP1P/RN1N1RKB[/diag]

I was not happy with Glaurung's opening play in this game. It plays too passively in the centre, and the pieces are poorly coordinated. How is white supposed to develop the knights? Black's position is also somewhat unharmonic. In order to activate the h8 bishop, he has to play g6 sooner or later, but this leaves the other bishop at h5 out of play.

5... Nbc6 6. f4 e6 7. Nf2 Qd7 8. Bg2 Re8 9. b4 Ne7 10. a4 d5 11. e3

[diag]r2nr1kb/pppqn1pp/4p3/3p1p1b/PP3P2/2BPP1P1/2PQ1NBP/RN3RK1[/diag]

White seems to be a bit better, although Glaurung's evaluation of 0.62 is probably on the optimistic side of the truth.

11... Nf7 12. Qc1 Ng6!?

During the game, Johannes and I both thought this was an ugly move. Both of black's bishops now look really silly. As it turns out, however, Glaurung is unable to make use of this fact, and after a few lame moves by my program (which still had an eval around +0.6 at this point of the game) white's advantage evaporates.

13. Nd2 e5 14. Qb2 c6 15. a5 Rad8 16. Qa2 d4!

Black has at least equal play.

17. exd4 exf4 18. a6 b6 19. Nc4 f3 20. Bh3

[diag]3rr1kb/p2q1npp/Ppp3n1/5p1b/1PNP4/2BP1pPB/Q1P2N1P/R4RK1[/diag]

A messy and unclear position. The pawn on f3 is a major problem for white, but black still has to solve the problem of the development of his minor pieces.

20... Qd5 21. Rfe1 Rxe1 22. Bxe1 Ne7 23. c3 g5 24. Bd2 g4 25. Bf1

I'm not sure exactly where white went wrong, but the position now looks better for black. The passed pawn on f3 is well protected, and the h8 bishop has finally entered the game. Glaurung, for some reason, is still optimistic, and gives a 0.00 score.

25... Bf6 26. Qb2 Ng6 27. Ne3 Qd7

[diag]3r2k1/p2q1n1p/Ppp2bn1/5p1b/1P1P2p1/2PPNpP1/1Q1B1N1P/R4BK1[/diag]

28. b5!?

A strange-looking pawn sacrifice. White doesn't seem to get much for the pawn except a passed pawn on d4, but because it looks difficult to advance the central pawns it is hard to see how this could give sufficient compensation. Surprisingly, Glaurung evaluated this position as +0.07 for white, and for the rest of the game the eval never drops more than a few centipawns below zero. Jonny, IIRC, thought that black had a plus score of almost a full pawn here. This looks more realistic to me.

28... cxb5 29. Rb1 Nd6 30. Nd5 Qf7 31. Nxf6+ Qxf6 32. Qc1 Rc8 33. Qa3 Rc6 34. Re1 Qf7 35. Qa1 Ne7 36. Re5 Bg6 37. Qe1 Nd5 38. Kh1 Rc8 39. Kg1 h5 40. Re6 Re8 41. Rxe8+ Nxe8 42. c4

At last, white is able to start pushing his central pawn mass, and obtain some compensation for the pawn. It is hard to believe that black really had to allow this: I guess black must have had better moves somewhere between move 30 and 40.

42... Ndc7 43. Qa1 Qf6 44. Bf4 h4 45. Be5 Qe7 46. gxh4 Qxh4 47. Qc1 Kh7 48. d5 bxc4 49. dxc4

[diag]4n3/p1n4k/Pp4b1/3PBp2/2P3pq/5p2/5N1P/2Q2BK1[/diag]

Black still has the extra pawn, but the passed pawn on d5 is now really dangerous, and white looks better. Glaurung had +0.48 for white here, which rises to +0.75 at moves 51 and 52. But Jonny defends precisely, and we quickly reach a drawn knight endgame.

49... Kg8 50. Qd2 Qe7 51. Qf4 Nxa6 52. Bd3 Qb4 53. Kf1 Nc5 54. Bxf5 Bxf5 55. Qg5+ Kf7 56. Qxf5+ Ke7 57. Qh7+ Kf8 58. Qc2 b5 59. Nxg4 Qxc4+ 60. Qxc4 bxc4 61. Ne3 Nd3 62. Nxc4 Nxe5 63. Nxe5 a5 64. Nxf3 Ke7 65. Nd2 Kd6 66. Nc4 Kc5 67. Nxa5 Nf6 68. Kg2 Ne4 69. Kf3 Nc3 70. Kf4 Nxd5+ 71. Ke5 1/2-1/2

An interesting and hard-fought game.


Round 4: Shredder-Glaurung

From round 4 until the end of the tournament, Glaurung played on Ray's quad Opteron.

The game against Shredder was probably Glaurung's worst game of the tournament. It played some highly dubious pawn moves in the opening, and was left with fatally weak squares on g5, g6 and d5. Shredder played a nice positional game and had no problems in exploiting black's numerous weaknesses.


Round 5: Glaurung-The Baron

This was a very short win for Glaurung, but more because of unexpectedly poor play by The Baron than because of strong play by Glaurung. The Baron castled queenside and encouraged Glaurung to attack by unnecessarily weakening the pawn shelter in front of the king. By move 20, Glaurung had a +1.57 score and probably a winning attack. Two moves later, The Baron made what looked like an elementary tactical blunder, and Glaurung's score instantly jumped to +5.85. The Baron resigned a few moves later.

Efficient attacking technique by Glaurung, but not a very memorable game. I suspect that The Baron had a bug or two.

The second part of my report, containing rounds 6-9, will be posted in a day or two.

Tord

[Volker Pittlik]

Thank you Tord for this detailed report!

Posts like yours and Aleksandar's give me a very good impression of the event. Some kind of TV - working by imagination .

Volker

[mike schoonover]

hi tom,
thanks for the report.
regards
mike

[Tuvoc]

Thanks Tord, a great write-up ! I'm eagerly awaiting the 2nd instalment