WCCC 2022

[Eduard]

Round 6 has started

[AdminX]

[pgn]
[Event "World Computer Chess Championship 2022"]
[Site "Vienna, Austria"]
[Date "2022.07.28"]
[Round "8"]
[White "Leela Chess Zero 'LC0'"]
[Black "Komodo Dragon"]
[Result "1/2-1/2"]
[EventDate "2022.07.28"]
[ECO "D02"]

1.d4 d5 2.Nf3 Nf6 3.g3 c5 4.Bg2 cxd4 5.O-O e6 6.Nxd4 e5 7.Nb3 Be6 8.e4 dxe4 9.Qxd8+ Kxd8 10.Nc3 Kc7 11.Be3 Nc6 12.Nxe4 Nxe4 13.Bxe4 h5 14.f4 h4 15.f5 Bc4 16.Rfe1 hxg3 17.hxg3 Rd8 18.Kg2 b6 19.Rh1 Rxh1 20.Rxh1 f6 21.Kf3 a5 22.g4 Bxb3 23.axb3 Nd4+ 24.Bxd4 exd4 25.Rh7 b5 26.Rh5 Bd6 27.Rh7 Rd7 28.Bd3 Kb6 29.Rh8 Kc5 30.Rg8 Re7 31.Rc8+ Kb6 32.Rd8 Kc5 33.b4+ axb4 34.b3 Re3+ 35.Kf2 Kc6 36.Rg8 Re7 37.Kg1 Rc7 38.Be4+ Kc5 39.Kg2 Kb6 40.Bd3 Kc5 41.Kh3 Kd5 42.Kh4 Re7 43.Ra8 Ke5 44.Ra6 Bc7 45.Kh5 Kf4 46.Re6 Rd7 47.Bxb5 Rd5 48.Bd3 Rd7 49.Rc6 Bd6 50.Rb6 Kg3 51.Rb5 Re7 52.g5 fxg5 53.Kxg5 Kf3 54.Rb6 Rd7 55.Kg6 Kf4 56.Bb5 Rd8 57.Be2 Be5 58.Bd3 Rf8 59.Rxb4 Rf6+ 60.Kh7 Rf8 61.Ra4 g5 62.Kg6 g4 63.Ra1 Rg8+ 64.Kf7 Rb8 65.Rf1+ Ke3 66.Re1+ Kf4 67.Re4+ Kxf5 68.Rxd4+ Kg5 69.Re4 Kf5 70.Ra4+ Kg5 71.Ke6 Bc3 72.Rc4 g3 73.Bf1 Bd2 74.Rd4 Be3 75.Rd7 Kf4 76.Bg2 Bf2 77.Kd5 Ke3 78.Kd6 Ke2 79.Kc6 Ke1 80.Kc7 Rf8 81.c4 Be3 82.Bd5 Rf6 83.b4 Rg6 84.c5 Rg4 85.Re7 Kd2 86.Be4 1/2-1/2
[/pgn]

[Eduard]

Leela plays practically without a book. Three book moves is not enough. Leela exchanged too many pieces for nothing. After that, 100 graphics cards don't help to win either.

If Lc0 were my engine, I would develop a book system (which I do for PlayChess for a 3060m laptop card).

Against Komodo a book system with g3, d3, Nd2 would have been better.

[lkaufman]

Looks very likely that we'll have a tie between Dragon and Lc0, and that the result will be essentially decided by lot when colors are drawn for the Armageddon game after the expected six draws. Whoever gets Black will get the draw and thus win the title with over 90% probability, I would estimate. As long as human operators are involved, which requires reasonable increment (at least five seconds), there doesn't seem to be any way to get a decisive result in a short amount of time at this level without grossly favoring one side or starting with a non-standard opening position. Maybe tiebreaks should at least be played on a server like chess.com as bullet matches, in pairs until someone wins.

[dkappe]

As Larry predicted, Dragon got black and converted the Armageddon game. It looked like 20. g5 was premature. Too much contempt?

With the 8xA100, leela was the heavy favorite, but the lack of a real opening book may have hurt it.

[peter]

As Larry predicted, Dragon got black and converted the Armageddon game. It looked like 20. g5 was premature.

Even if this one's in Hiarcs online Abo- Tournament- book yet.
21.Nxg5 would have been the cont. in book instead of Leela's Bxg5 there whith a single one drawn game- example too.

[pgn][[Event "WCCC Armageddon"]
[Site "?"]
[Date "2022.07.05"]
[Round "?"]
[White "LC0"]
[Black "Dragon"]
[Result "1/2-1/2"]
[ECO "C00"]
[PlyCount "181"]
[EventDate "2022.??.??"]
[EventType "blitz"]
[SourceVersionDate "2022.07.29"]

1. Nf3 d5 2. g3 c5 3. Bg2 Nf6 4. O-O e6 5. d3 Nc6 6. Nbd2 Be7 7. e4 O-O 8. Re1
b5 9. e5 Nd7 10. Nf1 a5 11. h4 a4 12. a3 Bb7 13. Bf4 Qb6 14. N1h2 Rfc8 15. h5
h6 16. Qd2 Qd8 17. g4 Ra6 18. Kh1 Rb6 19. Rg1 Nf8 20. g5 hxg5 21. Bxg5 Bxg5 22.
Nxg5 Nxe5 23. Rae1 Ned7 24. Bf3 f6 25. Nh3 Kh8 26. Rg3 Ne5 27. Bg2 Nf7 28. Nf4
Rc7 29. Ng6+ Kg8 30. Ng4 d4 31. Bxb7 Rcxb7 32. Qf4 Nh7 33. h6 Qd5+ 34. Kg1 Qg5
35. Qxg5 fxg5 36. hxg7 b4 37. Re2 e5 38. N6xe5 Re6 39. Rg2 Kxg7 40. f4 gxf4 41.
Nxf7 Rxe2 42. Rxe2 Kxf7 43. Re5 bxa3 44. bxa3 Nf6 45. Nxf6 Kxf6 46. Rxc5 Rg7+
47. Kf2 Rg3 48. Ra5 Rh3 49. Rxa4 Kf5 50. Ra8 Rh2+ 51. Kf3 Rh3+ 52. Kg2 Re3 53.
Kf2 Rh3 54. a4 Kg4 55. Rg8+ Kf5 56. Kg2 Re3 57. a5 Re2+ 58. Kf3 Rxc2 59. a6 Rc7
60. Ra8 Rg7 61. Rh8 Rg3+ 62. Kf2 Rg7 63. Rh5+ Kg4 64. Rd5 Rh7 65. Ra5 Ra7 66.
Ra1 Kf5 67. Kf3 Ke5 68. Ra5+ Kd6 69. Kxf4 Kc7 70. Ra3 Kb6 71. Ra4 Kc5 72. Ra2
Rf7+ 73. Ke5 Ra7 74. Rc2+ Kb5 75. Rb2+ Kc5 76. Ra2 Re7+ 77. Kf4 Ra7 78. Ra4
Rf7+ 79. Ke4 Re7+ 80. Kf5 Ra7 81. Kf4 Rf7+ 82. Ke5 Re7+ 83. Kf6 Ra7 84. Kf5
Rf7+ 85. Ke6 Ra7 86. Ke5 Re7+ 87. Kf5 Ra7 88. Ra2 Kb4 89. Ke4 Kc5 90. Ke5 Re7+
91. Kf4 1/2-1/2[/pgn]

5.d3 was a little tame maybe, c4 there has a somewhat better statistics, hasn't it.

Congrats to Dragon!

[Eduard]

The Armageddon game of Lc0 was strong. The book system played was good (I would have played like that too). The position was extremely closed which is good for Lc0. Maybe just a little luck was missing. The idea of ​​1.h3 before Armageddon was good too. This is how Lc0 should have played against Komodo and Ginkgo in the normal games!!

Incomprehensible to me was Komodo's white-color repetition of the same line played before. In the Repetition seemed to me that the operator had imported the long variant into the book between games. Because in the second game, Komodo didn't waste any time until move 28 (he even got bonus time). Lc0, on the other hand, took 50% of its time. However, the time did not matter.

I play a lot on PlayChess Blitz. If the "correct" variants are chosen, then a 10x faster computer cannot win a game. It is very easy to draw with 2000 kns against 20000 kns. This is why Armageddon is a loss for computer chess because 95% of the time Black will win by drawing!

[Dariusz]

As Larry predicted, Dragon got black and converted the Armageddon game. It looked like 20. g5 was premature.

Even if this one's in Hiarcs online Abo- Tournament- book yet.
21.Nxg5 would have been the cont. in book instead of Leela's Bxg5 there whith a single one drawn game- example too.

[pgn][[Event "WCCC Armageddon"]
[Site "?"]
[Date "2022.07.05"]
[Round "?"]
[White "LC0"]
[Black "Dragon"]
[Result "1/2-1/2"]
[ECO "C00"]
[PlyCount "181"]
[EventDate "2022.??.??"]
[EventType "blitz"]
[SourceVersionDate "2022.07.29"]

1. Nf3 d5 2. g3 c5 3. Bg2 Nf6 4. O-O e6 5. d3 Nc6 6. Nbd2 Be7 7. e4 O-O 8. Re1
b5 9. e5 Nd7 10. Nf1 a5 11. h4 a4 12. a3 Bb7 13. Bf4 Qb6 14. N1h2 Rfc8 15. h5
h6 16. Qd2 Qd8 17. g4 Ra6 18. Kh1 Rb6 19. Rg1 Nf8 20. g5 hxg5 21. Bxg5 Bxg5 22.
Nxg5 Nxe5 23. Rae1 Ned7 24. Bf3 f6 25. Nh3 Kh8 26. Rg3 Ne5 27. Bg2 Nf7 28. Nf4
Rc7 29. Ng6+ Kg8 30. Ng4 d4 31. Bxb7 Rcxb7 32. Qf4 Nh7 33. h6 Qd5+ 34. Kg1 Qg5
35. Qxg5 fxg5 36. hxg7 b4 37. Re2 e5 38. N6xe5 Re6 39. Rg2 Kxg7 40. f4 gxf4 41.
Nxf7 Rxe2 42. Rxe2 Kxf7 43. Re5 bxa3 44. bxa3 Nf6 45. Nxf6 Kxf6 46. Rxc5 Rg7+
47. Kf2 Rg3 48. Ra5 Rh3 49. Rxa4 Kf5 50. Ra8 Rh2+ 51. Kf3 Rh3+ 52. Kg2 Re3 53.
Kf2 Rh3 54. a4 Kg4 55. Rg8+ Kf5 56. Kg2 Re3 57. a5 Re2+ 58. Kf3 Rxc2 59. a6 Rc7
60. Ra8 Rg7 61. Rh8 Rg3+ 62. Kf2 Rg7 63. Rh5+ Kg4 64. Rd5 Rh7 65. Ra5 Ra7 66.
Ra1 Kf5 67. Kf3 Ke5 68. Ra5+ Kd6 69. Kxf4 Kc7 70. Ra3 Kb6 71. Ra4 Kc5 72. Ra2
Rf7+ 73. Ke5 Ra7 74. Rc2+ Kb5 75. Rb2+ Kc5 76. Ra2 Re7+ 77. Kf4 Ra7 78. Ra4
Rf7+ 79. Ke4 Re7+ 80. Kf5 Ra7 81. Kf4 Rf7+ 82. Ke5 Re7+ 83. Kf6 Ra7 84. Kf5
Rf7+ 85. Ke6 Ra7 86. Ke5 Re7+ 87. Kf5 Ra7 88. Ra2 Kb4 89. Ke4 Kc5 90. Ke5 Re7+
91. Kf4 1/2-1/2[/pgn]

5.d3 was a little tame maybe, c4 there has a somewhat better statistics, hasn't it.

Congrats to Dragon!

I join in the congratulations !

[Dariusz]

Noteworthy is the nice result of Shredder, which played on hardware using ARM processors.

[CornfedForever]

Differing hardware is just a wrong approach in my opinion.
As bad as an Armageddon decider? Who know...