Mscp 1.6 small bug fix update & great game against Glaur

[Jim Ablett]

Found a small bug in the evaluation and made a few
changes to King endgame.
Download from my homepage.

Mscp just played a great game against Glaurung
which ended in a draw, but I think Mscp could have won
perhaps. It had me on the edge of my seat watching!
I was willing Mscp on to win, but it just couldn't do it.
Maybe if Mscp used tablebases it could have found a win?

Here's the pgn >

[Event "Arena tournament"]
[Site "TYRELL-XBHMTDAN"]
[Date "2005.02.26"]
[Round "1"]
[White "Glaurung 2.1(bh)"]
[Black "Mscp 1.6"]
[Result "1/2-1/2"]
[ECO "E16"]
[Opening "Queen's Indian"]
[Time "13:08:13"]
[Variation "Capablanca, 6.Bd2 c5"]
[TimeControl "40/300:40/300:40/300"]
[Termination "normal"]
[PlyCount "187"]
[WhiteType "program"]
[BlackType "program"]

1.c4 Nf6 2.d4 e6 3.Nf3 b6 4.g3 Bb7 5.Bg2 Bb4+ 6.Bd2 c5 7.O-O O-O 8.Bxb4 cxb4
9.Qd3 d6 10.Nbd2 a5 {(10. ... a7a5 11.e2e4 Nb8d7 12.Nf3g5 h7h6 13.Ng5f3 Qd8e7
14.Ra1d1 Ra8d8 15.Rf1e1 Rf8e8 16.Qd3e3 Nf6g4) -0.32/13 12} 11.e4 {(1109423334) +0.91/7
9} e5 {(11. ... e6e5 12.Nf3g5 Rf8e8 13.Ra1e1 Qd8c7 14.f2f4 e5xd4 15.Ng5f3 Nb8c6
16.Re1d1 Ra8d8 17.Nf3xd4 Nc6xd4 18.Qd3xd4) -0.28/12 8} 12.a3 {(1109423370) +0.93/8 26}
bxa3 {(12. ... b4xa3 13.Ra1xa3 Nb8c6 14.Ra3b3 Nc6b4 15.Qd3e2 Qd8e7 16.Rf1d1 Ra8e8
17.Rb3e3 e5xd4 18.Nf3xd4 Qe7e5) -0.17/13 12} 13.Rxa3 {(1109423415) +0.90/8 30} Nc6 {(13.
... Nb8c6 14.Rf1d1 Qd8c7 15.Ra3b3 Bb7c8 16.d4xe5 d6xe5 17.Qd3e3 Nc6b4 18.Nf3g5
Rf8d8 19.f2f4 Bc8g4) -0.16/13 10} 14.Rfa1 {(1109423475) +0.86/8 47} Nb4 {(14. ...
Nc6b4 15.Qd3b1 Qd8c7 16.Ra3c3 Rf8e8 17.d4d5 Bb7c8 18.Qb1e1 Bc8g4 19.Qe1e2 Qc7c5
20.Nd2b3 Qc5c7) 0.00/13 9} 15.Qb1 {(1109423496) +0.67/7 10} Qe7 {(15. ... Qd8e7 16.Ra3e3
Ra8e8 17.d4d5 Qe7c7 18.Bg2h3 Bb7c8 19.Bh3xc8 Re8xc8 20.Qb1d1 Qc7c5 21.Qd1e2 Rf8e8)
0.00/13 8} 16.d5 {(1109423538) +0.67/8 32} Ng4 {(16. ... Nf6g4 17.Qb1d1 Bb7c8 18.Nf3h4
Ng4f6 19.Ra3e3 Bc8d7 20.Nh4f5 Bd7xf5 21.e4xf5 Qe7d7 22.Qd1f3) 0.00/12 7} 17.h3
{(1109423551) +0.94/7 4} Nf6 {(17. ... Ng4f6 18.Ra3e3 Bb7a6 19.b2b3 Qe7c7 20.g3g4 Ra8d8
21.Qb1c1 Ba6b7 22.Qc1c3 Qc7c5 23.Ra1d1 Rf8e8) 0.00/13 7} 18.Qc1 {(1109423564) +0.87/7 4}
Rad8 {(18. ... Ra8d8 19.Qc1c3 Rd8e8 20.Ra1d1 Bb7c8 21.g3g4 Bc8d7 22.Qc3b3 Re8d8
23.g4g5 Nf6h5 24.h3h4) +0.02/12 11} 19.Qc3 {(1109423582) +0.95/7 5} Qc7 {(19. ... Qe7c7
20.Nf3g5 Qc7c5 21.Nd2b3 Qc5c7 22.Nb3d2) 0.00/12 8} 20.Nh4 {(1109423597) +0.96/7 6} Bc8
{(20. ... Bb7c8 21.Ra1e1 Nf6h5 22.Re1c1 Qc7c5 23.Nd2b3 Qc5c7 24.Nb3d2) 0.00/12 9}
21.Nhf3 {(1109423614) +0.94/7 6} Bd7 {(21. ... Bc8d7 22.Ra1e1 Rd8c8 23.g3g4 Qc7c5
24.Re1e3 Rf8e8 25.Ra3a1 Rc8d8 26.g4g5 Nf6h5 27.Ra1e1) +0.03/12 6} 22.Nh2 {(1109423633)
+0.87/7 10} Rfe8 {(22. ... Rf8e8 23.Nh2f3 Rd8c8 24.g3g4 Re8e7 25.Ra1e1 Qc7c5 26.Re1e3
Rc8e8 27.g4g5 Nf6h5 28.h3h4) +0.03/12 7} 23.Nhf1 {(1109423648) +0.96/7 6} Qc5 {(23.
... Qc7c5 24.Nf1e3 Qc5d4 25.Ra3b3 Re8e7 26.f2f3 Nf6h5 27.Kg1h2 Rd8e8 28.Qc3xd4
e5xd4 29.Ne3g4 Bd7xg4 30.f3xg4) +0.01/12 6} 24.Ne3 {(1109423661) +1.03/7 5} Qd4 {(24.
... Qc5d4 25.Ra1b1 Re8e7 26.Nd2b3 Qd4d3 27.Qc3xd3 Nb4xd3 28.Nb3xa5 Nd3c5 29.Na5c6
Bd7xc6 30.d5xc6 Nc5xe4 31.Bg2xe4 Nf6xe4) 0.00/12 6} 25.Rb3 {(1109423676) +0.94/7 7}
Rdb8 {(25. ... Rd8b8 26.Rb3a3 Qd4xc3 27.Ra3xc3 Rb8d8 28.f2f3 Re8e7 29.Ra1d1 Rd8b8
30.Kg1f2 Rb8e8 31.Rc3c1) +0.05/12 6} 26.Qxd4 {(1109423688) +1.06/7 4} exd4 {(26. ...
e5xd4 27.Ne3d1 Nb4a6 28.Rb3d3 Na6b4 29.Rd3b3) 0.00/14 5} 27.Nef1 {(1109423698)
+0.83/7 3} Na6 {(27. ... Nb4a6 28.Rb3d3 Na6b4 29.Rd3b3) 0.00/13 5} 28.Rba3
{(1109423708) +1.07/7 3} Nb4 {(28. ... Na6b4 29.Ra1b1 Nf6h5 30.Nf1h2 Nh5f6 31.g3g4 d4d3
32.Ra3c3 Rb8d8 33.Rb1d1 Re8e7 34.g4g5 Nf6h5) -0.02/13 8} 29.Rc1 {(1109423721) +1.07/7 3}
Nh5 {(29. ... Nf6h5 30.Nf1h2 Nh5f6 31.g3g4 Re8e7 32.Nh2f1 h7h5 33.g4g5 Nf6h7
34.f2f4 Rb8e8 35.Nf1g3 g7g6) -0.10/13 9} 30.Nb3 {(1109423735) +1.28/7 2} d3 {(30. ...
d4d3 31.Nb3d4 Re8e7 32.Nf1d2 Rb8e8 33.Rc1d1 g7g6 34.Ra3c3 Nh5f6 35.f2f3 Bd7a4
36.Rd1e1 Ba4d7) -0.10/13 6} 31.Ne3 {(1109423746) +1.15/7 3} Nf6 {(31. ... Nh5f6 32.Nb3d2
Rb8d8 33.f2f3 Bd7c8 34.Kg1f2 Nf6d7 35.h3h4 Nd7e5 36.f3f4 Ne5g4+ 37.Ne3xg4 Bc8xg4
38.Bg2f3) -0.20/13 11} 32.Nd2 {(1109423763) +1.09/7 5} Rbd8 {(32. ... Rb8d8 33.f2f3 Bd7c8
34.h3h4 Nf6d7 35.Bg2h3 Bc8b7 36.Kg1f2 Nd7e5 37.Bh3f5 f7f6 38.Rc1d1) -0.30/12 6} 33.g4
{(1109423775) +1.45/7 4} Bc8 {(33. ... Bd7c8 34.f2f3 Nf6d7 35.Ne3f5 Nd7c5 36.Nf5d4 Bc8d7
37.Kg1f2 Re8e5 38.Kf2e3 Rd8e8 39.f3f4 Re5e7 40.Rc1d1) -0.22/14 11} 34.Rcc3 {(1109423789)
+1.50/6} Nd7 {(34. ... Nf6d7 35.f2f3 Nd7c5 36.Kg1f2 Bc8d7 37.Rc3c1 Re8e7 38.Ne3f5 Bd7xf5
39.g4xf5 Nb4c2 40.Ra3a2 Rd8e8 41.Rc1g1) -0.12/14 8} 35.f3 {(1109423802) +1.38/7 3} Nc5
{(35. ... Nd7c5 36.Kg1f2 Bc8d7 37.Rc3c1 Re8e5 38.Kf2e1 Rd8e8 39.f3f4 Re5e7 40.Bg2f3
f7f6 41.f4f5 Bd7a4 42.b2b3 Ba4d7) -0.09/15 8} 36.Bf1 {(1109423816) +1.31/7 4} Bd7
{(36. ... Bc8d7 37.Rc3c1 Re8e7 38.Bf1g2 Rd8e8 39.Kg1f2 f7f6 40.Kf2e1 Nc5a4 41.Ne3f5
Nb4c2+ 42.Ke1f2 Bd7xf5 43.Ra3xa4 Bf5d7 44.Ra4a2) -0.07/15 11} 37.b3 {(1109423834)
+1.17/7 5} Re5 {(37. ... Re8e5 38.Bf1g2 Rd8e8 39.Kg1f2 Kg8f8 40.Kf2e1 Kf8g8 41.f3f4
Re5e7 42.f4f5 Re7e5 43.Ra3a1 Re8e7 44.Ra1d1) -0.07/14 8} 38.Ng2 {(1109423850) +1.09/7
5} Nc2 {(38. ... Nb4c2 39.Ra3a2 Nc2b4 40.Ra2a3) 0.00/13 9} 39.Ra2 {(1109423862)
+1.17/7 2} Nb4 {(39. ... Nc2b4 40.Ra2a3) 0.00/14 15} 40.Ra1 {(1109423885) +1.35/7 6}
Nc2 {(40. ... Nb4c2 41.Ra1d1 Nc2b4 42.Rd1a1) 0.00/15 15} 41.Rb1 {(1109423945)
+1.61/8 43} Nb4 {(41. ... Nc2b4 42.Rb1b2 Rd8b8 43.Nd2b1 b6b5 44.c4xb5 Bd7xb5 45.Ng2e3
g7g6 46.Nb1d2 a5a4 47.Nd2c4 Bb5xc4 48.Ne3xc4) -0.08/14 14} 42.Rb2 {(1109424013)
+1.58/8 52} h5 {(42. ... h7h5 43.Rc3c1 Rd8e8 44.Ng2f4 h5xg4 45.h3xg4 g7g5 46.Nf4h5
Re8e7 47.Nh5f6+ Kg8g7 48.Nf6xd7 Nc5xd7) -0.13/11 8} 43.Ne1 {(1109424031) +2.64/7 8}
hxg4 {(43. ... h5xg4 44.h3xg4 b6b5 45.Rc3c1 Rd8c8 46.c4xb5 Bd7xb5 47.Rc1a1 Rc8b8
48.Bf1g2 a5a4 49.f3f4) -0.14/12 7} 44.hxg4 {(1109424055) +2.74/7 15} b5 {(44. ... b6b5
45.Rc3c1 Rd8c8 46.Rc1a1 Rc8a8 47.c4xb5 Bd7xb5 48.Bf1g2 Re5e8 49.Kg1f2 Re8e7 50.Rb2b1
Bb5d7) -0.24/13 11} 45.Kf2 {(1109424075) +2.03/7 7} a4 {(45. ... a5a4 46.b3xa4 Nc5xa4
47.Rc3b3 Na4xb2 48.Rb3xb2 b5xc4 49.Nd2xc4 d3d2 50.Rb2xd2 Re5e7 51.Nc4xd6) -0.31/12 13}
46.bxa4 {(1109424116) +2.25/8 27} Nxa4 {(46. ... Nc5xa4 47.Rc3b3 Na4xb2 48.Rb3xb2 b5xc4
49.Nd2xc4 d3d2 50.Nc4xd2 Nb4xd5 51.e4xd5 Re5xd5 52.Bf1c4 Rd5e5 53.Rb2b7) -0.30/14 9}
47.Rxb4 {(1109424133) +0.64/8 6} Nxc3 {(47. ... Na4xc3 48.Ne1xd3 Re5e7 49.c4c5 d6xc5
50.Nd3xc5 Bd7e8 51.Nd2b1 Nc3xb1 52.Rb4xb1 Re7c7 53.Rb1c1) +0.35/12 4} 48.Bxd3
{(1109424158) +0.99/8 19} Rb8 {(48. ... Rd8b8 49.Kf2e3 Re5e8 50.Bd3c2 f7f6 51.Ke3d4 Nc3e2+
52.Kd4e3 Ne2g1 53.Bc2d3 b5xc4 54.Rb4xb8 Re8xb8 55.Bd3xc4) +0.36/12 8} 49.cxb5
{(1109424196) +0.79/8 28} Nxb5 {(49. ... Nc3xb5 50.Nd2c4 Re5e8 51.Nc4a5 Kg8f8 52.Kf2e3 Kf8e7
53.Ke3f4 Nb5a7 54.Rb4xb8 Re8xb8 55.Na5c4 Na7b5 56.Nc4e3) +0.55/14 10} 50.Nc4
{(1109424221) +0.92/8 13} Ree8 {(50. ... Re5e8 51.Bd3c2 Re8d8 52.Bc2a4 Bd7e8 53.Ne1d3 Kg8f8
54.Kf2e3 Kf8e7 55.Ke3f4 Be8d7 56.Nc4a5 Ke7f6 57.Na5c6) +0.53/14 6} 51.Nc2 {(1109424244)
+0.92/8 15} Rec8 {(51. ... Re8c8 52.Nc4a5 Kg8f8 53.Na5c6 Bd7xc6 54.d5xc6 Nb5c7
55.Rb4xb8 Rc8xb8 56.Nc2d4 Kf8e8 57.Nd4f5 Rb8b2+ 58.Kf2e3) +0.56/13 8} 52.N2e3
{(1109424258) +1.13/7 3} Kf8 {(52. ... Kg8f8 53.Nc4a5 Nb5a3 54.Rb4xb8 Rc8xb8 55.Ne3f5 Rb8b2+
56.Kf2e3 Bd7xf5 57.g4xf5 Na3c2+ 58.Ke3f4 Nc2b4 59.Bd3c4 Kf8e7) +0.64/13 8} 53.Ra4
{(1109424272) +0.99/7 4} g6 {(53. ... g7g6 54.Ra4a6 Kf8e7 55.Nc4b6 Rc8c3 56.Nb6xd7 Ke7xd7
57.Bd3c4 f7f6 58.Ra6c6 Nb5d4 59.Rc6a6) +0.67/12 5} 54.Ra6 {(1109424282) +1.08/7 3} Ke7
{(54. ... Kf8e7 55.Bd3f1 Rb8a8 56.Ra6xa8 Rc8xa8 57.Nc4b6 Ra8a2+ 58.Kf2g3 Nb5d4
59.Nb6xd7 Ke7xd7 60.Kg3f4 g6g5+ 61.Kf4xg5 Nd4xf3+ 62.Kg5f6) +0.68/13 9} 55.f4
{(1109424299) +1.14/7 6} Rba8 {(55. ... Rb8a8 56.Ra6b6 Rc8b8 57.f4f5 Ra8a2+ 58.Kf2g3 Rb8xb6
59.Nc4xb6 Ra2a3 60.f5f6+ Ke7e8 61.Bd3xb5 Bd7xb5 62.Kg3f4 Ra3a2) +1.03/13 19} 56.Rxa8
{(1109424325) +1.16/7 5} Rxa8 {(56. ... Rc8xa8 57.Nc4b6 Ra8a2+ 58.Kf2e1 Bd7e8 59.g4g5 Nb5d4
60.Ne3c4 Nd4f3+ 61.Ke1d1 Be8d7 62.Nb6xd7 Ke7xd7) +1.06/11 1} 57.e5 {(1109424329) +1.02/7
2} dxe5 {(57. ... d6xe5 58.f4xe5 Nb5d4 59.Kf2g3 Ra8a2 60.Bd3b1 Ra2a1 61.Bb1d3
Nd4b3 62.Kg3f4 Nb3c5 63.Bd3c2) +1.17/12 6} 58.fxe5 {(1109424340) +1.04/7 3} Nd4 {(58.
... Nb5d4 59.Kf2g3 Ra8a2 60.Bd3b1 Ra2a1 61.Bb1d3 Nd4b3 62.Bd3c2 Nb3c5 63.Kg3f4
Ra1c1 64.g4g5) +1.14/12 8} 59.Kg3 {(1109424352) +1.40/7 3} Ra2 {(59. ... Ra8a2
60.Bd3b1 Ra2a1 61.Bb1d3 Nd4b3 62.Nc4b6 Ra1e1 63.Kg3f4 Nb3c1 64.Bd3c4 Nc1e2+ 65.Kf4f3
Ne2d4+ 66.Kf3e4) +1.04/12 8} 60.Kf4 {(1109424364) +1.48/7 2} Rf2+ {(60. ... Ra2f2+
61.Kf4e4 Nd4b3 62.Bd3c2 Nb3c1 63.d5d6+ Ke7d8 64.Ke4d5 Bd7e6+ 65.Kd5c5 Be6xg4 66.Bc2xg6
Nc1b3+ 67.Kc5b6 Bg4d7) +1.12/11 5} 61.Ke4 {(1109424371) +1.00/7} Nb3 {(61. ... Nd4b3
62.Bd3c2 Nb3c1 63.Nc4a3 Rf2h2 64.Ke4f4 Nc1e2+ 65.Kf4e4 Rh2h4 66.Ke4f3 Ne2d4+ 67.Kf3f4
Nd4xc2 68.Na3xc2) +1.21/12 7} 62.Bc2 {(1109424381) +1.63/7 1} Nc1 {(62. ... Nb3c1
63.d5d6+ Ke7d8 64.Nc4b6 Bd7xg4 65.Ne3xg4 Rf2xc2 66.Ng4h6 Rc2f2 67.Ke4e3 Rf2f1 68.Ke3e4)
+1.20/11 3} 63.Bd1 {(1109424389) +1.46/7 2} Na2 {(63. ... Nc1a2 64.Bd1b3 Na2c3+ 65.Ke4d4
Nc3e2+ 66.Kd4c5 Ne2c1 67.d5d6+ Ke7e8 68.Bb3c2 Bd7xg4 69.Bc2xg6 Nc1b3+ 70.Kc5b6 Bg4d7)
+1.10/11 7} 64.Bb3 {(1109424399) +1.55/7 1} Nc3+ {(64. ... Na2c3+ 65.Ke4d4 Nc3e2+
66.Kd4c5 Ne2f4 67.Kc5d4 Nf4g2 68.d5d6+ Ke7d8 69.Ne3xg2 Rf2xg2 70.Nc4e3 Rg2d2+ 71.Kd4e4
Kd8e8 72.Bb3d5) +1.08/12 6} 65.Kd3 {(1109424410) +1.61/8 3} Ne2 {(65. ... Nc3e2
66.Bb3d1 Ne2f4+ 67.Kd3d4 f7f6 68.e5e6 Bd7b5 69.Bd1b3 Nf4e2+ 70.Kd4c5 Ne2c1 71.Kc5xb5
Nc1xb3 72.Kb5c6 Nb3d4+ 73.Kc6c7) +1.27/12 7} 66.d6+ {(1109424421) +1.51/7 2} Kd8 {(66.
... Ke7d8 67.Bb3c2 Ne2c1+ 68.Kd3e4 f7f5+ 69.e5xf6/ep Rf2xf6 70.Nc4e5 Nc1e2 71.Bc2d3
Ne2f4 72.Ne5xd7 Kd8xd7) +1.23/11 3} 67.Bc2 {(1109424428) +1.16/7 2} f6 {(67. ... f7f6
68.e5xf6 Ne2c1+ 69.Kd3e4 Rf2xf6 70.Nc4e5 Nc1e2 71.Ke4d5 Ne2c3+ 72.Kd5c5 Rf6e6 73.Ne3c4
g6g5 74.Ne5xd7 Kd8xd7) +1.28/11 5} 68.exf6 {(1109424435) +1.92/7 1} Nc1+ {(68. ...
Ne2c1+ 69.Kd3e4 Rf2xf6 70.Nc4e5 Nc1e2 71.Bc2d3 Ne2c3+ 72.Ke4d4 Nc3b5+ 73.Kd4d5 g6g5
74.Ne3f5 Bd7e6+ 75.Kd5c5 Nb5a7 76.Bd3e4) +1.09/12 5} 69.Kd4 {(1109424443) +1.69/8 1}
Rxf6 {(69. ... Rf2xf6 70.g4g5 Nc1e2+ 71.Kd4c5 Rf6f2 72.Bc2xg6 Ne2f4 73.Bg6f5 Bd7xf5
74.Ne3xf5 Nf4d3+ 75.Kc5d4 Rf2xf5 76.Kd4xd3 Rf5d5+ 77.Kd3e4 Rd5xg5) +0.75/14 7} 70.g5
{(1109424454) +1.49/7 1} Ne2+ {(70. ... Nc1e2+ 71.Kd4c5 Rf6f2 72.Bc2xg6 Ne2f4 73.Bg6f5 Bd7xf5
74.Ne3xf5 Nf4d3+ 75.Kc5d4 Rf2xf5 76.Kd4xd3 Rf5f1 77.Nc4e5 Rf1d1+ 78.Kd3e4 Rd1xd6)
+0.57/13 4} 71.Kc5 {(1109424461) +1.16/8 2} Rf2 {(71. ... Rf6f2 72.Bc2xg6 Ne2f4 73.Bg6f5
Bd7xf5 74.Ne3xf5 Nf4d3+ 75.Kc5d4 Rf2xf5 76.Kd4xd3 Rf5xg5 77.Kd3d4 Rg5g6 78.Kd4d5
Kd8d7) +0.82/14 5} 72.Bxg6 {(1109424469) +1.81/7} Nf4 {(72. ... Ne2f4 73.Bg6f5 Bd7xf5
74.Ne3xf5 Nf4d3+ 75.Kc5d4 Rf2xf5 76.Kd4xd3 Kd8d7 77.g5g6 Rf5g5 78.Kd3e4 Kd7e6 79.Ke4d4
Rg5g1 80.Kd4c5 Rg1xg6) +0.75/15 7} 73.Bf7 {(1109424479) +1.68/7 1} Nd3+ {(73. ...
Nf4d3+ 74.Kc5d4 Rf2xf7 75.Kd4xd3 Rf7f3 76.Nc4e5 Rf3g3 77.Kd3e4 Rg3g1 78.g5g6 Bd7e8
79.Ke4d5 Be8d7 80.Ne5xd7 Kd8xd7) +0.69/13 4} 74.Kd4 {(1109424485) +1.61/8} Rxf7 {(74.
... Rf2xf7 75.Kd4xd3 Rf7g7 76.Nc4e5 Bd7e8 77.Kd3e4 Rg7xg5 78.Ne3g4 Rg5g7 79.Ng4f6
Be8a4 80.Ke4d5 Ba4b3+ 81.Kd5c6 Rg7g1) +0.63/14 6} 75.Kxd3 {(1109424494) +1.49/8 1}
Rg7 {(75. ... Rf7g7 76.Nc4e5 Bd7e8 77.Ne3g4 Rg7xg5 78.Kd3e4 Rg5g7 79.Ng4f6 Be8a4
80.Ke4d5 Rg7g2 81.Ne5c6+ Ba4xc6+ 82.Kd5xc6 Rg2c2+ 83.Kc6b7 Rc2d2) +0.64/14 6} 76.Ne5
{(1109424502) +1.99/7} Be6 {(76. ... Bd7e6 77.g5g6 Rg7b7 78.Kd3d4 Rb7b4+ 79.Kd4c5 Rb4b1
80.Ne3g4 Rb1c1+ 81.Kc5b6 Rc1f1 82.Kb6c5 Kd8e8 83.Kc5c6 Rf1c1+ 84.Kc6b7) +0.36/13 12}
77.g6 {(1109424516) +2.62/7} Rb7 {(77. ... Rg7b7 78.Kd3d4 Rb7b1 79.Kd4e4 Rb1b4+
80.Ne3c4 Kd8e8 81.Ke4d4 Rb4b1 82.Kd4c5 Rb1g1 83.Nc4b6 Rg1g5 84.Kc5d4) +0.29/13 5} 78.Kd4
{(1109424524) +2.36/7 1} Rb1 {(78. ... Rb7b1 79.g6g7 Rb1g1 80.Ne3g4 Kd8c8 81.d6d7+ Kc8d8
82.Ng4f6 Kd8c7 83.Ne5c6 Kc7xc6 84.d7d8N+ Kc6d6 85.Nd8xe6 Kd6xe6) 0.00/13 11} 79.g7
{(1109424537) +2.38/6} Rg1 {(79. ... Rb1g1 80.Ne3g4 Rg1f1 81.Kd4e4 Rf1g1 82.Ke4d4) 0.00/14 7}
80.N3g4 {(1109424546) +2.13/6} Rd1+ {(80. ... Rg1d1+ 81.Kd4c5 Rd1d5+ 82.Kc5b6 Rd5xd6+
83.Ne5c6+ Kd8c8 84.Ng4f6 Rd6d2 85.g7g8Q+ Be6xg8 86.Nf6xg8 Rd2b2+ 87.Kb6c5 Rb2c2+ 88.Kc5b5
Rc2xc6 89.Kb5xc6) 0.00/14 7} 81.Nd3 {(1109424594) +2.37/9 39} Rg1 {(81. ... Rd1g1
82.Nd3e5) 0.00/16 10} 82.Nde5 {(1109424625) +2.19/9 18} Rf1 {(82. ... Rg1f1 83.Kd4e4
Kd8c8 84.Ng4h6 Rf1g1 85.Ne5g4 Kc8d7 86.g7g8Q Be6xg8 87.Ng4f6+ Kd7xd6 88.Nh6xg8 Rg1xg8
89.Nf6xg8) 0.00/14 12} 83.Ke3 {(1109424648) +2.35/8 10} Ba2 {(83. ... Be6a2 84.Ke3e2 Rf1f5
85.Ne5c6+ Kd8d7 86.Nc6e7 Rf5g5 87.Ng4f6+ Kd7xd6 88.Nf6e4+ Kd6xe7 89.Ne4xg5 Ke7f6 90.g7g8R
Ba2xg8 91.Ke2e3) 0.00/13 14} 84.Nc6+ {(1109424673) +5.38/8 9} Kd7 {(84. ... Kd8d7
85.Nc6e7 Kd7xd6 86.g7g8Q Ba2xg8 87.Ne7xg8 Kd6e6 88.Ng4f6 Rf1xf6 89.Ng8xf6) 0.00/16 7}
85.Ne7 {(1109424701) +5.38/9 18} Kxd6 {(85. ... Kd7xd6 86.g7g8Q Ba2xg8 87.Ne7xg8 Kd6e6
88.Ng4f6 Rf1xf6 89.Ng8xf6) 0.00/16 6} 86.g8=Q {(1109424729) +6.02/10 20} Bxg8 {(86. ...
Ba2xg8 87.Ne7xg8 Kd6e6 88.Ng4f6 Rf1xf6 89.Ng8xf6) 0.00/20 10} 87.Nxg8 {(1109424747)
+5.92/9 6} Ra1 {(87. ... Rf1a1 88.Ng4f6 Kd6c5 89.Ke3e4 Kc5d6 90.Ke4d4 Kd6e6 91.Kd4e3
Ke6f7 92.Ke3d4 Ra1e1 93.Kd4c5 Re1b1 94.Kc5d4 Rb1e1) 0.00/19 6} 88.N8f6 {(1109424763)
+6.40/9 7} Rf1 {(88. ... Ra1f1 89.Ke3e4 Rf1xf6 90.Ng4xf6) 0.00/18 6} 89.Ne4+
{(1109424788) +6.75/9 17} Kd5 {(89. ... Kd6d5 90.Ke3d3 Rf1g1 91.Ng4e3+ Kd5e5 92.Ne4c5 Rg1c1
93.Ne3g4+ Ke5f5 94.Ng4e3+ Kf5e5) 0.00/18 8} 90.Kd3 {(1109424814) +6.80/9 16} Rh1 {(90.
... Rf1h1 91.Ng4e3+ Kd5e5 92.Ne4g5 Ke5f4 93.Ng5e6+ Kf4e5 94.Ne6f8 Rh1g1 95.Kd3e2
Ke5f4 96.Ke2d3 Kf4e5) 0.00/17 6} 91.Ne3+ {(1109424839) +7.36/9 16} Ke6 {(91. ...
Kd5e6 92.Kd3e2 Ke6e5 93.Ne4c3 Ke5f4 94.Ke2d3 Kf4e5 95.Ne3c4+ Ke5f5 96.Nc3d5 Rh1b1
97.Nd5e3+ Kf5e6 98.Kd3d4 Rb1e1 99.Kd4c5 Re1xe3 100.Nc4xe3) 0.00/18 8} 92.Kd4
{(1109424853) +7.61/8 4} Rh4 {(92. ... Rh1h4 93.Ne3c4 Rh4xe4+ 94.Kd4xe4) 0.00/22 6} 93.Nd5
{(1109424866) +7.64/9 6} Rxe4+ {(93. ... Rh4xe4+ 94.Kd4xe4) 0.00/24 6} 94.Kxe4 {(1109424879)
+10.40/12 5} {Insufficient material} 1/2-1/2

Jim.

[Guenther Simon]

Hi Jim,

Have you noticed the strange depths of Glaurung?
Glaurung was almost always 4-6 plies behind MSCP.
I think something was clearly wrong?

Regards,
Guenther

[Jim Ablett]

Hi Guenther,

Oh no!
Thanks for bursting my bubble!
I thought it was too good to be true.

I'll re-run with a different build - That was a
Bryan Hofmann one, I try Dann Corbit's
instead.

Regards,
Jim.

[Guenther Simon]

My first thought was you were testing Glaurungs
new rating adjustment ability and had it set to 1700

Regards,
Guenther

[Jim Ablett]

I think I know what it is.

I implemented in Mscp a liitle routine I call Aggressive Time Management
(a.t.m). It makes mscp think very deeply in the early to middle stages of
the game (where it is most critical in my view) and play very quickly the
last 10 moves or so up to the time check, where as there is usually less
material on the board to calculate for, it can think less deeply and usually
get away with it. I said usually - it does come a cropper sometimes, so I
put in an option to switch it off in the .ini file. Once over the time-check
threshold (in this case 40th move), it then repeats the process again,
thinking more deeply till up to the maybe the 70th move when it lowers
it's depth again, and so on and so on. It's still experimental, but it seems
to have given mscp a massive boost, at least in fast time control games.
I think I'm giving away my trade secrets here!


Jim.

[Marcus Prewarski]

Increasing the time will not make that much a difference in search depths. If you really want to improve an engine like MSCP you should focus on the search and branch factor. Small changes in evaluation and time allocation will have very minor changes in performance unless you uconver serious bugs in evaluation.

[Jim Ablett]

Hi Marcus,
Well I haven't done that much else strength-wise
to mscp, just tinkered with the piece square tables a bit.

Here's a game I just played against a non-uci engine - Horizon,
on paper a lot stronger than Mscp, again I think Horizon was
lucky to escape with a draw and was taken apart material-wise.

[Event "Arena tournament"]
[Site "TYRELL-XBHMTDAN"]
[Date "2005.02.26"]
[Round "1"]
[White "Horizon 4.1 b13"]
[Black "Mscp 1.6"]
[Result "1/2-1/2"]
[ECO "A04"]
[Opening "Reti Opening"]
[Time "16:32:16"]
[Variation "1...c5 2.g3 g6 3.Bg2 Bg7 4.O-O Nc6 5.d3 d6 6.e4"]
[TimeControl "40/300:40/300:40/300"]
[Termination "normal"]
[PlyCount "79"]
[WhiteType "program"]
[BlackType "program"]

1.e4 c5 2.Nf3 Nc6 3.d3 g6 4.g3 Bg7 5.Bg2 d6 6.O-O {(1109435562) -3.90/7 10} e5
7.c4 {(1109435596) -3.98/8 32} Nge7 8.Qe2 {(1109435649) +0.29/8 52} O-O {(8. ... 00
9.Bc1e3 f7f5 10.Nb1c3 f5xe4 11.Nc3xe4 Nc6d4 12.Nf3xd4) +0.20/11 13} 9.Nc3 {(1109435704)
+0.26/8 40} Bg4 {(9. ... Bc8g4 10.Bc1e3 Nc6d4 11.Be3xd4 c5xd4 12.Nc3b5 f7f5 13.Nb5a3
Rf8f6 14.h2h3 Bg4xf3) +0.72/11 8} 10.Be3 {(1109435731) +0.31/7 17} Nd4 {(10. ...
Nc6d4 11.Be3xd4 c5xd4 12.Nc3d5 f7f5 13.h2h3 f5xe4 14.d3xe4 Rf8xf3 15.Bg2xf3 Bg4xh3)
+0.69/11 9} 11.Bxd4 {(1109435759) +0.46/8 16} cxd4 {(11. ... c5xd4 12.Nc3b5 f7f5
13.e4xf5 a7a6 14.Nb5a3 Ne7xf5 15.Qe2e4 Bg4xf3 16.Qe4xf3 Nf5e3) +0.86/11 8} 12.Nb5
{(1109435773) +0.62/7 4} Qb6 {(12. ... Qd8b6 13.b2b3 Bg4d7 14.Nb5a3 f7f5 15.Na3c2 Bd7c6
16.b3b4 f5xe4 17.d3xe4) +0.94/10 8} 13.Rab1 {(1109435791) +0.60/7 8} f5 {(13. ... f7f5
14.e4xf5 Rf8xf5 15.h2h3 Bg4xf3 16.Bg2xf3) +1.11/10 8} 14.Rfd1 {(1109435815) +0.51/7 14}
fxe4 {(14. ... f5xe4 15.d3xe4 Rf8f6 16.Rd1d3 Qb6c5 17.Nb5a3 Ne7c6 18.Rb1d1 Ra8f8)
+1.32/9 7} 15.dxe4 {(1109435831) +0.44/7 6} d5 {(15. ... d6d5 16.e4xd5 Qb6f6 17.Rd1d3
Bg4f5 18.Rb1e1 Bf5xd3 19.Qe2xd3 a7a6) +1.61/9 8} 16.exd5 {(1109435854) +1.35/7 13}
Qf6 {(16. ... Qb6f6 17.d5d6 Ne7c6 18.Nb5c7 Ra8d8 19.Nc7d5 Qf6f7 20.Rd1d3 Rd8xd6)
+0.94/9 8} 17.d6 {(1109435871) +1.31/7 8} Nc6 {(17. ... Ne7c6 18.Nb5c7 Ra8c8 19.Nc7d5
Qf6xd6 20.Rb1a1 b7b5 21.b2b3 Bg4f5) +1.22/10 6} 18.Rd3 {(1109435881) +0.13/6 2} Bf5
{(18. ... Bg4f5 19.Nf3e1 a7a6 20.Nb5c7 Ra8c8 21.Nc7d5 Qf6xd6 22.Bg2e4 Bf5h3) +1.50/9
8} 19.Ne1 {(1109435899) +0.69/7 8} Bxd3 {(19. ... Bf5xd3 20.Bg2d5+ Kg8h8 21.Ne1xd3
Ra8c8 22.d6d7 Rc8d8 23.Nd3c5 Rd8b8) +1.21/9 8} 20.Bd5+ {(1109435910) +0.75/6 1} Kh8
{(20. ... Kg8h8 21.Ne1xd3 Ra8c8 22.c4c5 Qf6f5 23.Bd5e4 Qf5d7 24.Be4xc6 Qd7xc6)
+1.14/10 7} 21.Nxd3 {(1109435920) +0.77/6 1} Rac8 {(21. ... Ra8c8 22.c4c5 Qf6f5 23.Qe2e4
Qf5h3 24.f2f4 e5xf4 25.Nd3xf4) +0.88/9 8} 22.c5 {(1109435932) +0.84/6 2} Qd8 {(22.
... Qf6d8 23.Bd5xc6 Rc8xc6 24.b2b4 Rf8e8 25.Nb5xa7 Rc6a6 26.Na7b5) +0.61/9 8} 23.b4
{(1109435945) +1.49/6 3} a6 {(23. ... a7a6 24.Bd5xc6 a6xb5 25.Bc6xb5 Rc8a8) 0.00/9 8} 24.Bxc6
{(1109435956) +2.04/6 1} Rxc6 {(24. ... Rc8xc6 25.Nb5a7 Rc6xd6 26.c5xd6) -1.62/10 7} 25.Na7
{(1109435969) +2.73/7 3} Re8 {(25. ... Rf8e8 26.Na7xc6 b7xc6 27.Qe2e4 Qd8d7 28.Rb1e1 g6g5
29.f2f4 g5xf4 30.Nd3xf4) -2.29/10 7} 26.Nxc6 {(1109435979) +2.81/6 1} bxc6 {(26. ...
b7xc6 27.Rb1e1 Qd8d7 28.Nd3c1 Qd7c8 29.Qe2f3 Re8d8 30.Qf3f7 Qc8g4 31.Nc1d3) 0.00/11
2} 27.Qc2 {(1109435985) +2.44/6} Qb8 {(27. ... Qd8b8 28.Qc2c4 Qb8b5 29.Qc4xb5
a6xb5 30.f2f3 Bg7f6) -1.86/10 8} 28.a4 {(1109435996) +2.78/6 1} e4 {(28. ... e5e4
29.Nd3f4 Bg7f6 30.Rb1e1 Qb8xb4) -2.04/10 8} 29.Nf4 {(1109436009) +2.99/7 2} Qc8 {(29.
... Qb8c8 30.Qc2c4 d4d3 31.Rb1d1 Bg7h6 32.Qc4f7 Re8f8 33.Qf7e6) -2.09/10 9} 30.Qc4
{(1109436020) +2.83/6} g5 {(30. ... g6g5 31.Nf4h5 Bg7e5 32.Rb1e1 Qc8f5 33.Qc4xa6 Qf5h3
34.d6d7 Qh3xd7 35.f2f3) -2.48/10 9} 31.Ne2 {(1109436033) +3.08/6 1} Qg4 {(31. ... Qc8g4
32.Rb1e1 d4d3 33.Ne2c3 d3d2 34.Re1e2 Qg4e6 35.Qc4xe6) -1.92/9 6} 32.Nxd4 {(1109436044)
+4.12/6 3} e3 {(32. ... e4e3 33.f2xe3 Re8xe3 34.d6d7 Qg4xd7 35.Nd4c2 Qd7d3 36.Nc2xe3)
-2.12/10 8} 33.f4 {(1109436055) +4.18/6} gxf4 {(33. ... g5xf4 34.Qc4f7 Re8f8 35.Qf7e6
Qg4h5 36.Qe6c4 f4xg3 37.h2xg3) -1.04/9 8} 34.Nxc6 {(1109436067) +4.31/6 1} Qf3 {(34.
... Qg4f3 35.d6d7 Re8d8 36.Nc6xd8 Qf3f2+ 37.Kg1h1 Qf2f3+ 38.Kh1g1) 0.00/9 9} 35.d7
{(1109436080) 0.00/6 1} Rd8 {(35. ... Re8d8 36.Nc6xd8 Qf3f2+ 37.Kg1h1 Qf2f3+ 38.Kh1g1) 0.00/9
9} 36.Nxd8 {(1109436092) 0.00/6 1} Qf2+ {(36. ... Qf3f2+ 37.Kg1h1 Qf2f3+ 38.Kh1g1)
0.00/11 10} 37.Kh1 Qf3+ {(37. ... Qf2f3+ 38.Kh1g1 Qf3f2+ 39.Kg1h1) 0.00/11 7} 38.Kg1
Qf2+ {(38. ... Qf3f2+ 39.Kg1h1 Qf2f3+ 40.Kh1g1) 0.00/12 11} 39.Kh1 Qf3+ {(39. ...
Qf2f3+ 40.Kh1g1 Qf3f2+ 41.Kg1h1) 0.00/12 11} 40.Kg1 {3-fold repetition} 1/2-1/2

Jim.

[Guenther Simon]

Hi Jim,

Sorry to destroy your bubbles again.
In the game versus Horizon there is the same error
happening as in the first game vs. Glaurung posted.

Is it possible that MSCP had always pondered and thus
Glaurung and Horizon were severely handicapped?
Anything something in that games is very wrong.

depth 8 only after 32-40 seconds for Horizon??

(BTW where does the strange eval score come there too?)

6.O-O {(1109435562) -3.90/7 10} e5
7.c4 {(1109435596) -3.98/8 32} Nge7 8.Qe2 {(1109435649) +0.29/8 52} O-O {(8. ... 00
9.Bc1e3 f7f5 10.Nb1c3 f5xe4 11.Nc3xe4 Nc6d4 12.Nf3xd4) +0.20/11 13} 9.Nc3 {(1109435704)
+0.26/8 40}

depth 8 only for Glaurung in 26-30 seconds??

12.a3 {(1109423370) +0.93/8 26}
bxa3 {(12. ... b4xa3 13.Ra1xa3 Nb8c6 14.Ra3b3 Nc6b4 15.Qd3e2 Qd8e7 16.Rf1d1 Ra8e8
17.Rb3e3 e5xd4 18.Nf3xd4 Qe7e5) -0.17/13 12} 13.Rxa3 {(1109423415) +0.90/8 30}

This are only a few snippets, but that pattern goes through
the whole games (sometimes it also looks like both opponents
moved much too fast, without need?)

Regards,
Guenther

[Uri Blass]

As far as I see Mscp simply outsearched horizon

horizon (white) got often depth 6 or 7

for example it for example I see 26.Nxc6 2.81/6 when mscp replied bxc6 with depth 11.

It seems strange to me and I suspect that something is wrong here.

[Jim Ablett]

Yes, it seems that maybe mscp is a system hog,
taking all the resources/threads of the computer
for itself and leaving nothing for the oppponent.

More testing needed.

Jim.

[Marcus Prewarski]

Now that I've looked at it a bit more and tried your MSCP on my system it looks to me that unlike what the PGN header says MSCP is actually white in both cases. I think that explains the search depths and the strange eval numbers early. On the plus side, MSCP still got the draws. It probably benefited from somewhat closed games.

[Uri Blass]

looking at the game against horizon

If mscp was white
It seems that horizon blundered in the following position by 15...d5

r4rk1/pp2n1bp/1q1p2p1/1N2p3/2PpP1b1/5NP1/PP2QPBP/1R1R2K1 b - - 0 15

I fail to understand why to give the opponent a protected passed pawn when black is better without it.

[Uri Blass]

looking at the game against glaurung I really cannot understand the evaluation of mscp if mscp is white.

It says more than +6 for KNN vvs KR and later evaluates +10.40 for
KN vs K.

Uri

[Guenther Simon]

Now that I've looked at it a bit more and tried your MSCP on my system it looks to me that unlike what the PGN header says MSCP is actually white in both cases. I think that explains the search depths and the strange eval numbers early. On the plus side, MSCP still got the draws. It probably benefited from somewhat closed games.

looking at the game against horizon

If mscp was white
It seems that horizon blundered in the following position by 15...d5

r4rk1/pp2n1bp/1q1p2p1/1N2p3/2PpP1b1/5NP1/PP2QPBP/1R1R2K1 b - - 0 15

I fail to understand why to give the opponent a protected passed pawn when black is better without it.
Marcus Prewarski
PostPosted: Sat Feb 26, 2005 9:04 pm

Thanks to Marcus and the *hint* by Uri
and thanks to my curiosity about riddles I suppose
I have solved the whole thing!
It really seems MSCP was actually White in both games,
because I really could verify Horizons move as _Black_
in the second game with the exactly same depth and
score! A move Uri did not believe Horizon would ever
play, but my debug shows it was Horizon without any doubt.

Thus, it looks like MSCP had really got 2 draws against
very strong opponents in 2 blitz games.

Regards,
Guenther
P.S. The debug is from my slow PIII 650 therefore Horizon
needed some time to reach depth 9 and that was also the reason
for my test set up with 40/15.

Code: Select all

recognized 'normal' (-1) as variant normal
WinBoard 4.2.7 + Horizon_41b13
Reset(1, 0) from gameMode 0
recognized 'normal' (-1) as variant normal
GameEnds(0, (null), 2)
StartChildProcess (dir="c:\Winboard\Horizon_41b13") Horizon_41b13
671 >first : xboard
protover 2
711 <first : Welcome to Horizon Version 4.1b13 !
711 <first :
1412 <first : Loading Opening Book C:\Winboard\\Horizon_41b13\Horizon.dat ... OK
1412 <first : [96201 Positions]
1432 <first :
1432 <first :
1432 <first : Checking for Tablebase Files
2283 <first : Including 5-man Endgame Table Bases
2283 <first :
2283 <first : r n b q k b n r
2283 <first : p p p p p p p p
2283 <first : . . . . . . . .
2283 <first : . . . . . . . .
2283 <first : . . . . . . . .
2283 <first : . . . . . . . .
2283 <first : P P P P P P P P
2283 <first : R N B Q K B N R
2283 <first :
2283 <first : [1]W >
10675 >first : new
random
10675 >first : level 40 15 0
10675 >first : post
10675 >first : hard
10675 >first : easy
PasteTextFromClipboard(): lock count 0
PasteTextFromClipboard(): lock count 0
PasteTextFromClipboard(): err 0 locked 0
PasteFenFromClipboard(): fen 'r4rk1/pp2n1bp/1q1p2p1/1N2p3/2PpP1b1/5NP1/PP2QPBP/1R1R2K1 b - - 0 15'
18967 >first : force
18967 >first : new
random
18967 >first : level 40 15 0
18967 >first : post
18967 >first : hard
18967 >first : easy
18967 >first : force
18967 >first : a2a3
18967 >first : edit
18967 >first : #
18967 >first : Nb5
18967 >first : Pc4
18967 >first : Pe4
18967 >first : Nf3
18967 >first : Pg3
18967 >first : Pa2
18967 >first : Pb2
18967 >first : Qe2
18967 >first : Pf2
18967 >first : Bg2
18967 >first : Ph2
18967 >first : Rb1
18967 >first : Rd1
18967 >first : Kg1
18967 >first : c
18967 >first : Ra8
18967 >first : Rf8
18967 >first : Kg8
18967 >first : Pa7
18967 >first : Pb7
18967 >first : Ne7
18967 >first : Bg7
18967 >first : Ph7
18967 >first : Qb6
18967 >first : Pd6
18967 >first : Pg6
18967 >first : Pe5
18967 >first : Pd4
18967 >first : Bg4
18967 >first : .
21901 >first : white
21901 >first : time 90000
otim 90000
21901 >first : black
go
21931 <first : Current Position = 0.97
21931 <first :   2     71       0         38   Bxf3 Bxf3
21941 <first :   2     76       1        204   Rac8 b4
21951 <first :   2     84       2        386   Bd7 Qd3
21951 <first :   2     86       2        434   Be6 b4
21971 <first :   3    102       3        810   Be6 b4 Qc6
21991 <first :   3    109       6       2195   Rf6 h3 Bd7
22001 <first :   3    123!      7       2819   a6 Na3 Be6
22011 <first :   3    132       8       2913   a6 Na3 Qc5
22022 <first :   4     97?      9       3702   a6 Na3 a5
22202 <first :   4     97      27      13146   a6 Na3 a5
22242 <first :   5    110      31      16101   a6 Na3 Bxf3 Bxf3 Qc5
22332 <first :   5    114      40      22605   Rf6 h3 Bxf3 Bxf3 Raf8
22723 <first :   6    119      79      63871   Rf6 h3 Bd7 b4 a5 bxa5 Qxa5
23914 <first :   7    142     198     167953   Rf6 b4 Raf8 Rd3 Be6 Ra3 Qc6
27479 <first :   8    127     555     555321   Rf6 Na3 Raf8 Rd3 Qb4 Rbd1 Qc5 h4
34760 <first :   9    144    1283    1317308   Rf6 h3 Bxf3 Bxf3 Raf8 Rd3 a6 Na3 Qc5
41730 <first :   9    161    1980    2053657   d5 exd5 Qf6 Rd3 Bf5 Re1 Bxd3 Qxd3 a6
46537 <first :   9    161    2460    2614980   d5 exd5 Qf6 Rd3 Bf5 Re1 Bxd3 Qxd3 a6
46537 <first : move d6d5

[Jim Ablett]

You should really take Mscp's evaluation scores with a pinch of salt. I've been messing around with it, changing values here and it needs a lot of sorting out. Thanks for everyones input. So back to an earlier point, I'm still not clear, is mscp still searching deeper than both glaurung and horizon or not?

[Marcus Prewarski]

No, Horizon and Glaurung are searching several plys deeper than MSCP. This is the main reason why they are much stronger.

You can tweak the evaluation all you want but if the engine is getting outsearched by that much, it will have trouble winning games. Adding more time won't change this much either. Going back to my original reply, the best way to improve MSCP is to work on the search and move ordering to make it search deeper.

[Jim Ablett]

Hello again Marcus,

I agree with you that all things being equal
the deepest searcher will always be stronger,
although top human players I believe, don't win
their majority of games through searching
deeper, they can never hope to compete against a
computer with regards to search depth. So how do
the repeatedly outplay computers, through better
evaluation and tactics, true? . This is the area
I'm more interested in.

Here's a game I just played with a new version of
Mscp (1.6.1) against your very strong 'Drunkenmaster'
engine which should search a lot deeper than mscp.
I know short time controls are a great 'leveller'
in these matches, but nonetheless this is a great win
for mscp in 33 moves so I must be doing something right.

[Event "Arena tournament"]
[Site "TYRELL-XBHMTDAN"]
[Date "2005.02.27"]
[Round "1"]
[White ["Mscp 1.6"]
[Black "Drunkenmaster 1.2"]
[Result "1-0"]
[ECO "B43"]
[Opening "Sicilian"]
[Time "16:16:38"]
[Variation "Kan, 5.Nc3 Qc7 6.Bd3 Nf6 7.O-O"]
[TimeControl "40/300:40/300:40/300"]
[Termination "normal"]
[PlyCount "64"]
[WhiteType "program"]
[BlackType "program"]

1.Nc3 c5 {(d7d5<66> c7c5<33> ) 0.00/1} 2.e4 e6 {(b8c6<60> a7a6<10> e7e6<17>
d7d6<10> ) 0.00/1} 3.Nf3 a6 {(a7a6<40> b8c6<40> d7d6<20> ) 0.00/1} 4.d4 cxd4 {(c5d4<100>
) 0.00/1} 5.Nxd4 Qc7 {(b7b5<45> d8c7<41> d7d6<12> ) 0.00/1} 6.Bd3 Nf6 {(g8f6<100>
) 0.00/1} 7.O-O Bd6 {(7. ... Bf8d6 8.Nd4f3 Nf6g4 9.h2h3 Ng4h2 10.Rf1e1 Nh2xf3+
11.Qd1xf3 Bd6h2+) -0.64/9 7} 8.f4 {(1109521039) +0.43/7 15} Qb6 {(8. ... Qc7b6 9.Nc3e2
e6e5 10.f4xe5 Bd6xe5 11.c2c3 00 12.Qd1b3 Qb6c5) -0.49/9 7} 9.Be3 {(1109521055)
+0.57/7 7} Qxb2 {(9. ... Qb6xb2 10.Nc3a4 Qb2b4 11.c2c4 Bd6c7 12.e4e5 Nf6g8 13.Ra1b1
Qb4a3) -0.43/9 7} 10.Na4 {(1109521073) +0.50/7 9} Qb4 {(10. ... Qb2b4 11.c2c4 Bd6c7
12.e4e5 Nf6g8 13.Ra1b1 Qb4a3 14.Qd1b3 Qa3e7) -0.64/9 7} 11.c4 {(1109521098) +0.72/7 16}
Bc7 {(11. ... Bd6c7 12.e4e5 Nf6g8 13.Ra1b1 Qb4e7 14.Na4b6 Bc7xb6 15.Rb1xb6 Nb8c6)
-1.11/9 15} 12.e5 {(1109521129) +1.04/7 12} Ng8 {(12. ... Nf6g8 13.Ra1b1 Qb4e7 14.Na4b6
Bc7xb6 15.Rb1xb6 Nb8c6 16.Qd1g4 f7f5) -1.05/9 7} 13.c5 {(1109521162) +1.09/7 24} Qa3
{(13. ... Qb4a3 14.Na4b6 Bc7xb6 15.Nd4c2 Qa3c3 16.c5xb6 Nb8c6 17.Qd1d2 Qc3xd2)
-0.95/9 11} 14.Rf3 {(1109521187) +1.41/7 12} Nc6 {(14. ... Nb8c6 15.Nd4xe6 Qa3c3
16.c5xb6/ep Rh8xh7) -1.33/8 7} 15.Nxc6 {(1109521207) +5.98/7 9} dxc6 {(15. ... d7xc6
16.Be3d2 Bc7b6 17.Na4xb6 Qa3e7) -3.98/9 14} 16.Bd2 {(1109521233) +5.93/7 9} Ne7 {(16.
... Ng8e7 17.Bd3c2 Qa3xf3 Qf3a3) -3.98/9 6} 17.Bc4 {(1109521255) +8.68/7 13} Qxf3
{(17. ... Qa3xf3 18.Qd1xf3 Ne7d5 19.Ra1b1 00 20.Na4b6 Nd5xb6 21.c5xb6 Rf8d8 22.Bc4d3
Bc7b8) -3.96/11 6} 18.Qxf3 {(1109521284) +8.35/8 21} Nd5 {(18. ... Ne7d5 19.Na4b6
Ra8b8 20.Nb6a8) -3.97/10 7} 19.Bxd5 {(1109521301) +8.67/7 9} exd5 {(19. ... e6xd5
20.Na4b6 Bc7xb6 21.c5xb6 00 22.Qf3d3 Bc8g4 23.Bd2b4 Rf8e8 24.Ra1c1 a6a5) -3.02/11 7}
20.Nb6 {(1109521315) +8.55/7 4} Bxb6 {(20. ... Bc7xb6 21.c5xb6 00 22.Bd2b4 Rf8e8
23.Qf3d3 g7g6 24.Ra1c1 Bc8f5) -3.05/11 7} 21.cxb6 {(1109521326) +8.59/7 2} O-O {(21. ...
00 22.Bd2b4 Rf8e8 23.Qf3d3 Bc8g4 24.Bb4c5 Re8e6 25.Ra1c1 Ra8e8 26.Rc1e1 a6a5)
-3.08/11 7} 22.Qd3 {(1109521339) +8.55/7 5} f6 {(22. ... f7f6 23.Bd2b4 Rf8f7 24.Ra1e1
f6xe5 25.Re1xe5 Bc8g4 26.Qd3d4 Bg4f5 27.b6c7 Kg8xf7) -2.96/11 7} 23.Re1 {(1109521352)
+8.51/7 4} fxe5 {(23. ... f6xe5 24.Re1xe5 Bc8d7 25.Bd2c3 Rf8xf4 26.Re5e7 Rf4f7
27.Re7xf7 Kg8xf7 28.Qd3xh7 Ra8g8) -2.93/11 7} 24.Rxe5 {(1109521368) +8.61/7 6} Bd7 {(24.
... Bc8d7 25.Re5g5 h7g6) -2.93/9 7} 25.Re7 {(1109521379) +9.72/6 2} Bf5 {(25. ...
Bd7f5 26.Qd3d4 Rf8f7 27.Re7xf7 Kg8xf7 28.Qd4e5 Bf5c8 29.Qe5f5+ Kf7xb7) -3.67/9 7}
26.Qe2 {(1109521394) +9.51/6 5} Rf7 {(26. ... Rf8f7 27.Re7xf7 Kg8xf7 Kf7xf5 Kf5e4)
-4.19/9 12} 27.Rxf7 {(1109521410) +9.65/7 3} Kxf7 {(27. ... Kg8xf7 28.Qe2e5 Bf5d7
29.Qe5c7 Kf7e8 30.Qc7xb7 Ra8c8 31.Qb7xa6 Rc8b8 32.Qa6e2+ Ke8f7) -5.42/11 11} 28.Qe5
{(1109521427) +10.51/7 3} Bc8 {(28. ... Bf5c8 29.Qe5c7+ Kf7e6 30.Qc7xg7 a6a5 31.Qg7xh7 Ke6d6
32.Qh7c7+ Kd6c5 33.g2g3) -6.27/10 14} 29.Bc3 {(1109521445) +12.07/6 2} a5 {(29. ... a6a5
30.Qe5c7+ Kf7e8 31.Qc7xg7 d5d4 32.Qg7g8+ Ke8d7 33.Qg8xh7+ Kd7e8) -7.18/9 10} 30.Qxg7+
{(1109521460) +13.56/6 3} Ke6 {(30. ... Kf7e6 31.f4f5+ Ke6d6 32.Qg7f8+ Kd6d7 33.f5f6 c6c5
34.Qf8xc5 Kd7e6 35.Qc5e7+) -M0/10 9} 31.f5+ {(1109521476) +14.00/6 5} Kd6 {(31. ... Ke6d6
32.Qg7f8+ Kd6c5) -M8/9 6} 32.Bd4 {(1109521489) +16.27/6 4} Bxf5 {(32. ... Bc8xf5
33.Qg7e5+ Kd6d7 34.Bd4c5 Ra8b8 35.Qe5d6+ Kd7e8 36.Qd6e7+ Bf5d7) -M4/10 10} {Black
resigns}

Jim.

[Pallav Nawani]

Hi jim,

What sort of changes have you made to Mscp's eval? Can you post something here?

Pallav

[Marcus Prewarski]

Frist, I object to you calling DrunkenMaster a strong engine Also this is just one game and DM got its queen trapped, it happens with weak engines like mine. Occaisionally my engine gets draws or even wins against much superior engines, individual games don't mean much. I was a little worried and just ran a quick test match of DM vs your MSCP and DM won 4-0. Trust me, these games weren't won due to a great evaluation in DM. I can provide games if you are interested. Conditions were:
time control: 2/1
cpu: 1.6Ghz Centrino
MSCP using about 31Megs memory
DM using about 9: 8Meg hash size

I'm not disagreeing that evaluation is important, but until you have a good search and move ordering, it won't help much. All good engines are good tactically. Good luck with your work.

[Guenther Simon]

Hi Marcus and Jim,

First of all congrats Jim for MSCPs win vs. DrunkenMaster.
Looking at the game though, it is clear that DM went into
an opening, which it did not understand or may be I should say
which is very unfavourable for programs at all?
After 2 moves out of book 7...Bd6?! 8...Qb6?? and 9...Qxb2???
it was already completely lost. It was already strategically lost though
after 8...Qb6?? (IMHO).
The good news is that due to DMs nice book stat output I could calculate
the probability it would run into this line:
0.5 - 0.6%

Best regards,
Guenther