WAC again

[Henk]

Skipper used Simplified evaluation function in the past. That was a success for then it could beat Fairy-max almost always or most of the time. Looks like you can't do without piece square table.

[jdart]

It is always a bit of a hard choice whether to focus on search or eval.

But starting simple is generally a good approach. It is better to have a bug-free simple implementation than something more elaborate that is buggy. Complex code is harder to debug, and also tuning something that is buggy is probably a futile exercise, because as soon as the bugs are fixed it is going to behave differently.

--Jon

[Henk]

Improving WAC scores probably won't make a chess engine play stronger.

The main purpose for a test like that is to show missing or poorly implemented features in analysis like king safety ....

Skippers king safety makes it also score worse on (fifty position) WAC test. I think it's implementation is too expensive or king safety is not needed as well for that test.

[Henk]

In Skippers current implementation ranking/ordering remaining non captures/quiet moves also gives no improvement. Only bad for speed.

[Henk]

Version of Skipper scoring best on WAC test.

[pgn]
[Event "Computer Chess Game"]
[Site "HP"]
[Date "2017.08.22"]
[Round "-"]
[White "Fairy-Max 4.8S"]
[Black "SkipperWinb"]
[Result "1-0"]
[TimeControl "120"]
[Annotator "1. +0.14 1... +0.06"]

1. Nc3 {+0.14/8} Nh6 {+0.06/120 2.0} 2. f4 {+0.09/8 3} f6 {+0.08/110 2.0}
3. Nf3 {+0.13/8 2.5} Rg8 {+0.05/120 1.9} 4. e4 {+0.13/9 2.4} Rh8
{+0.01/110 1.9} 5. Be2 {+0.10/9 1.4} Kf7 {+0.06/120 1.9} 6. O-O
{+0.09/8 1.9} Nc6 {-0.04/110 1.8} 7. Ng5+ {+0.10/9 2.2} fxg5
{+0.00/110 1.8} 8. fxg5+ {+0.23/9 3} Kg8 {+0.07/120 1.8} 9. gxh6
{+0.05/9 2.4} gxh6 {+0.05/120 1.8} 10. d4 {+0.33/9 4} Bg7 {+0.09/110 1.7}
11. Bh5 {+0.53/9 2.2} Bxd4+ {-0.01/110 1.7} 12. Kh1 {+1.33/9 1.6} d5
{-4.58/130 1.7} 13. Bf7+ {+79.93/9 2.8} Kg7 {-4.48/130 1.7} 14. Bxh6+
{+79.94/10 1.2} Kxh6 {-2621.38/150 1.6} 15. Qc1+ {+79.95/12 1.5} Kg7
{-2621.34/150 1.6} 16. Qg5+ {+79.96/16 1.4} Kf8 {-2621.36/140 1.6} 17. Qh6+
{+79.97/28 0.3} Bg7 {-2621.38/150 1.5} 18. Be6+ {+79.98/28 0.2} Ke8
{-2621.40/150 1.5} 19. Qh5# {+79.99/28 0.2}
{Xboard adjudication: Checkmate} 1-0
[/pgn]

Code: Select all

WAC.007-- passed
WAC.017-- passed
WAC.004-- passed
WAC.019-- passed
WAC.018-- passed
WAC.011-- passed
WAC.006-- passed
WAC.008-- passed
WAC.016-- passed
WAC.012-- passed
WAC.001-- passed
WAC.015-- passed
WAC.005-- passed
WAC.013-- passed
WAC.009-- passed
WAC.003-- passed
WAC.010-- passed
WAC.014-- passed
WAC.020-- passed
WAC.021-- passed
WAC.023-- passed
WAC.024-- passed
WAC.025-- passed
WAC.026-- passed
WAC.027-- passed
WAC.028-- passed
WAC.029-- passed
WAC.030-- passed
WAC.032-- passed
WAC.034-- passed
WAC.035-- passed
WAC.036-- passed
WAC.037-- passed
WAC.039-- passed
WAC.040-- passed
WAC.042-- passed
WAC.044-- passed
WAC.045-- passed
WAC.046-- passed
WAC.047-- passed
WAC.048-- passed
WAC.049-- failed
WAC.050-- passed
WAC.051-- passed
WAC.053-- passed
WAC.054-- passed
WAC.055-- failed
WAC.056-- passed
WAC.057-- passed
WAC.058-- passed
Total = 50
Score = 48
Percentage = 0.96

[Henk]

Renamed it and now playing with white

[pgn]
[Event "Computer Chess Game"]
[Site "HP"]
[Date "2017.08.22"]
[Round "-"]
[White "SkipperWinb_WAC50"]
[Black "Fairy-Max 4.8S"]
[Result "0-1"]
[TimeControl "120"]
[Annotator "1. +0.08 1... +0.04"]

1. g3 {+0.08/140} d6 {+0.04/8 4} 2. h4 {-0.09/120 2.0} Bf5 {+0.10/8 4} 3.
b3 {+0.08/110 2.0} c5 {+0.16/8 5} 4. Rh2 {-0.02/110 1.9} Nc6 {+0.03/8 2.9}
5. Rh1 {-0.01/110 1.9} Qd7 {-0.02/8 4} 6. d3 {+0.06/110 1.9} O-O-O
{+0.04/8 2.6} 7. h5 {+0.09/110 1.8} Nf6 {-0.02/8 3} 8. f3 {+0.06/100 1.8}
Qe6 {-0.20/8 4} 9. Kf2 {+0.07/100 1.8} Nd5 {-0.21/7 1.4} 10. Qd2
{+0.02/100 1.7} Qe5 {-0.27/8 2.8} 11. c3 {+0.07/120 1.7} Nc7 {-0.24/8 2.3}
12. Qd1 {-0.03/100 1.7} h6 {-0.13/8 3} 13. f4 {+0.04/100 1.6} Qf6
{-0.15/8 1.0} 14. e3 {+0.05/100 1.6} Be6 {-0.17/7 1.3} 15. Rh2
{+0.09/100 1.6} Nd5 {-0.11/8 2.8} 16. d4 {-0.01/110 1.6} Bf5 {-0.24/8 4}
17. Qf3 {-0.01/100 1.5} Ndb4 {-0.11/8 2.3} 18. cxb4 {-0.11/100 1.5} Nxd4
{+1.09/10 2.2} 19. exd4 {-2.21/110 1.5} Qxd4+ {+1.09/10 1.3} 20. Be3
{-2.21/110 1.5} Qxa1 {+1.17/10 2.6} 21. Nd2 {-2.22/110 1.4} Qxa2
{+1.38/9 2.0} 22. b5 {-2.22/100 1.4} e6 {+1.51/8 1.0} 23. g4
{-2.19/110 1.4} Bc2 {+1.77/9 3} 24. Kg3 {-2.49/90 1.4} Bxb3 {+1.86/7 0.9}
25. Nxb3 {-2.83/100 1.3} Qxb3 {+1.78/9 1.8} 26. g5 {-2.77/100 1.3} hxg5
{+1.85/8 1.8} 27. fxg5 {-2.76/120 1.3} d5 {+1.75/8 1.1} 28. Kg4
{-2.76/90 1.3} f5+ {+2.19/7 1.1} 29. gxf6 {-2.86/100 1.3} gxf6
{+2.26/8 1.2} 30. Rg2 {-2.83/90 1.2} d4 {+2.70/8 1.2} 31. Bd2
{-2.78/100 1.2} f5+ {+2.82/8 0.8} 32. Kh4 {-2.88/110 1.2} d3 {+2.91/7 0.8}
33. Bg5 {-2.89/100 1.2} Rd4+ {+2.78/7 1.2} 34. Kh3 {-2.83/110 1.2} d2
{+3.38/8 1.6} 35. Qxb3 {-2.93/140 1.2} Rxh5+ {+3.84/10 1.0} 36. Kg3
{-3.03/140 1.1} Rxg5+ {+3.72/10 1.0} 37. Kf3 {-3.13/140 1.1} Rxg2
{+3.58/10 0.9} 38. Qxe6+ {-4.45/130 1.1} Kd8 {+3.57/9 1.6} 39. Kxg2
{-4.51/120 1.1} d1=Q {+3.58/9 0.7} 40. Qf6+ {-4.50/110 1.0} Be7
{+4.05/11 1.0} 41. Qxf5 {-4.51/110 1.0} Rg4+ {+3.96/11 0.9} 42. Kh1
{-4.61/110 1.0} c4 {+3.90/10 0.7} 43. Qf2 {-4.71/100 1.0} Rh4+
{+5.93/10 0.6} 44. Bh3 {-14.70/120 1.0} Rxh3+ {+7.08/11 0.6} 45. Kg2
{-17.00/140 1.0} Qg4+ {+10.97/10 0.6} 46. Kf1 {-17.01/150 1.0} Rd3
{+11.56/12 1.6} 47. Qe2 {-16.94/140 0.9} Rd1+ {+14.19/13 0.5} 48. Qxd1+
{-18.00/120 0.9} Qxd1+ {+14.95/12 1.5} 49. Kg2 {-18.02/120 0.9} c3
{+14.99/9 0.6} 50. Nh3 {-23.90/90 0.9} Qg4+ {+19.01/9 0.8} 51. Kf1
{-24.00/120 0.9} c2 {+79.96/10 0.4} 52. Nf2 {-24.10/110 0.9} c1=Q+
{+79.97/28 0.2} 53. Nd1 {-24.20/100 0.9} Qcxd1+ {+79.98/28 0.2} 54. Kf2
{-24.30/110 0.9} Bc5# {+79.99/28 0.2}
{Xboard adjudication: Checkmate} 0-1
[/pgn]

At least a king safety test position should be added.

[Henk]

Doubt if Skippers center control term is useful. Here SkipperWinB is using an extra center control term in evaluation.

[pgn]
[Event "Computer Chess Game"]
[Site "HP"]
[Date "2017.08.22"]
[Round "-"]
[White "SkipperWinb"]
[Black "SkipperWinb_WAC50"]
[Result "0-1"]
[TimeControl "120"]
[Annotator "1. -0.01 1... +0.06"]

1. d4 {-0.01/100} Nh6 {+0.06/120 2.0} 2. Nc3 {+0.12/100 2.0} b6
{+0.07/120 2.0} 3. e4 {+0.48/80 1.9} Ba6 {-0.03/120 2.0} 4. h4
{+0.49/110 1.9} g6 {+0.07/110 1.9} 5. Be2 {+0.57/100 1.9} c6
{-0.03/120 1.9} 6. Bxa6 {+0.35/80 1.8} Nxa6 {+0.04/130 1.9} 7. g3
{+0.42/80 1.8} Rg8 {-0.04/110 1.8} 8. Qf3 {+0.45/90 1.8} Rh8
{+0.00/100 1.8} 9. Be3 {+0.29/90 1.8} Bg7 {+0.03/100 1.8} 10. Qe2
{+0.39/90 1.7} Qc8 {-0.02/100 1.7} 11. Bf4 {+0.40/80 1.7} Rg8
{-0.06/100 1.7} 12. h5 {+0.55/100 1.7} Rh8 {+0.00/110 1.7} 13. Qd2
{+0.45/100 1.7} Ng4 {+0.04/120 1.6} 14. h6 {+0.47/100 1.6} Bf8
{+0.08/120 1.6} 15. e5 {+0.57/110 1.6} d6 {-0.02/110 1.6} 16. f3
{+0.91/120 1.6} dxe5 {-0.12/130 1.6} 17. dxe5 {+0.86/130 1.5} g5
{-0.55/120 1.5} 18. fxg4 {+0.76/120 1.5} gxf4 {-0.59/120 1.5} 19. Qxf4
{+0.81/110 1.5} Rg8 {-0.59/100 1.5} 20. Rh4 {+0.87/100 1.5} Rh8
{-0.52/100 1.5} 21. Rh5 {+0.69/90 1.5} Qe6 {-0.50/90 1.4} 22. Nf3
{+0.87/90 1.4} Rg8 {-0.57/100 1.4} 23. Ng5 {+0.77/100 1.4} Qg6
{-0.51/110 1.4} 24. Qa4 {+0.84/80 1.4} Nc5 {-0.51/110 1.4} 25. Qc4
{+0.74/110 1.3} Rb8 {-0.53/90 1.3} 26. Qf4 {+0.77/90 1.3} Ra8
{-0.63/100 1.3} 27. b4 {+0.83/100 1.3} Nd7 {-0.33/90 1.3} 28. Qe4
{+0.53/80 1.3} Nxe5 {+0.07/110 1.3} 29. Nxh7 {+0.43/110 1.3} Nxg4
{+0.17/110 1.3} 30. Nxf8 {+0.33/100 1.3} Qf6 {+1.00/110 1.2} 31. O-O-O
{+0.23/110 1.2} Nf2 {+1.10/110 1.2} 32. h7 {-2.27/100 1.2} Rh8
{+1.20/120 1.2} 33. Qe5 {-2.37/110 1.2} Qxe5 {+3.30/120 1.2} 34. Rxe5
{-3.09/120 1.2} Nxd1 {+3.39/130 1.2} 35. Kxd1 {-3.05/130 1.1} Kxf8
{+3.49/120 1.2} 36. b5 {-3.34/110 1.1} Rd8+ {+3.64/120 1.1} 37. Kc1
{-3.26/120 1.1} cxb5 {+3.54/130 1.1} 38. Nxb5 {-3.34/110 1.1} Rd7
{+3.58/120 1.1} 39. Rh5 {-3.26/110 1.1} Kg7 {+3.64/120 1.1} 40. Rg5+
{-3.36/110 1.0} Kh6 {+3.60/130 1.1} 41. Rf5 {-3.40/110 1.0} e6
{+3.59/120 1.0} 42. Rf4 {-3.41/120 1.0} Rxh7 {+3.63/110 1.0} 43. Re4
{-3.51/100 1.0} Rg7 {+3.66/120 1.0} 44. g4 {-3.51/110 1.0} Rb7
{+3.67/120 1.0} 45. Nd6 {-3.34/120 1.0} Rb8 {+3.58/130 1.0} 46. Nc4
{-3.25/110 1.0} Rh8 {+3.65/120 1.0} 47. Rf4 {-3.35/110 0.9} Rd8
{+3.60/120 0.9} 48. Ne5 {-3.34/110 0.9} Rf8 {+3.52/120 0.9} 49. Nd7
{-3.37/100 0.9} Rfg8 {+3.62/110 0.9} 50. Ne5 {-3.47/110 0.9} f5
{+3.59/120 0.9} 51. gxf5 {-3.41/120 0.9} Kg5 {+3.62/130 0.9} 52. Rf3
{-3.38/110 0.9} exf5 {+3.66/120 0.9} 53. Nc6 {-3.37/100 0.9} Rf8
{+3.67/110 0.8} 54. Rg3+ {-3.43/90 0.9} Kh6 {+3.64/130 0.8} 55. Rh3+
{-3.26/110 0.8} Kg6 {+3.67/130 0.8} 56. Rg3+ {-3.33/100 0.8} Kf6
{+3.58/120 0.8} 57. Rxg7 {-3.43/120 0.8} Kxg7 {+3.52/130 0.8} 58. Kb2
{-3.53/110 0.8} a6 {+3.48/110 0.8} 59. Ne5 {-3.55/100 0.8} Rh8
{+3.56/110 0.8} 60. Nc4 {-3.55/90 0.8} Rh6 {+3.51/110 0.8} 61. Ne5
{-3.56/100 0.8} Re6 {+3.58/110 0.8} 62. Nd3 {-3.51/100 0.7} Re7
{+3.67/100 0.8} 63. Nf4 {-3.49/100 0.7} Ra7 {+3.67/110 0.7} 64. Ng2
{-3.49/100 0.7} b5 {+3.63/110 0.7} 65. a3 {-3.54/100 0.7} Rb7
{+3.56/110 0.7} 66. Nh4 {-3.53/110 0.7} Kf6 {+3.58/120 0.7} 67. c3
{-3.49/110 0.7} Rd7 {+3.65/100 0.7} 68. Kb3 {-3.59/110 0.7} Rd2
{+3.66/110 0.7} 69. Nf3 {-3.53/130 0.7} Rg2 {+3.56/110 0.7} 70. Ne1
{-3.63/110 0.7} Rg7 {+3.63/110 0.7} 71. Kb4 {-3.51/100 0.7} Rh7
{+3.60/100 0.6} 72. Kc5 {-3.49/100 0.6} Rh1 {+3.53/100 0.6} 73. Ng2
{-3.56/110 0.7} Kg7 {+3.58/100 0.6} 74. Kb4 {-3.60/100 0.6} Rh8
{+3.66/100 0.6} 75. Ka5 {-3.59/100 0.6} Ra8 {+3.66/120 0.6} 76. Nf4
{-3.52/110 0.6} Kf8 {+3.56/110 0.6} 77. Kb6 {-3.42/120 0.6} Ke7
{+3.46/120 0.6} 78. Kb7 {-2.58/100 0.6} Re8 {+3.36/120 0.6} 79. Ka7
{-2.68/100 0.6} Kd6 {+3.58/120 0.6} 80. Kb7 {-3.53/90 0.6} Re3
{+3.59/110 0.6} 81. Kxa6 {-3.58/120 0.6} Kc6 {+3.69/120 0.6} 82. a4
{-3.68/110 0.6} bxa4 {+5.88/130 0.5} 83. c4 {-3.78/120 0.6} a3
{+6.89/130 0.5} 84. Nd5 {-3.88/130 0.5} a2 {+6.88/120 0.5} 85. Nb4+
{-3.98/140 0.6} Kc5 {+6.96/120 0.5} 86. Nxa2 {-4.08/130 0.5} Kxc4
{+6.87/110 0.5} 87. Ka5 {-6.88/110 0.5} f4 {+6.91/110 0.5} 88. Kb6
{-17.09/110 0.5} f3 {+7.01/120 0.5} 89. Nb4 {-17.07/110 0.5} Kxb4
{+7.11/120 0.5} 90. Ka6 {-2621.32/100 0.5} Re7 {+2621.33/110 0.5} 91. Kb6
{-2621.34/130 0.5} f2 {+2621.25/110 0.5} 92. Kc6 {-2621.36/120 0.5} Re1
{+2621.39/110 0.5} 93. Kd5 {-2621.34/100 0.5} Re2 {+2621.35/100 0.5} 94.
Kd6 {-2621.34/110 0.5} Re3 {+2621.30/100 0.5} 95. Kc6 {-2621.34/110 0.5}
Kc4 {+2621.37/100 0.4} 96. Kb6 {-2621.34/100 0.5} Re7 {+2621.35/100 0.4}
97. Kc6 {-2621.36/110 0.4} f1=Q {+2621.36/100 0.4} 98. Kb6
{-2621.38/100 0.4} Qf2+ {+2621.45/90 0.4} 99. Ka6 {-2621.40/90 0.4} Qa7#
{+2621.49/90 0.4}
{Xboard adjudication: Checkmate} 0-1
[/pgn]

[hgm]

So there is a lot of chess knowledge needed to win games that is not needed for doing well on WAC... And Fairy-Max already seems to have some of it.

[Henk]

Skippers Center Control term may be useful. That version won 8 out of 10 games. So posted game probably bad luck. My computer making too much noise on longer tests. I even fear it might get broken. Don't know yet what causes it.

[Daniel Anulliero]

I have a feeling , speaking with this guy and speaking with a wall is the same
Poor Skipper , it'll be never improve. ..