The cause of extreme piece shuffling

Discussion of anything and everything relating to chess playing software and machines.

Moderators: hgm, Rebel, chrisw

User avatar
hgm
Posts: 27812
Joined: Fri Mar 10, 2006 10:06 am
Location: Amsterdam
Full name: H G Muller

Re: The cause of extreme piece shuffling

Post by hgm »

When engines don't see a way to make progress, but can easily defend against the opponent making progress, they just randomly pick one of the moves that maintain the status quo. They have no tendency to organize their own pieces in what they would consider the optimal way, and are prepared to move to positions they actuall consider pretty poor, as long as they see that within their search horizon they will be able to move back to that optimum. So they basically just make a random walk in the region of game-state space that surrounds the perceived local optimum with a maximum distance of their search depth.

A delayed-loss bonus would give them an incentive to actually move towards the local optimum, from which they might be able to see a way to make progress (while far away from it they need all their depth to worry about how to get to that local optimum, and can certainly not see what lies behind it). But not many engines have that.
User avatar
Nordlandia
Posts: 2821
Joined: Fri Sep 25, 2015 9:38 pm
Location: Sortland, Norway

Re: The cause of extreme piece shuffling

Post by Nordlandia »

Here is the full game.

[pgn][Event "EXPERT-LENOVA-1, Slow 150m+60s"]
[Site "EXPERT-LENOVA-1"]
[Date "2016.01.10"]
[Round "1"]
[White "SugaR 2.0 64 POPCNT"]
[Black "Komodo 9.3 64-bit"]
[Result "1-0"]
[ECO "C60"]
[Annotator "0.38;0.22"]
[PlyCount "267"]
[TimeControl "9000+60"]

{Intel(R) Core(TM) i5-5200U CPU @ 2.20GHz 2194 MHz W=39.6 plies; 2,951kN/s; 8,
255,934 TBAs B=35.4 plies; 2,893kN/s; 23,148,237 TBAs} 1. e4 {Both last book
move} e5 2. Nf3 Nc6 3. Bb5 Nge7 4. O-O {0.38/31 285} g6 {0.22/27 173} 5. c3 {
0.28/33 280} a6 {0.21/29 193} 6. Ba4 {0.25/34 419 (Bc4)} Bg7 {0.21/30 284} 7.
d4 {0.24/31 67} exd4 {0.21/29 400} 8. cxd4 {0.28/33 151} O-O {0.18/29 219} 9.
d5 {0.17/27 39 (Bxc6)} Na5 {0.14/27 154} 10. Nc3 {0.16/31 465 (Re1)} b5 {
0.12/29 527 (d6)} 11. Bc2 {0.13/32 302 (Bb3)} d6 {0.12/27 109} 12. Rb1 {
0.27/32 392 (h3)} Re8 {0.13/29 231 (Nc4)} 13. Re1 {0.10/31 519} Bg4 {0.18/29
475 (Nc4)} 14. h3 {0.20/32 238} Bxf3 {0.22/30 258} 15. Qxf3 {0.21/33 121} c5 {
0.13/32 228 (Nc4)} 16. b3 {0.34/31 320 (dxc6)} Nf5 {0.22/29 231} 17. Ne2 {
0.22/33 229} Nd4 {0.23/28 102} 18. Nxd4 {0.23/31 41} Bxd4 {0.22/29 385} 19. Be3
{0.34/31 239} Bg7 {0.21/28 206 (Bxe3)} 20. g3 {0.42/30 106 (Qe2)} c4 {0.32/30
234} 21. h4 {0.49/32 670 (Rec1)} h5 {0.30/28 143 (Qd7)} 22. Kg2 {0.56/29 143
(Rec1)} Qd7 {0.37/31 375 (Qc8)} 23. Qd1 {0.68/29 182 (Rec1)} Rec8 {0.51/29 637}
24. b4 {0.77/34 276 (Re2)} Nb7 {0.60/28 143} 25. a4 {0.82/32 80} Bc3 {0.60/28
236} 26. Bd2 {0.84/32 144} Bh8 {0.65/29 398 (Bd4)} 27. a5 {0.91/32 131 (Re2)}
Re8 {0.95/30 345} 28. Be3 {0.96/31 77} Re7 {0.93/31 355 (Qe7)} 29. f4 {1.13/30
196 (Bd4)} Bc3 {1.00/30 356} 30. Bd2 {1.13/34 136 (Rh1)} Bf6 {1.14/31 378
(Bxd2)} 31. f5 {1.38/31 188} Be5 {1.27/29 282} 32. Rf1 {1.55/33 877} Nd8 {
1.41/29 599 (Kg7)} 33. f6 {1.90/33 155 (Be3)} Re8 {1.66/31 122} 34. Qf3 {
2.05/31 68} c3 {2.35/31 272} 35. Bxc3 {2.09/35 215} Rc8 {2.35/31 150} 36. Rb3 {
2.15/31 35} Rc4 {2.41/34 172} 37. Rf2 {2.34/33 252 (Qe3)} Qc8 {2.62/31 264
(Kh7)} 38. Bxe5 {2.46/37 223} dxe5 {2.32/32 186} 39. Bd3 {2.53/31 36} Rc1 {
2.71/32 137} 40. Qe3 {2.63/35 120} Kh7 {2.86/31 51} 41. Qe2 {2.70/31 41} Nb7 {
2.42/32 109 (Qc7)} 42. Rf5 {2.92/35 212} Ra1 {2.92/33 104} 43. Rxh5+ {2.93/34
31} Kg8 {3.09/34 73} 44. Qb2 {2.93/36 95 (Qe3)} Qc1 {3.01/38 159} 45. Qxc1 {
2.93/38 73} Rxc1 {3.10/34 57} 46. Rg5 {2.93/40 118} Nd6 {3.11/36 108} 47. Kf2 {
2.93/40 151} Kh7 {3.03/36 135 (Kh8)} 48. Ke2 {2.93/40 103} Kh6 {3.02/37 244
(Nc4)} 49. Rb2 {2.93/39 155 (Rb1)} Kh7 {3.03/39 138 (Nc4)} 50. Rc2 {2.93/43 208
} Rxc2+ {3.03/40 47} 51. Bxc2 {2.93/46 33} Rc8 {3.04/37 44 (Kh6)} 52. Bd3 {
2.93/41 207} Re8 {3.04/41 55} 53. Kf2 {2.93/45 72 (Kf3)} Kh8 {3.04/41 51 (Kg8)}
54. Ke3 {2.93/48 82 (Kf3)} Kh7 {3.03/39 235} 55. Kf3 {2.93/52 84 (Be2)} Kh8 {
3.04/34 97 (Kh6)} 56. Bb1 {2.93/49 244 (Kf2)} Kg8 {3.03/38 42 (Kh7)} 57. Ke2 {
2.93/51 102 (Kg2)} Kh7 {3.10/41 55} 58. Kd3 {2.93/52 38 (Bd3)} Kh6 {3.04/42 101
} 59. Ke3 {2.93/52 90 (Bc2)} Kh7 {3.04/42 115} 60. Bd3 {2.93/52 78} Kh6 {
3.04/43 66 (Kg8)} 61. Kf3 {2.93/50 335 (Kd2)} Kh7 {3.04/42 70} 62. Kg2 {
2.93/49 24 (Be2)} Kh6 {3.07/42 102 (Kh8)} 63. Kf2 {2.93/46 70} Kh7 {3.07/41 35}
64. Bb1 {2.93/45 26 (Ke2)} Nc4 {3.11/43 106 (Kg8)} 65. Bc2 {2.93/48 84 (Ke1)}
Nd6 {3.10/41 144 (Kh6)} 66. Kf1 {2.93/43 118 (Ke3)} Rc8 {3.08/37 55 (Kh6)} 67.
Bd3 {2.93/42 75} Re8 {3.06/40 54} 68. Ke2 {2.93/44 27 (Ke1)} Kg8 {3.07/39 49
(Kh6)} 69. Bc2 {2.93/45 118} Kh8 {3.02/38 85 (Kh7)} 70. Kd3 {2.93/45 75 (Ke3)}
Kh7 {3.03/37 32} 71. Bd1 {2.93/44 34} Nc4 {3.03/39 84 (Kh6)} 72. Bb3 {2.93/50
143 (Be2)} Nd6 {3.03/39 47} 73. Bc2 {2.93/51 22 (Ba2)} Nc4 {3.02/39 112 (Kh6)}
74. Bb1 {2.93/48 73 (Ke2)} Kh6 {3.02/41 75 (Nd6)} 75. Ke2 {2.93/48 161} Kh7 {
3.02/39 141 (Nd6)} 76. Bc2 {2.93/45 78 (Ba2)} Nd6 {2.99/34 31 (Kg8)} 77. Ke1 {
2.93/39 108} Kh6 {2.94/37 44 (Kg8)} 78. Bd3 {2.87/43 234} Kh7 {2.84/38 43} 79.
Bb1 {2.93/43 112 (Kd2)} Kh8 {2.87/35 77 (Kh6)} 80. Kf2 {2.93/43 150 (Bc2)} Kh7
{2.82/34 32} 81. Kf1 {2.84/41 231} Kg8 {2.78/36 79 (Kh6)} 82. Kg2 {2.75/41 129
(Bc2)} Kh7 {2.79/38 85} 83. g4 {2.74/38 67 (Kh2)} Kh6 {2.77/41 81 (Kh8)} 84.
Kf3 {2.74/46 57 (Bd3)} Nc4 {2.72/40 142 (Kh7)} 85. Bd3 {2.74/50 69 (Kf2)} Nd6 {
2.78/38 40} 86. Ke2 {2.74/48 139 (Kf2)} Nb7 {2.79/38 34 (Kh7)} 87. Ke3 {
2.74/41 61} Nd6 {2.73/38 28} 88. Bb1 {2.74/46 24 (Kf3)} Nc4+ {2.74/38 37 (Kh7)}
89. Ke2 {2.74/50 53 (Kf2)} Nd6 {2.73/40 47} 90. Kd2 {2.74/51 58 (Kd3)} Nc4+ {
2.73/38 47 (Kh7)} 91. Kd3 {2.74/49 50 (Ke1)} Kh7 {2.73/39 49 (Nd6)} 92. Ke2 {
2.74/50 68 (Ba2)} Nd6 {2.74/42 63} 93. Ke3 {2.74/43 17} Nc4+ {2.74/41 95 (Kh6)}
94. Kd3 {2.74/46 65 (Kf2)} Kh6 {2.74/39 75 (Nd6)} 95. Bc2 {2.74/50 63 (Ba2)}
Kh7 {2.76/41 49 (Nd6)} 96. Bb3 {2.74/50 67 (Ke2)} Nd6 {2.74/39 32} 97. Bd1 {
2.74/52 25} Nc4 {2.74/39 46} 98. Be2 {2.74/52 54} Kh6 {2.74/40 131 (Nd6)} 99.
Kc3 {2.74/49 67} Nd6 {2.74/39 32} 100. Bd3 {2.74/51 30} Kh7 {2.74/42 49 (Rc8+)}
101. Kb3 {2.74/49 53} Kh6 {2.74/42 31} 102. Kb2 {2.74/52 86} Nb7 {2.71/41 75}
103. Kc3 {2.74/52 37 (Kc2)} Nd6 {2.74/40 87 (Rc8+)} 104. Kb3 {2.74/42 57 (Kc2)}
Kh7 {2.76/39 84} 105. Kc2 {2.74/42 36} Rc8+ {2.73/37 35} 106. Kd1 {2.74/46 89}
Re8 {2.73/41 44} 107. Bb1 {2.74/49 580 (Kd2)} Kg8 {2.72/38 52 (Nb7)} 108. Kd2 {
2.70/47 145 (Kc1)} Kh7 {2.73/42 44} 109. Bc2 {2.70/46 271 (Bd3)} Kh6 {2.71/35
33} 110. Ke3 {2.70/39 31} Rc8 {2.71/32 34} 111. Bb1 {2.70/40 64 (Kd3)} Re8 {
2.72/42 145} 112. Bd3 {2.70/42 46 (Kf2)} Kh7 {2.72/41 112 (Nb7)} 113. Kf2 {
2.70/39 36 (Bf1)} Kh6 {2.59/39 99} 114. Kf3 {2.63/38 14 (Bb1)} Kh7 {2.59/38 63
(Nb7)} 115. Ke2 {2.70/40 73 (Bf1)} Kh8 {2.59/35 68 (Nb7)} 116. Kf2 {2.59/39
201 (Bc2)} Kh7 {2.27/35 72} 117. Kf1 {2.62/37 63 (Ke1)} Kh8 {2.41/35 96 (Kh6)}
118. Kg2 {2.70/32 80} Kh7 {2.26/33 28 (Kg8)} 119. Kh2 {2.53/36 60 (Kg1)} Kh6 {
2.26/32 44} 120. Kg3 {2.21/38 84 (Kg1)} Kh7 {1.93/31 37 (Nb7)} 121. h5 {
2.63/30 30} Kh6 {2.71/32 54} 122. Kh4 {2.71/31 22} Kh7 {3.07/32 56} 123. Be2 {
2.74/33 42} Nxe4 {3.07/33 43} 124. hxg6+ {2.78/30 16} fxg6 {3.21/34 79} 125. f7
{2.81/33 29} Rh8 {3.31/33 52} 126. f8=R {2.98/31 69 (f8Q)} Rxf8 {3.31/31 43}
127. Rxe5 {3.05/28 10} Rf4 {3.32/33 89} 128. Re6 {3.05/30 32 (Bxb5)} Kg7 {
2.45/29 78} 129. Bxb5 {3.14/27 14} axb5 {3.66/30 238 (Nf6)} 130. a6 {4.37/27 44
} g5+ {4.62/31 78} 131. Kh5 {4.56/24 14} Kf7 {6.53/25 39 (Nf6+)} 132. a7 {
38.96/33 43} Rf2 {250.00/46 23 (Rf3)} 133. Rxe4 {123.38/40 44} Ra2 {250.00/45
23} 134. Kxg5 {123.41/44 15} 1-0[/pgn]
User avatar
Marek Soszynski
Posts: 582
Joined: Wed May 10, 2006 7:28 pm
Location: Birmingham, England

Re: The cause of extreme piece shuffling

Post by Marek Soszynski »

hgm wrote:A delayed-loss bonus would give [engines] an incentive to actually move towards the local optimum, from which they might be able to see a way to make progress (while far away from it they need all their depth to worry about how to get to that local optimum, and can certainly not see what lies behind it). But not many engines have that.
Which engines do?
Marek Soszynski
User avatar
hgm
Posts: 27812
Joined: Fri Mar 10, 2006 10:06 am
Location: Amsterdam
Full name: H G Muller

Re: The cause of extreme piece shuffling

Post by hgm »

I know that Fairy-Max/micro-Max, Joker, Spartacus, KingSlayer, HaQiKi D, Shokidoki, HaChu all do it.
User avatar
Evert
Posts: 2929
Joined: Sat Jan 22, 2011 12:42 am
Location: NL

Re: The cause of extreme piece shuffling

Post by Evert »

hgm wrote:I know that Fairy-Max/micro-Max, Joker, Spartacus, KingSlayer, HaQiKi D, Shokidoki, HaChu all do it.
I sense a pattern! :lol: