Man vs machine - how to beat chess engine

[lkaufman]

Komodo 4 is much stronger than Komodo 3 from the human point of view...after maybe 30 loses here is the first draw...

[Event "5 min. game (+3 s/per move for man)"]
[Date "2012.06.13"]
[White "Kvicala, Miroslav"]
[Black "Komodo64 no-sse Version 4"]
[Result "1/2-1/2"]
[ECO "C02"]
[WhiteElo "2035"]

1. e4 e6 2. d4 d5 3. e5 c5 4. c3 b6 5. Nf3 Qd7 6. Bd3 Ba6 7. Qe2 Bxd3 8. Qxd3
Nc6 9. O-O Nge7 10. a3 c4 11. Qc2 Na5 12. Nbd2 Nec6 13. Rb1 Be7 14. Re1 h6 15.
Nf1 Nb3 16. Bf4 O-O 17. Ng3 a6 18. Nd2 Nca5 19. Nxb3 Nxb3 20. Qe2 Kh8 21. Nf1
Rae8 22. Nd2 Nxd2 23. Qxd2 g5 24. Bg3 b5 25. f4 Rg8 26. Rf1 Ref8 27. Rf2 Bd8
28. Rbf1 Rg7 29. f5 Rfg8 30. Rf3 Qc6 31. f6 Rh7 32. Bf2 g4 33. Rf4 Bb6 34. Be3
h5 35. g3 a5 36. R4f2 Rg6 37. Ra1 a4 38. Bg5 Qc8 39. Raf1 Qg8 40. Bh4 Qf8 41.
Kg2 Qh6 42. Qxh6 Rhxh6 43. Rf4 Kg8 44. h3 Kf8 45. hxg4 hxg4 46. Rh1 Ke8 47.
Rhf1 Kd7 48. R1f2 Kc6 49. Rf1 Bc7 50. Re1 Rg8 51. Re2 Bb6 52. Re1 Rh5 53. Re2
Rg6 54. Re1 Rh7 55. Re2 Rg8 56. Re1 Rh6 57. Re2 Kb7 58. Re1 Rhg6 59. Re2 Rh8
60. Re1 Kc6 61. Re2 Bc7 62. Re1 Rc8 63. Re2 Rgg8 64. Re1 Ra8 65. Re2 Kb7 66.
Re1 Kb6 67. Re2 Rac8 68. Re1 Rcd8 69. Re2 Kc6 70. Re1 Rde8 71. Re2 Ra8 72. Re1
Rad8 73. Re2 Ba5 74. Re1 Bb6 75. Re2 Ra8 76. Re1 Rae8 77. Re2 Rb8 78. Re1 Rbc8
79. Kf2 Rg6 80. Kg2 Ra8 81. Ra1 Bc7 82. Rf2 Rag8 83. Re2 Rh8 84. Re3 Rh5 85.
Re2 Bb6 86. Rf2 Rhh6 87. Re2 Kd7 88. Rd2 Bc7 89. Re2 Rh8 90. Re3 Rh5 91. Re2
Rgh6 92. Re3 Rf5 93. Re2 Kc6 94. Rf2 Rxf2+ 95. Kxf2 Rh5 96. Rf1 Bxe5 97. dxe5
Rxe5 98. Re1 Rxe1 99. Kxe1 Kd6 100. Ke2 e5 101. Ke3 Ke6 102. Kf2 d4 103. Ke2
Kd5 104. Kd2 d3 105. Ke3 Ke6 106. Bg5 Kf5 107. Bh4 e4 108. Kd2 Kg6 109. Ke3 Kf5
110. Kd2 Kg6 111. Ke3 Kf5 1/2-1/2

Assuming this was played on a phone game, not a full computer, I would expect Komodo to play around the same level as the world's top few human players, i.e. around 2800. Playing White every game and taking 3" per move handicap probably reduces that to something like 2650. A score of 0.5 to 30.5 implies about a 600 elo difference, about the difference between 2650 and your rating. So at least in this match, results are perfectly in line with expectations.

[overlord]

Yes. I can make draw against every engine on smartphone relative easily (on PC it is much more difficult but possible)...10..h6 (engines doesn´t play always same moves in same postions - one guy in this discussion explained it in detail). Many people simply thinks that human has better chances against engines in long game but is is not true, it is better to play blitz due to engine hozizon when engines underestimates kings attacks.

[overlord]

Larry,

first I would like to say why I posted all those games. Harvey Williams some time ago accused some player that he is cheater when he poster one or two draw against Hiarcs (he was using iphone and ELO of the player was almost 2300). S o I wanted to show that it is possible to make some win or draw against every engine on smartphones - number of loses is not relevant...if I will lose against Kramnik 1000 times and win once...nobody will care about those 1000 loses It is the same with engines. Moreover I am very experienced with computer chess (we have a lot of successes in free style - Equidistance, Etaoin Shrdlu), so I believe that I can achieve better results against engines that much stronger players. There is not one chess...if you are playing against engines you have change openings, playing style...simply everything. On the contrary for me was very suspicious matches as Kramnik-Fritz and so on...I still believe that strongest GM´s would be able totally destroy Fritz or similar engines...OK Houdini is different story...

At the end I discovered that most pleasant engine to play against it is Critter

[overlord]

This game was played without any increment for human...I really don´t know why so many people believe that they can´t make good result against engine...it is just enough to make normal moves...please see following game...
[Event "5 min. game"]
[Date "2012.06.20"]
[White "Kvicala, Miroslav"]
[Black "Critter 1.6 32-bit"]
[Result "1/2-1/2"]
[WhiteElo "2035"]

1. d4 Nf6 2. c4 e5 3. dxe5 Ng4 4. Bf4 Nc6 5. Nf3 Bb4+ 6.
Nbd2 Qe7 7. a3 Ngxe5 8. Nxe5 Nxe5 9. e3 Bxd2+ 10. Qxd2 d6 11. Be2 O-O 12. O-O
Re8 13. Rfd1 b6 14. Rac1 Bb7 15. b4 Rad8 16. Qc3 Nd7 17. Bf3 Bxf3 18. gxf3 Ne5
19. Bxe5 dxe5 20. Qc2 c5 21. b5 Qe6 22. Rxd8 Rxd8 23. Rd1 Rxd1+ 24. Qxd1 h6 25.
Qd5 Qxd5 26. cxd5 Kf8 27. e4 Ke7 28. Kf1 h5 29. Ke2 h4 30. a4 Kd6 31. Ke1 h3
32. Ke2 Kd7 33. Ke1 f6 34. Ke2 Kd6 35. Ke1 Kc7 36. Ke2 Kd8 37. Ke1 g6 38. Ke2
g5 39. Ke1 Ke7 40. Ke2 Kd6 41. Ke1 Kc7 42. Ke2 Kd7 43. Ke1 Kd8 44. Ke2 Ke7 45.
Ke1 Kd7 46. Ke2 Kc8 47. Ke1 Kb8 48. Ke2 Kc7 49. Ke1 Kb7 50. Ke2 Kc8 51. Ke1 Kc7
52. Ke2 Kd7 53. Ke1 Ke7 54. Ke2 Kf7 55. Ke1 Kg7 56. Ke2 Kg8 57. Ke1 Kg7 58. Ke2
Kf7 59. Ke1 Kg6 60. Ke2 1/2-1/2

[Kyodai]

Yes, a very solid game vs Critter 1.6. Natural, simple moves from white
in a variation that doesn't give black much play. White has a slightly better structure the game through and the draw was easy. An instructive example of how to get half a point in blitz vs a strong engine. This is no miracle...

[overlord]

The game between England and Italy was little bit boring so I played also litlle bit boring game against Critter

[Event "15 min. game"]
[Site "Microsoft"]
[Date "2012.06.24"]
[Round "?"]
[White "Kvicala, Miroslav"]
[Black "Critter 1.6 32-bit"]
[Result "1/2-1/2"]
[ECO "A84"]
[WhiteElo "2035"]
[PlyCount "351"]

1. d4 f5 2. Nf3 Nf6 3. c4 e6 4. g3 b6 5. Bg2 Bb7 6. O-O Bd6 7. Nc3 O-O 8. Bf4
Bxf4 9. gxf4 Na6 10. e3 Bxf3 11. Bxf3 c6 12. Rc1 Qe8 13. Kh1 Qf7 14. Rg1 Nc7
15. a3 Kh8 16. Qd2 Rac8 17. b4 a6 18. c5 b5 19. a4 g6 20. axb5 axb5 21. Ra1 Ra8
22. Qb2 Rfe8 23. Rxa8 Rxa8 24. Ra1 Rxa1+ 25. Qxa1 Qf8 26. Qa7 Qc8 27. Kg2 Kg8
28. Kf1 Kg7 29. Ke2 Kf7 30. Kd2 Nfe8 31. Kc2 Qd8 32. Qa1 Qh4 33. Qg1 Qh3 34.
Qg2 Qxg2 35. Bxg2 Ke7 36. Kb3 Kf6 37. h4 Kf7 38. Ne2 Nf6 39. Bf3 Kg7 40. Ng3
Ncd5 41. Be2 h5 42. Bf3 Ng4 43. Nh1 Nh2 44. Bg2 Kf8 45. Ng3 Kf7 46. Nh1 Ke7 47.
Ng3 Ng4 48. Nh1 Kf6 49. Bf3 Nh2 50. Bg2 Kf7 51. Ng3 Ng4 52. Nh1 Ke7 53. Bf3 Nh2
54. Bg2 Ke8 55. Ng3 Ng4 56. Nh1 Nc7 57. Bf3 Nh2 58. Bg2 Kf7 59. Kc3 Ng4 60. Bf3
Ke7 61. Bg2 Ke8 62. Bf3 Kf8 63. Bg2 Kg8 64. Bf3 Kf7 65. Bg2 Kf8 66. Bf3 Nh2 67.
Bg2 Ke7 68. Ng3 Ke8 69. Nh1 Kf8 70. Ng3 Kg7 71. Nh1 Ng4 72. Bf3 Kh7 73. Bg2 Nf6
74. Ng3 Ncd5+ 75. Kb3 Ng4 76. Nh1 Kh6 77. Bf3 Kg7 78. Bg2 Kh7 79. Bf3 Nc7 80.
Bg2 Kg8 81. Bf3 Nh2 82. Bg2 Nd5 83. Ng3 Kh7 84. Nh1 Kg7 85. Ng3 Ne7 86. Nh1 Kf8
87. Ng3 Ng4 88. Nh1 Ke8 89. Bf3 Kd8 90. Bg2 Kc7 91. Bf3 d6 92. Bg2 Nd5 93. Bf3
Nh2 94. Bg2 Kd8 95. Ng3 Ng4 96. Nh1 Kd7 97. Bf3 Ndf6 98. Bg2 d5 99. Bf3 Nh2
100. Bg2 Ne4 101. Kc2 Ng4 102. Kd3 Ngxf2+ 103. Nxf2 Nxf2+ 104. Ke2 Ng4 105. Bf3
Kd8 106. Bg2 Ke7 107. Bf3 Kf6 108. Bg2 Kf7 109. Bf3 Nh2 110. Kf2 Ke7 111. Ke2
Ng4 112. Bg2 Kf6 113. Bf3 Kf7 114. Bg2 Ke8 115. Bf3 Kd7 116. Bg2 Kc7 117. Bf3
Kb7 118. Bg2 Kc8 119. Bf3 Nf6 120. Bg2 Kd7 121. Bf3 Ke7 122. Bg2 Ne8 123. Bf3
Kd7 124. Bg2 Nc7 125. Kd3 Kd8 126. Bf3 Ke7 127. Bg2 Kf7 128. Bf3 Ne8 129. Bg2
Nf6 130. Bf3 Ke7 131. Ke2 Kf8 132. Bg2 Ke8 133. Bf3 Ne4 134. Bxe4 fxe4 135. Kf2
Kf8 136. Kg3 Ke7 137. Kh3 Kf7 138. Kg3 Kf6 139. Kh3 Ke7 140. Kg3 Kd8 141. Kh3
Kc7 142. Kg3 Kb8 143. Kh3 Kb7 144. Kg3 Kc7 145. Kh3 Kd8 146. Kg3 Kd7 147. Kh3
Kc8 148. Kg3 Kb7 149. Kh3 Ka8 150. Kg3 Kb8 151. Kh3 Kc7 152. Kg3 Kc8 153. Kh3
Kd7 154. Kg3 Ke7 155. Kh3 Kf6 156. Kg3 Kg7 157. Kh3 Kg8 158. Kg3 Kh8 159. Kh3
Kg7 160. Kg3 Kf6 161. Kh3 Kf5 162. Kg3 e5 163. fxe5 g5 164. hxg5 Kxg5 165. e6
Kf6 166. Kh4 Kxe6 167. Kxh5 Kf5 168. Kh4 Kg6 169. Kg4 Kf6 170. Kf4 Ke6 171. Kg5
Ke7 172. Kf5 Kf7 173. Ke5 Ke7 174. Kf5 Kf7 175. Ke5 Ke7 176. Kf5 1/2-1/2

[overlord]

When Critter goes out of book it plays couple of strategically very suspicious moves...the end comes very soon...

[Event "5 min. game"]
[Date "2012.06.28"]
[White "Critter 1.6 32-bit"]
[Black "Kvicala, Miroslav"]
[Result "0-1"]
[ECO "E97"]
[BlackElo "2035"]

1. d4 Nf6 2. c4 g6 3. Nc3 Bg7 4. e4 d6 5. Nf3 O-O 6. Be2 e5 7. O-O Nc6 8. d5
Ne7 9. h3 Ne8 10. Qb3 h6 11. Bd2 f5 12. Rad1 f4 13. c5 g5 14. Bd3 Ng6 15. cxd6
cxd6 16. Rc1 h5 17. Ne1 g4 18. hxg4 hxg4 19. a4 Qh4 20. Ne2 f3 21. Ng3 Nf4 22.
Bb5 Rf6 23. Bxf4 exf4 24. Nh5 Qxh5 25. Bxe8 Qh4 26. Bh5 Rh6 27. Rxc8+ Rxc8 28.
Qxb7 Rf8 29. Qe7 Qxe7 30. Nxf3 gxf3 31. gxf3 Rxh5 32. Rc1 Qg5+ 33. Kf1 Rh1+ 34.
Ke2 Rxc1 35. Kd3 Qe5 36. b3 Qd4+ 37. Ke2 Qd1# 0-1

[overlord]

One more win in king´s indian defence...9.h3 is simply losing mistake...

[Event "10 min game"]
[Date "2012.06.30"]
[White "Critter 1.6 32-bit"]
[Black "Kvicala, Miroslav"]
[Result "0-1"]
[BlackElo "2035"]

1. d4 Nf6 2. c4 g6 3. Nc3 Bg7 4. e4 d6 5. Nf3 O-O 6. Be2
e5 7. O-O Nc6 8. d5 Ne7 9. h3 Ne8 10. Qd3 h6 11. Nd2 f5 12. b4 f4 13. Bb2 g5
14. c5 a6 15. Nc4 Ng6 16. Na5 Nh4 17. f3 h5 18. Rf2 Rf7 19. Rc1 Bf8 20. Ba3 Rg7
21. Kh1 g4 22. hxg4 hxg4 23. fxg4 Bxg4 24. Bxg4 Rxg4 25. Qh3 Qg5 26. c6 b6 27.
Nc4 Rg3 28. Qe6+ Kg7 29. Ne2 Rxg2 30. Rcf1 Nf6 31. Qh3 Rxf2 32. Rxf2 Be7 33.
Bb2 Rh8 34. Nxf4 exf4 35. Rh2 Ng6 36. Qxh8+ Nxh8 37. Rg2 Qxg2+ 38. Kxg2 b5 39.
Na5 Nf7 40. Kf3 Ne5+ 41. Bxe5 dxe5 42. a3 Ne8 43. Nb3 Kg6 44. Kg4 Nf6+ 45. Kf3
Kh5 46. Nc1 Ne8 47. Nb3 Kh4 48. Na1 Nd6 49. Nb3 Nc4 50. Nc5 a5 51. Na6 axb4 52.
axb4 Bd6 53. Nc5 Kh3 54. Kf2 Na3 55. Nd7 Nc2 56. Nc5 Nxb4 57. Nb7 Na6 58. Kf3
b4 59. Na5 Bc5 60. Nc4 b3 61. Nb2 Ba3 62. d6 Bxd6 63. Kf2 Nc5 64. Kf3 Ne6 65.
Nd1 Nd4+ 66. Kf2 Ba3 67. Ke1 b2 68. Nxb2 Bxb2 69. Kf2 f3 70. Kg1 Kg3 71. Kf1
Nc2 72. Kg1 f2+ 73. Kf1 Ne3+ 74. Ke2 Bd4 75. Kd2 Nc4+ 76. Ke2 Kg2 77. Kd1 f1=Q+
78. Kc2 Qe2+ 79. Kb3 Nd2+ 80. Kc2 Nxe4+ 81. Kb3 Qb2+ 82. Ka4 Bc5 83. Ka5 Qb4+
84. Ka6 Qb6# 0-1

[Nelson Hernandez]

Miroslav, if you want to be taken seriously you need, as a minimum, to provide annotation. We need to know when your opponent left book and how long both sides thought in each position.

Many people have preceded you in making claims similar to yours. Years ago we heard about the wonders of blockaded positions. From time to time various charlatans have come along making outlandish or provocative statements. I am not saying you are one of them, but you must understand the history and in some way address those concerns in the same way an academic would confront contra-arguments in peer review.

My own observation is that games such as these are entirely decided in the first 15 moves or so--they are purely an opening book trick. There are several ways to pull it off in practice. Your games tend to fall into one of these two general categories:

1. Opening book leads game directly into drawish position where better alternatives were available.
2. Opening book establishes slightly superior position and then one or more moves are played (in or out of book) that are subtly suboptimal, leading to draw or defeat.

In all cases I note that your play after leaving book is exceptionally strong for a person playing blitz, leading to natural doubts. A superficial impression: you don't blunder. It seems the engine blunders more often than you do! What are we to make of that?

In short, engines without professional quality books are vulnerable, yes, but if that is your whole point then you are only recapitulating what has been known for over 40 years. And I might add that the current definition of "professional quality" is much, much higher than it was five or ten years ago.

[overlord]

Nelson,

in some points I have agree with you. Engines are much weaker without good book. In most of my games default book of Stockfish was used (for smartphones in Droidfish). You said that many openings are drawish, yes they are but it is normal in chess...of course that I will not play Najdorf sicilian against engine. In engines testing 8move book is normally used - to test real engine strength (not quality of book). I guess that 8 move book ahould be enough for top engines to beat average human easily but reality is different (in first phase of the game - I don´t mean only opening people are superior to engines - knowledge of structures, patterns are something what engines simply doesn´t have. Here you can see game played under absolutly fair conditions against Rybka 4.1 64 bit on my old Q6600 (1 CPU was used), 5 minutes for both of us (no increment, no take backs or something like that). Everybody can check it...

[Event "Blitz:5'"]
[Site "Microsoft"]
[Date "2012.07.02"]
[Round "?"]
[White "Kvicala, Miroslav"]
[Black "Deep Rybka -4.1-x64"]
[Result "1/2-1/2"]
[WhiteElo "2035"]
[PlyCount "66"]
[TimeControl "300"]

{256MB, Mirekbook.ctg} 1. d4 {0} d5 {0} 2. c4 {3} e6 {0} 3. Nc3 {1} Be7 {0} 4.
Nf3 {2} Nf6 {0} 5. Bg5 {1} c6 {0} 6. e3 {3} Nbd7 {0} 7. Rc1 {1} h6 {0} 8. Bh4 {
2} O-O {0} 9. Qc2 {2} a6 {0} 10. a3 {1} b5 {9} 11. c5 {10} a5 {0} 12. b4 {6}
Bb7 {16} 13. Bd3 {4} g5 {19} 14. Bg3 {8} Nh5 {0} 15. O-O {3} Nxg3 {16} 16. hxg3
{0} g4 {8} 17. Nd2 {20} Nf6 {3} 18. f4 {30} a4 {9} 19. Kf2 {15} Re8 {9} 20. Rh1
{2} Bf8 {3} 21. Ne2 {13} Qc7 {12} 22. Rh4 {10} Rad8 {11} 23. Rch1 {1} Bg7 {20}
24. Qd1 {7} Rc8 {17} 25. Bb1 {3} Rb8 {7} 26. Qc2 {3} Kf8 {6} 27. Nc1 {12} h5 {7
} 28. Nd3 {10} Ke7 {0} 29. Ne5 {3} Rg8 {0} 30. Qd3 {23} Rh8 {5} 31. Qc2 {2}
Rhg8 {8} 32. Qd3 {1} Rh8 {7} 33. Qc2 {1} Rhg8 {7} 1/2-1/2