Why computer versus computer games should always go to mate

[Monsieur Plastique]

I know many will say I am overly pedantic with my computer versus computer games, since I always use a 40 in 2 time control for slow games and always play them to mate, regardless of how far ahead one machine may be. This is how, for instance, I found the stalemate bug in the Saitek handheld Maestro (where it will sometimes end up drawing with stalemate in overwhelmingly won positions).

I am currently playing a match between Star Opal and the Saitek Cavalier where prior to the game below, Star Opal was leading 2-0.

In the third game, however, Star Opal did not capitalise on it's opening play, subsequently played inaccurately and ended up on the receiving end of a queenside attack by black with two connected passed pawns. After black queened a pawn the game was hopelessly lost. White was a queen down and mate was soon to follow.

But I continued to play the game, as I have seen too many swindles in computer versus computer games over the years. Here, Star Opal in a hopeless position manages to swindle a perpetual check out of Cavalier thanks to an inaccurate 45th move.

I am sure most people would have resigned for Star Opal at worst when black queened. Although for the most part the winning side will end up winning the game, I am glad I have stuck to my principal of playing to mate, despite it taking up more time.

As my brother (who has played computer versus computer games for over three decades now) says:


"...always, always, always play until the bitter end! NEVER assume! NEVER adjudicate!”

to which I would add:

"especially if you are debugging or beta testing an engine!!"

[Event "?"]
[Site "?"]
[Date "2013.07.28"]
[Round "?"]
[White "Novag Star Opal 24 MHz"]
[Black "Saitek Cavalier"]
[Result "1/2-1/2"]
[ECO "D39"]
[PlyCount "105"]

1. d4 Nf6 2. c4 e6 3. Nc3 Bb4 4. Nf3 d5 5. Bg5 dxc4 6. e4 b5 7. e5 h6 8. Bh4 g5
9. Bxg5 hxg5 10. exf6 Qxf6 11. Be2 Nc6 12. O-O Bxc3 13. bxc3 g4 14. Ne1 Qh6 15.
h3 gxh3 16. g4 O-O 17. Rb1 Bd7 18. a4 a6 19. axb5 axb5 20. Nc2 Qg5 21. Re1 Rfb8
22. Bf3 Kf8 23. Nb4 Nxb4 24. Bxa8 Nd3 25. Bf3 Nxe1 26. Qxe1 c5 27. Qe5 Qxe5 28.
dxe5 Ke7 29. Kh2 b4 30. cxb4 cxb4 31. Kxh3 b3 32. g5 c3 33. Be4 b2 34. Bc2 Rb5
35. f4 Rb4 36. Rf1 Ba4 37. Bxa4 Rxa4 38. Rb1 c2 39. Rxb2 c1=Q 40. Rb7+ Kf8 41.
Rb8+ Kg7 42. Rb3 Qxf4 43. Rg3 Ra2 44. g6 fxg6 45. Rxg6+ Kh8 46. Rg8+ Kh7 47.
Rg7+ Kh6 48. Rg6+ Kh5 49. Rg5+ Kh6 50. Rg6+ Kh7 51. Rg7+ Kh8 52. Rg8+ Kh7 53.
Rg7+ 1/2-1/2

[Theo]

Here's another one about "should always go to mate":

Mephisto Modena (white) vs Renaissance Maestro D 6Mhz

[fen]2b5/6R1/4p3/1k3qP1/3Q4/3P3r/6K1/8 b - - 0 73[/fen]

Black (Maestro) to move. The last two moves Maestro played a tempo, signalizing mate in 4 and mate in 3. Now, its mate in 2 starting with Qf3+.

Now Maestro plays 73 ... Ka6 instead of mating !!!! Why ????



The whole game:

[Date "2013.08.03"]
[White "Mephisto Modena"]
[Black "Maestro D 6Mhz"]
[Comment "mod2, maestro1"]
[Result "1-0"]

1.Nf3 c5 2.g3 e6 3.Bg2 Qa5 4.0-0 d5 5.d3 Bd6 6.e4 Ne7 7.e5 Bc7
8.Nbd2 Qa6 9.Nb3 Nd7 10.Be3 d4 11.Bf4 Ng6 12.Qd2 0-0 13.Rfe1
Nxf4 14.Qxf4 f6 15.Qe4 fxe5 16.a3 c4 17.dxc4 Qxc4 18.Nfd2 Qa6
19.Bf1 Qc6 20.Qd3 Nc5 21.Nxc5 Qxc5 22.Bg2 Kh8 23.Nc4 Rf7 24.Re2
a6 25.Rae1 Rf5 26.Be4 Rf6 27.Bxh7 Rh6 28.Be4 Kg8 29.Bf3 Qb5 30.b3
Rf6 31.a4 Qc5 32.Nxe5 Rf5 33.Nc4 Rf6 34.Be4 Rh6 35.f4 Rb8 36.Bg6
b5 37.axb5 axb5 38.Ne5 Bb6 39.b4 Qc3 40.Bf7+ Kf8 41.Rb1 Qxd3
42.cxd3 Bc7 43.Rf1 Bd6 44.Rb2 Ra8 45.Bg6 Ra6 46.Be4 Bxe5 47.fxe5+
Ke8 48.Rbf2 Rh8 49.Bg6+ Kd8 50.Rf8+ Rxf8 51.Rxf8+ Kc7 52.Rf7+
Kc6 53.Rxg7 Ra1+ 54.Kg2 Kb6 55.Be4 Ra2+ 56.Kh3 Rd2 57.g4 Rb2
58.g5 Rxb4 59.Kg4 Ra4 60.h4 b4 61.h5 b3 62.h6 b2 63.h7 b1=Q 64.h8=Q
Qg1+ 65.Kh3 Qe3+ 66.Kh2 Ra2+ 67.Bg2 Qxe5+ 68.Kh3 Qf5+ 69.Kh4
Rxg2 70.Qd8+ Kb5 71.Qxd4 Rh2+ 72.Kg3 Rh3+ 73.Kg2 Ka6 74.Ra7+
Kb5 75.Qc4+ Kb6 76.Qc7+ Kb5 77.Ra5+ Kb4 78.Rxf5 exf5 79.Qc4+
Ka3 80.Kxh3 Bd7 81.Qc5+ Kb3 82.Qd5+ Kc3 83.Qxd7 f4 84.Kg4 f3
85.Kxf3 Kd2 86.g6 Kc3 87.g7 Kb4 88.g8=Q Kc3 89.Qc7+ Kb4 90.Qgb8+
Ka4 91.Qca7# 1-0

[Steve B]

Hi Jon

basically i agree with you however i have one exception
there are a variety of computers(mostly the high end ones) that have a "Resign" feature
sometimes this is nothing more then the computer announcing a mate against itself
some however will resign when the eval hits -9.99

if a computer will resign of its own accord i will end the game there even if mate is not imminent
i consider the Resign feature part and parcel of the Program and if a computer chooses to quit..then keeping the game going is not correct

IMHO Regards
Steve

[Sargon1972]

Hello Guys,

Well the Risc 1mb ore Risc 2, resign very "quick" in tournament games
When they come below a scrore of -500 they resign.

But yes , they mostly resign when mate is coming , ore when they see it

Kind regards,Hans

[spacious_mind]

I also agree generally with playing to the end, especially with any computer below 1800 ELO (2000 USCF) unless they announce mate or resign.

[Event "U1400 Blue Stage 1 Group A"]
[Site "Active Chess 0:30/1 Hoover AL"]
[Date "2007.02.18"]
[Round "6"]
[White "Fidelity CC Sensory Voice, LV 4."]
[Black "Fidelity Mini Chess Challenge, LV 3."]
[Result "1/2-1/2"]
[ECO "C65"]
[WhiteElo "1394"]
[BlackElo "1300"]
[PlyCount "305"]
[EventDate "2007.02.01"]
[EventType "tourn (rapid)"]
[EventRounds "18"]
[EventCountry "USA"]
[EventCategory "6"]

1. e4 Nc6 2. Nf3 e5 3. Bb5 Nf6 4. Bxc6 dxc6 5. d3 Bg4 6. O-O Bc5 7. Nc3 O-O 8.
Bg5 Bxf3 9. Qxf3 Re8 10. Bxf6 Qxf6 11. Qxf6 gxf6 12. Rfe1 Rad8 13. h4 a6 14. a3
h5 15. Rad1 a5 16. Re2 Ra8 17. Red2 a4 18. Kf1 Red8 19. Ne2 Kg7 20. g3 Re8 21.
Nc3 b6 22. Kg2 Red8 23. Ne2 b5 24. c3 Re8 25. f4 Rad8 26. Kf3 exf4 27. gxf4 Ra8
28. d4 Bd6 29. Ng3 Rh8 30. Rd3 Rab8 31. R3d2 Rh7 32. Ke3 Ra8 33. Rd3 Kg6 34.
Nf5 Rhh8 35. Kf3 Rab8 36. R3d2 Rhg8 37. Ke3 Rbc8 38. Rd3 Rge8 39. Rg1+ Kh7 40.
Rg7+ Kh8 41. Rxf7 Rf8 42. Rxf8+ Rxf8 43. Rd1 Kh7 44. Kf3 Rg8 45. e5 fxe5 46.
dxe5 Bc5 47. Ke4 Kg6 48. Nd4 Bxd4 49. Rxd4 Re8 50. Rd7 Rc8 51. f5+ Kh6 52. Re7
c5 53. Kd5 c4 54. Kc6 Rf8 55. Re6+ Kg7 56. Kxb5 Rxf5 57. Kxa4 Kf7 58. Rc6 Rxe5
59. Rxc7+ Ke6 60. Rxc4 Re2 61. b4 Kf5

[fen]8/8/8/5k1p/KPR4P/P1P5/4r3/8 w - - 0 62[/fen]

62. b5 Re1 63. b6 Re7 64. Rd4 Ke5 65. Rb4
Rb7 66. Kb5 Kf5 67. Kc5 Ke5 68. Rb2 Ke4 69. Kb5 Rf7 70. Kc4 Rb7 71. Rb4 Ke5 72.
Kd3 Rd7+ 73. Rd4 Rb7 74. Re4+ Kf5 75. Rb4 Rd7+ 76. Ke3 Re7+ 77. Kf3 Rb7 78. Rb1
Ke5 79. Ke3 Ke6 80. Kd2 Rd7+ 81. Kc1 Rb7 82. Kd1 Kd6 83. Rb5 Kc6 84. Rxh5 Rxb6
85. Kc2 Kd7 86. c4 Ra6 87. Rd5+ Ke6 88. Rd3 Ra5 89. Rh3 Ke5 90. Kd2 Ra6 91. Rc3
Rh6 92. Rh3 Kd4 93. h5 Kxc4 94. a4 Rd6+ 95. Ke2 Ra6 96. Ra3 Kb4 97. Rh3 Rxa4
98. h6 Ra2+ 99. Kf3 Ra3+ 100. Kg4 Rxh3 101. Kxh3 Kc4 102. h7 Kd4 103. h8=Q+ Ke4
104. Qg8 Kf4 105. Qg3+ Kf5 106. Qc3 Kf4 107. Qc1+ Kf3 108. Qe1 Kf4 109. Qg1 Kf5
110. Qh1 Kf4 111. Qb1 Kf3 112. Qc1 Ke4 113. Qc3 Kf4 114. Qf6+ Ke4 115. Qf2 Kd5
116. Qe2 Kd4 117. Qe1 Kd5 118. Qf1 Ke5 119. Qg1 Kf5 120. Qh1 Kf4 121. Qb1 Kf3
122. Qd1+ Kf4 123. Qc1+ Kf3 124. Qe1 Kf4 125. Qb1 Kf3 126. Qc1 Ke4 127. Qg1 Kf4
128. Qc1+ Kf3 129. Qe1 Kf4 130. Qa1 Kf3 131. Qf6+ Ke4 132. Qa6 Kf4 133. Qa5 Kf3
134. Qa2 Kf4 135. Qe2 Kg5 136. Qa2 Kf4 137. Qe2 Kg5 138. Qf2 Kh5 139. Qc2 Kg5
140. Qh2 Kf5 141. Qg2 Kf4 142. Qd2+ Kf3 143. Qa2 Kf4 144. Qe2 Kg5 145. Qa2 Kf4
146. Qd2+ Kf3 147. Qb2 Kf4 148. Qf6+ Ke4 149. Qf1 Kd4 150. Qf3 Ke5 151. Qb3 Kf4
152. Qd3 Kg5 153. Qa3 1/2-1/2

Regards

[Steve B]

Hello Guys,

Well the Risc 1mb ore Risc 2, resign very "quick" in tournament games
When they come below a scrore of -500 they resign.

But yes , they mostly resign when mate is coming , ore when they see it

Kind regards,Hans

Hi Hans
i dont recall the Risc I and II resigning after a -5 Eval
you mention tournament games
do you mean 40/2 games ?

i usually only play 1 Min Avg Games and never 40/2
is this "premature resignation" bug noticed in all time controls or only 40/2?

Short Fuse Regards
Steve

[Sargon1972]

Hello Steve,

Yes i do mean 40/120 games

Well it resigns when it is lost when i play 30 sec game with it but that is a long time ago now, so i do not know for sure in every level this will hapen, and i think not it,s a bug ore so

Kr,Hans

[Theo]

Hello Steve,

Yes i do mean 40/120 games

Well it resigns when it is lost when i play 30 sec game with it but that is a long time ago now, so i do not know for sure in every level this will hapen, and i think not it,s a bug ore so

Kr,Hans

Guys,

this is not a bug, this happens with any normal level. Risc and Polgar and Milano resign if

evaluation drops below -9.99

or

if evaluation drops below -4.99 AND its queen is exchanged already.

They resign only once. There is a text message displayed after the move, and it is supported by a acoustic signal.

KR,
Theo

[Monsieur Plastique]

Hi Jon

basically i agree with you however i have one exception
there are a variety of computers(mostly the high end ones) that have a "Resign" feature
sometimes this is nothing more then the computer announcing a mate against itself
some however will resign when the eval hits -9.99

if a computer will resign of its own accord i will end the game there even if mate is not imminent
i consider the Resign feature part and parcel of the Program and if a computer chooses to quit..then keeping the game going is not correct

IMHO Regards
Steve

Hi Steve,

I should clarify that I 100% agree with you in these situations. If a a machine resigns, no problem - that is a loss - regardless of the merits of the position.