Draw recognition problems

[Lyapko George]

Geschrieben von: / Posted by: Lyapko George at 16 June 2000 11:06:12:
Many middle strength engines does not recognize 50 moves rule and or 3-fold repetitions. After all their games you must check PGN-files to award a draw in some cases. That is my opinion. Here are 2 games which I've played yesterday. In the first, Averno recognizes 3-fold repetition against Knightx in absolutely hopeless position. In the second, Chop mates Colchess, but there was 50 moves rule (why Chop didn't mate 30-40 moves before?), so I must award a draw.
Am I wright?
Best regards,
George
[Event "Computer chess game"]
[Site "LYAPKO"]
[Date "2000.06.15"]
[Round "1"]
[White "Knightx_153"]
[Black "Averno_029"]
[Result "1/2-1/2"]
[TimeControl "40/300"]
1. d4 d6 2. e4 Nf6 3. Nc3 g6 4. Nf3 Bg7 5. Be2 O-O 6. O-O Bg4 7. Be3 Kh8 8.
Ng5 Bxe2 9. Qxe2 h6 10. Nf3 Kh7 11. Qb5 Qc8 12. e5 Ng4 13. Bf4 Nc6 14. exd6
cxd6 15. Rad1 e5 16. dxe5 dxe5 17. Bg3 f5 18. Rd6 e4 19. Nd2 e3 20. fxe3
Nxe3 21. Rc1 f4 22. Bf2 Bxc3 23. bxc3 Qc7 24. Ne4 Qe7 25. Bh4 Qxe4 26.
Qxb7+ Kh8 27. Qxc6 Qxc6 28. Rxc6 Kg7 29. Rc7+ Rf7 30. Rxf7+ Kxf7 31. Rb1
Nxc2 32. Rb7+ Ke6 33. Rh7 h5 34. Bg5 Kf5 35. h4 Ke4 36. Rg7 a5 37. Rxg6 Rc8
38. Re6+ Kf5 39. Rf6+ Ke5 40. Rh6 Kf5 41. Rxh5 Kg4 42. Rh7 Rxc3 43. Rg7 Rd3
44. Bd8+ Kh5 45. Rg5+ Kh6 46. Bxa5 Nd4 47. Bb4 Rg3 48. Rxg3 fxg3 49. a4 Kh5
50. Be7 Nc6 51. Bf6 Kg4 52. Bg5 Na5 53. Kf1 Nc6 54. Ke2 Nd4+ 55. Kd3 Nc6
56. Kc4 Ne5+ 57. Kd5 Nd3 58. a5 Kf5 59. Bd2 Ne5 60. a6 Nd7 61. a7 Nb6+ 62.
Kc6 Na8 63. h5 Ke5 64. Kb7 Ke4 65. h6 Kd3 66. Bf4 Ke2 67. Bxg3 Kf1 68. h7
Kxg2 69. Bd6 Kf3 70. Kxa8 Ke4 71. Kb8 Kf5 72. h8=Q Ke6 73. a8=Q Kxd6 74.
Qh5 Ke6 75. Qc5 Kf7 76. Kb7 Ke6 77. Kb8 Kf7 78. Kb7 Ke6 79. Kb8
{Draw by repetition} 1/2-1/2
[Event "Computer chess game"]
[Site "LYAPKO"]
[Date "2000.06.15"]
[Round "4"]
[White "Colchess_623"]
[Black "Chop_710"]
[Result "1/2-1/2"]
[TimeControl "40/300"]
1. e4 Nf6 2. e5 Nd5 3. d4 d6 4. Nf3 Bg4 5. h3 Bxf3 6. Qxf3 dxe5 7. dxe5 Nc6
8. Bb5 Qd7 9. Nc3 e6 10. Nxd5 Qxd5 11. Qxd5 exd5 12. Be3 a6 13. Bxc6+ bxc6
14. O-O-O Kd7 15. f4 Rb8 16. Rd3 Rb4 17. Rhd1 Be7 18. b3 Re4 19. g3 Rb8 20.
Ba7 Rb7 21. Bd4 c5 22. Bb2 c6 23. c4 d4 24. Kd2 Bd8 25. Ba3 Ba5+ 26. Kc2
Re2+ 27. Kb1 Bb4 28. Bb2 Ke6 29. h4 f6 30. exf6 Kxf6 31. h5 Kf5 32. Rg1 Rh2
33. Rdd1 Kg4 34. Rd3 Kxh5 35. g4+ Kh4 36. g5 Re7 37. a3 Bd2 38. Bc1 Bxc1
39. Kxc1 Ra2 40. Kb1 Rf2 41. Rh1+ Kg4 42. Rdh3 Kxf4 43. Rh4+ Kxg5 44. Rxh7
Kf6 45. Rh8 Kf7 46. Rb8 Re3 47. Rb6 Ke7 48. Ka1 Kd6 49. Rxa6 Ree2 50. Rg1
d3 51. Rd1 Ra2+ 52. Kb1 Rfb2+ 53. Kc1 Rxb3 54. Rxd3+ Rxd3 55. Ra8 Raxa3 56.
Rg8 Rdc3+ 57. Kd2 Rg3 58. Ke1 Rg2 59. Kf1 Rc2 60. Rd8+ Kc7 61. Rd1 Rac3 62.
Ra1 Rc1+ 63. Rxc1 Rxc1+ 64. Ke2 Rxc4 65. Kd3 Rd4+ 66. Ke3 g5 67. Kf3 c4 68.
Ke3 c5 69. Kf3 c3 70. Ke2 c2 71. Ke3 c1=Q+ 72. Kf3 Rf4+ 73. Ke2 Qc3 74. Kd1
Rf3 75. Ke2 Qd3+ 76. Ke1 Qb1+ 77. Kd2 Qf1 78. Kc2 Qa1 79. Kd2 Qb1 80. Ke2
Qf1+ 81. Kd2 Qf2+ 82. Kd1 Rd3+ 83. Kc1 Qd2+ 84. Kb1 Qe1+ 85. Kc2 Qf1 86.
Kb2 Qe2+ 87. Kc1 Qd1+ 88. Kb2 Rb3+ 89. Ka2 Rc3 90. Kb2 Qc1+ 91. Ka2 Qd2+
92. Kb1 Qc2+ 93. Ka1 Qe2 94. Kb1 Rc4 95. Ka1 Qd1+ 96. Kb2 Qd2+ 97. Ka3 Qb4+
98. Ka2 Rh4 99. Ka1 Qc3+ 100. Kb1 Rh1+ 101. Ka2 Rh2+ 102. Kb1 Rh5 103. Ka2
Qc4+ 104. Kb1 Rh1+ 105. Kb2 Rc1 106. Ka3 Qb4+ 107. Ka2 Qa4+ 108. Kb2 Rc2+
109. Kb1 Qb3+ 110. Ka1 Qe3 111. Kb1 Rh2 112. Ka1 Qa3+ 113. Kb1 Rh1+ 114.
Kc2 Re1 115. Kd2 Qe3+ 116. Kc2 Re2+ 117. Kd1 Qd3+ 118. Kc1 Rc2+ 119. Kb1
Rh2+ 120. Kc1 Qd2+ 121. Kb1 Qa2+ 122. Kc1 g4 123. Kd1 Rh1#
{Black mates,but draw awarded:50 moves after 121...Qa2+} 1/2-1/2

[Gábor Szõts]

Geschrieben von: / Posted by: Gábor Szõts at 16 June 2000 12:29:40:
Als Antwort auf: / As an answer to: Draw recognition problems geschrieben von: / posted by: Lyapko George at 16 June 2000 11:06:12:

Many middle strength engines does not recognize 50 moves rule and or 3-fold repetitions. After all their games you must check PGN-files to award a draw in some cases. That is my opinion. Here are 2 games which I've played yesterday. In the first, Averno recognizes 3-fold repetition against Knightx in absolutely hopeless position. In the second, Chop mates Colchess, but there was 50 moves rule (why Chop didn't mate 30-40 moves before?), so I must award a draw.
Am I wright?
Best regards,
George

Hello,
According to FIDE laws, draw on 3-fold repetition or 50-move rule must be claimed. If the player doesn't claim, he has to continue the game. So in my opinion you don't have the right to adjust the score. You are the arbiter, not the player.
There is, however, a case where you can (and must) adjust the score. The FIDE laws say:
9.6. The game is drawn when a position is reached from which
a checkmate cannot occur by any possible series of legal
moves, even with the most unskilled play. This immediately
ends the game.
That is, there is no win on time in this case.
Regards,
Gábor

[Colin Frayn]

Geschrieben von: / Posted by: Colin Frayn at 16 June 2000 12:55:31:
Als Antwort auf: / As an answer to: Draw recognition problems geschrieben von: / posted by: Lyapko George at 16 June 2000 11:06:12:

Many middle strength engines does not recognize 50 moves rule and or 3-fold repetitions. After all their games you must check PGN-files to award a draw in some cases. That is my opinion. Here are 2 games which I've played yesterday. In the first, Averno recognizes 3-fold repetition against Knightx in absolutely hopeless position. In the second, Chop mates Colchess, but there was 50 moves rule (why Chop didn't mate 30-40 moves before?), so I must award a draw.

You must strictly claim for draws by repetition & 50 moves. ColChess doesn't support the 50 move rule yet, though I'll probably add it soon. Just one of those things that doesn't come up very often with top level programs. It does support all other draws though, and claims them too.
Cheers,
Colin

ColChess Homepage

[Lyapko George]

Geschrieben von: / Posted by: Lyapko George at 16 June 2000 13:39:07:
Als Antwort auf: / As an answer to: Re: Draw recognition problems geschrieben von: / posted by: Gábor Szõts at 16 June 2000 12:29:40:

Hello,
According to FIDE laws, draw on 3-fold repetition or 50-move rule must be claimed. If the player doesn't claim, he has to continue the game. So in my opinion you don't have the right to adjust the score. You are the arbiter, not the player.
There is, however, a case where you can (and must) adjust the score. The FIDE laws say:
9.6. The game is drawn when a position is reached from which
a checkmate cannot occur by any possible series of legal
moves, even with the most unskilled play. This immediately
ends the game.
That is, there is no win on time in this case.
Regards,
Gábor

When I play at MMEICS and make three repetitions, the game is immediately draw. I don't need to claim it...
NB,about insufficient material: Crafty recently claims a draw in KBkn position, but there exists a position, where black mates (eg. KB6/8/kn/8/8/8/8/8).So, Crafty is wrong?
Szia,
Gyorgy

[Inmann Werner]

Geschrieben von: / Posted by: Inmann Werner at 16 June 2000 13:58:48:
Als Antwort auf: / As an answer to: Re: Draw recognition problems geschrieben von: / posted by: Gábor Szõts at 16 June 2000 12:29:40:

Many middle strength engines does not recognize 50 moves rule and or 3-fold repetitions. After all their games you must check PGN-files to award a draw in some cases. That is my opinion. Here are 2 games which I've played yesterday. In the first, Averno recognizes 3-fold repetition against Knightx in absolutely hopeless position. In the second, Chop mates Colchess, but there was 50 moves rule (why Chop didn't mate 30-40 moves before?), so I must award a draw.
Am I wright?
Best regards,
George

Hello,
According to FIDE laws, draw on 3-fold repetition or 50-move rule must be claimed. If the player doesn't claim, he has to continue the game. So in my opinion you don't have the right to adjust the score. You are the arbiter, not the player.
There is, however, a case where you can (and must) adjust the score. The FIDE laws say:
9.6. The game is drawn when a position is reached from which
a checkmate cannot occur by any possible series of legal
moves, even with the most unskilled play. This immediately
ends the game.
That is, there is no win on time in this case.
Regards,
Gábor

I am not of your opinion.
For me, the GUI has to detect this rules and claim the draws. In Chessbase GUI, the GUI claims it, in my own GUI (privatware) also the GUI claims it. The Engine only has to think about the move and return a best move.
IMHO, also checkmate has to be recognized by the GUI.
Werner

[Gábor Szõts]

Geschrieben von: / Posted by: Gábor Szõts at 16 June 2000 14:02:24:
Als Antwort auf: / As an answer to: Re: Draw recognition problems geschrieben von: / posted by: Lyapko George at 16 June 2000 13:39:07:

When I play at MMEICS and make three repetitions, the game is immediately draw. I don't need to claim it...
NB,about insufficient material: Crafty recently claims a draw in KBkn position, but there exists a position, where black mates (eg. KB6/8/kn/8/8/8/8/8).So, Crafty is wrong?

It is the deficiency of the server, or you don't comply with FIDE laws.
If we take the FIDE law literally, KBkn is not a draw, since the bishop side can 'mate itself'. But it would be an unnecessary wasting of time to continue in such a position, since the claim was based upon tablebases, so the claimant won't make such a fatal error, nor will it exceed time.
Szia
Gábor

[Volker Pittlik]

Geschrieben von: / Posted by: Volker Pittlik at 16 June 2000 14:09:27:
Als Antwort auf: / As an answer to: Re: Draw recognition problems geschrieben von: / posted by: Lyapko George at 16 June 2000 13:39:07:
[...]

NB,about insufficient material: Crafty recently claims a draw in KBkn position, but there exists a position, where black mates (eg. KB6/8/kn/8/8/8/8/8).So, Crafty is wrong?

Interesting position indeed. If you set the black knight to c4 Crafty finds the mate immediatly:
8272 _first : edit
8272 _first : #
8272 _first : Ka8
8272 _first : Bb8
8272 _first : c
8272 _first : Ka6
8272 _first : Nc4
8272 _first : .
11657 _first : analyze
11937 _first : Analyze Mode: type "exit" to terminate.
11997 _first : 1 -32766 0 7 1. ... Nb6#
11997 _first : 1 -32766 0 12 1. ... Nb6#
11997 _first : 2 -32766 0 13 1. ... Nb6# _EGTB_
12007 _first : 2 -32766 0 23 1. ... Nb6# _EGTB_
12007 _first : 3 -32766 0 24 1. ... Nb6# _HT_
12007 _first : 3 -32766 0 34 1. ... Nb6# _HT_
12007 _first : 4 -32766 0 35 1. ... Nb6# _HT_
12007 _first : 4 -32766 0 45 1. ... Nb6# _HT_
13940 _first : .
13940 _first : .
13940 _first : stat01: 0 0 0 0 0
13940 _first : 1 -32766 0 7 1. ... Nb6# _HT_
13940 _first : 1 -32766 0 12 1. ... Nb6# _HT_
13940 _first : 2 -32766 0 13 1. ... Nb6# _HT_
13940 _first : 2 -32766 0 23 1. ... Nb6# _HT_
13940 _first : 3 -32766 0 24 1. ... Nb6# _HT_
13950 _first : 3 -32766 0 34 1. ... Nb6# _HT_
13950 _first : 4 -32766 0 35 1. ... Nb6# _HT_
13950 _first : 4 -32766 0 45 1. ... Nb6# _HT_
[...]
18677 _first : move c4b6
18767 _first : 0-1 {Black mates}
18767 _first : result 0-1 {Black mates}
18767 _first : force
But when you play the move yourself Crafty claims a draw:
32647 _first : edit
32647 _first : #
32647 _first : Ka8
32647 _first : Bb8
32647 _first : c
32647 _first : Ka6
32647 _first : Nc4
32647 _first : .
36813 _first : c4b6
36903 _first : game is a draw due to insufficient material.
36903 _first : 1/2-1/2 {Insufficient material}
42391 _first : quit
Congrats! I guess you found an error in Crafty.
Volker

[Gábor Szõts]

Geschrieben von: / Posted by: Gábor Szõts at 16 June 2000 14:25:05:
Als Antwort auf: / As an answer to: Re: Draw recognition problems geschrieben von: / posted by: Volker Pittlik at 16 June 2000 14:09:27:

[Gábor Szõts]

Geschrieben von: / Posted by: Gábor Szõts at 16 June 2000 14:47:34:
Als Antwort auf: / As an answer to: Re: Draw recognition problems geschrieben von: / posted by: Inmann Werner at 16 June 2000 13:58:48:

I am not of your opinion.
For me, the GUI has to detect this rules and claim the draws. In Chessbase GUI, the GUI claims it, in my own GUI (privatware) also the GUI claims it. The Engine only has to think about the move and return a best move.
IMHO, also checkmate has to be recognized by the GUI.
Werner

I don't dispute your approach. We could apply different rules for engines and humans. But, as far as I know, no such rules have been agreed upon. And then we are left with the FIDE laws, whether we like them or not.
Look, WinBoard is also a GUI, still it checks neither draw rules, nor the validity of draw claims. Its philosophy is that it is the job of the engines.
For me, either method is acceptable.
Gruß
Gábor
P.S. And what if we play without a GUI? Who will then check draws?

[Inmann Werner]

Geschrieben von: / Posted by: Inmann Werner at 16 June 2000 14:58:42:
Als Antwort auf: / As an answer to: Re: Draw recognition problems geschrieben von: / posted by: Gábor Szõts at 16 June 2000 14:47:34:

I am not of your opinion.
For me, the GUI has to detect this rules and claim the draws. In Chessbase GUI, the GUI claims it, in my own GUI (privatware) also the GUI claims it. The Engine only has to think about the move and return a best move.
IMHO, also checkmate has to be recognized by the GUI.
Werner

I don't dispute your approach. We could apply different rules for engines and >humans. But, as far as I know, no such rules have been agreed upon. And then we >are left with the FIDE laws, whether we like them or not.
Look, WinBoard is also a GUI, still it checks neither draw rules, nor the >validity of draw claims. Its philosophy is that it is the job of the engines.
For me, either method is acceptable.
Gruß
Gábor
P.S. And what if we play without a GUI?
Who will then check draws?

I agree.
But always a combination between Engine and GUI plays (an engine alone can not really play. It can be a "simple GUI" as in Crafty or whatever.
If you make a WCC, I agree full to you. But playing with engines under different GUIs, to only calculate the strength of engines, you have to make your own choices....
not my philosophy, who cares...
That is not possible (IMHO).
Maybe we speak of different things, if we speak of a GUI.
Werner

[Gábor Szõts]

Geschrieben von: / Posted by: Gábor Szõts at 16 June 2000 15:24:35:
Als Antwort auf: / As an answer to: Re: Draw recognition problems geschrieben von: / posted by: Inmann Werner at 16 June 2000 14:58:42:

P.S. And what if we play without a GUI?

That is not possible (IMHO).
Maybe we speak of different things, if we speak of a GUI.

Yes, this term (GUI) is rather nebulous for me.
Good-bye
Gábor

[Thomas Mayer]

Geschrieben von: / Posted by: Thomas Mayer at 16 June 2000 16:32:26:
Als Antwort auf: / As an answer to: Re: Draw recognition problems geschrieben von: / posted by: Inmann Werner at 16 June 2000 14:58:42:
Hi Werner

I am not of your opinion.
For me, the GUI has to detect this rules and claim the draws. In Chessbase GUI, the GUI claims it, in my own GUI (privatware) also the GUI claims it. The Engine only has to think about the move and return a best move.
IMHO, also checkmate has to be recognized by the GUI.
Werner

I don't dispute your approach. We could apply different rules for engines and >humans. But, as far as I know, no such rules have been agreed upon. And then we >are left with the FIDE laws, whether we like them or not.

I agree.
But always a combination between Engine and GUI plays (an engine alone can not really play. It can be a "simple GUI" as in Crafty or whatever.
If you make a WCC, I agree full to you. But playing with engines under different GUIs, to only calculate the strength of engines, you have to make your own choices....

Well, for the engine it is important to know about 3-fold-repetition and 50-move-rule, my engine knows nothing about this stuff and stumbles so very often in totaly won positions in the 3-fold-repetition because in his short horizon can't find the way to win... --> Because of this I think most of the 2000+ - engines knows about that...
With some engines I have the problem that they don't send the right result-code to winboard, so winboard remarks a lost by time, sometimes for the winning engine... In this case we should adjust the result to the truth...
Perhaps it's an good deal to say when the engine show's that it has understand the position (eval 0.00 or something like this) but didn't protest, we should set the score to draw. If both engines knows nothing, we play as long as time goes... (The evals are in most case in the debug-files, I play my tourneys with debug, in analysis it's often better to understand the games when you see, what the programs think with this and that move... especially when you don't watch every game at night...)
By the way: If the engine is a combination of GUI and engine you are absolutely write, it's not the part of the engine to remark draw, mate...
Greets, Thomas
P.S.: Huh... Not always easy to impress your thoughts in english, hopefully anyone understand what I have written...

[Inmann Werner]

Geschrieben von: / Posted by: Inmann Werner at 16 June 2000 23:22:13:
Als Antwort auf: / As an answer to: Re: Draw recognition problems geschrieben von: / posted by: Thomas Mayer at 16 June 2000 16:32:26:

Well, for the engine it is important to know about 3-fold-repetition and 50-move-rule, my engine knows nothing about this stuff and stumbles so very often in totaly won positions in the 3-fold-repetition because in his short horizon can't find the way to win... --> Because of this I think most of the 2000+ - engines knows about that...
With some engines I have the problem that they don't send the right result-code to winboard, so winboard remarks a lost by time, sometimes for the winning engine... In this case we should adjust the result to the truth...

I agree, that for an engine, especially the 3 fold repetition is more than important in search . I declare draw in search, if there is one repetition!
makes it fast and misses not often IMHO. (endless discussion..)
But when I write my engine, I do not know, if somebody uses my Dos GUI, Fritz, Winboard, Chess Assistant or whatelse. I do not know, if he wants to play a game, if he wants to analyze a position ....
Thats enough about this stuff....
Werner

[Dann Corbit]

Geschrieben von: / Posted by: Dann Corbit at 16 June 2000 23:58:29:
Als Antwort auf: / As an answer to: Re: Draw recognition problems geschrieben von: / posted by: Colin Frayn at 16 June 2000 12:55:31:

Many middle strength engines does not recognize 50 moves rule and or 3-fold repetitions. After all their games you must check PGN-files to award a draw in some cases. That is my opinion. Here are 2 games which I've played yesterday. In the first, Averno recognizes 3-fold repetition against Knightx in absolutely hopeless position. In the second, Chop mates Colchess, but there was 50 moves rule (why Chop didn't mate 30-40 moves before?), so I must award a draw.

You must strictly claim for draws by repetition & 50 moves. ColChess doesn't support the 50 move rule yet, though I'll probably add it soon. Just one of those things that doesn't come up very often with top level programs. It does support all other draws though, and claims them too.
Cheers,
Colin

[Dann Corbit]

Geschrieben von: / Posted by: Dann Corbit at 16 June 2000 23:59:57:
Als Antwort auf: / As an answer to: Re: Draw recognition problems geschrieben von: / posted by: Colin Frayn at 16 June 2000 12:55:31:

Many middle strength engines does not recognize 50 moves rule and or 3-fold repetitions. After all their games you must check PGN-files to award a draw in some cases. That is my opinion. Here are 2 games which I've played yesterday. In the first, Averno recognizes 3-fold repetition against Knightx in absolutely hopeless position. In the second, Chop mates Colchess, but there was 50 moves rule (why Chop didn't mate 30-40 moves before?), so I must award a draw.

You must strictly claim for draws by repetition & 50 moves. ColChess doesn't support the 50 move rule yet, though I'll probably add it soon. Just one of those things that doesn't come up very often with top level programs. It does support all other draws though, and claims them too.

5 out of 7000+ in the current calibration runs. There could possibly be more, if the reason for the draw was not written.
all.pgn ( 4255): 106. Ke4 Ra1 107. Rc3 Rb1 108. Kf5 Ra1 109. Ke5 Rb1 { Draw by 50 moves rule
all.pgn ( 5242): Ra6 Ne5 128. Ra3+ Kd2 { 50 move rule } 1/2-1/2
all.pgn ( 5533): 111. Rb5 Ke4 112. Rb6 Kf3 113. Rb3+ { Draw by 50 moves rule } 1/2-1/2
all.pgn ( 7405): Kh1 Ke3 129. Kh2 Kf4 130. Kh1 Ke4 { Draw by 50 moves rule } 1/2-1/2
all.pgn ( 9112): Qd7 155. Qb4 Qe8 156. Kf4 Kg8 157. Qc4 Kh8 { Draw by 50 moves rule }


My FTP site