Bugs in SzG Swiss - Dorky, Nejmet, Djinn

[Gábor Szõts]

Geschrieben von: / Posted by: Gábor Szõts at 16 March 2004 07:34:56:

Hi,
The Betsy-Dorky game indicated an unusual bug of Dorky (9.-Kd7), probably related to pondering (pondering was on at the beginning of the game because of my mistake).
[Event "SzG Swiss 3rd division 7th season"]
[Site "Szentendre"]
[Date "2004.03.12"]
[Round "7"]
[White "Betsy_651"]
[Black "Dorky_348"]
[Result "1-0"]
[TimeControl "2700+15"]
1. d4 Nf6 2. Bg5 Ne4 3. h4 d5 4. Nd2 Bf5 5. e3 h6 6. Bf4 e6 7. Nxe4 dxe4 8.
g4 Bh7 9. c3 Kd7 10. d5 Kc8 11. dxe6 Qxd1+ 12. Rxd1 fxe6 13. Bb5 Be7 14.
Ne2 Rd8 15. Nd4 Bg8 16. Bc4 e5 17. Bxg8 Rxg8 18. Bxe5 Nd7 19. Bf4 Nc5 20.
Nf5 Bf6 21. g5 hxg5 22. hxg5 Bd8 23. Rh7 Ne6 24. Rd5 b6 25. Ng3 Nf8 26. Rh3
Ng6 27. Rh7 Nf8 28. Rh2 Ne6 29. Nxe4 Nxf4 30. exf4 c6 31. Rf5 a6 32. Rf7
Re8 33. f3 Re7 34. Rxe7 Bxe7 35. Rh8+ Kb7 36. Rh7 Bf8 37. f5 Rd8 38. f6 Rd7
39. g6 Kc7 40. Rh8 Rd8 41. Ng5 Re8+ 42. Kf1 gxf6 43. Nh7 Kd6 44. Nxf8 Re7
45. Rh7 b5 46. b4
{Black resigns} 1-0

It seems to me that Nejmet's time usage, at least with incremental clock, is broken under WinBoard. Generally it uses only 20-30 seconds pro move, while in UCI mode under Arena this is around 1 minute. I believe this translates into a 100 Élõ decrease in strength. No wonder that the program cannot make it to the 2nd division. In the future I will use Nejmet only under Arena, until this bug is fixed.

However, perhaps this was the most unfortunate game I have ever seen. Look at Djinn's 42nd move, practically burying his own king.
[Event "SzG Swiss 3rd division 7th season"]
[Site "Szentendre"]
[Date "2004.03.13"]
[Round "8"]
[White "Djinn 0.848"]
[Black "Queen_242a"]
[Result "0-1"]
[TimeControl "2700+15"]
1. e4 c5 2. Nf3 e6 3. d4 cxd4 4. Nxd4 Nf6 5. Nc3 Bb4 6. e5 Nd5 7. Bd2 Bxc3
8. bxc3 d6 9. Qg4 g6 10. Bb5+ Bd7 11. exd6 Bxb5 12. Nxb5 a6 13. Qd4 Rg8 14.
Nc7+ Nxc7 15. dxc7 Qxc7 16. O-O Nd7 17. Rad1 f6 18. Rfe1 e5 19. Qb4 O-O-O
20. Be3 Rge8 21. Rd6 h5 22. Rb1 Rf8 23. a4 g5 24. a5 Rfe8 25. c4 h4 26. c5
h3 27. c6 bxc6 28. Ba7 Nf8 29. Rxf6 Ne6 30. Rf7 Rd7 31. Qb8+ Qxb8 32. Rxb8+
Kc7 33. Rxe8 Rxf7 34. Rxe6 g4 35. Rxe5 Kd6 36. Bb8+ Kd7 37. Rh5 hxg2 38. c4
Rf3 39. Rh7+ Ke6 40. Rc7 Kd6 41. Rf7+ Ke6 42. Rxf3 gxf3 43. Bf4 Kf5 44. Be3
Kg4 45. Bd4 Kg5 46. Be5 Kg4 47. Bf6 Kf5 48. Bc3 Ke4 49. c5 Kf5 50. Bb4 Ke4
51. Bd2 Kf5 52. Bc3 Ke4 53. Bg7 Kf4 54. Bb2 Kg5 55. Bd4 Kg6 56. h4 Kf5 57.
Bc3 Kg6 58. Be5 Kf5 59. Bg7 Ke4 60. Kh2 Kf5 61. Bc3 Kg6 62. Kg1 Kf5 63. Bd4
Kg6 64. Kh2 Kf5 65. Bg7 Kg4 66. Bf6 Kf5 67. Bg5 Kg6 68. Bd8 Kh5 69. Bf6 Kg6
70. Bg5 Kf5 71. Bd8 Kg6 72. Kg1 Kh6 73. Be7 Kg6 74. Bg5 Kh5 75. Bf6 Kg6 76.
Be7 Kh5 77. Bg5 Kg6 78. Bf4 Kh5 79. Bg3 Kg4 80. Kh2 Kh5 81. Be5 Kxh4 82.
Bg3+ Kg5 83. Kg1 Kf5 84. Bd6 Ke4 85. Bc7 Kd4 86. Bf4 Kxc5 87. Kh2 Kc4 88.
Bc7 Kb5 89. Bf4 c5 90. Kg1 c4 91. Bc7 Ka4 92. Be5 Kb3 93. Bd4 Kc2 94. Kh2
c3 95. Kg1 Kd2 96. Be3+
{White resigns} 0-1
Gábor

[Tom Likens]

Geschrieben von: / Posted by: Tom Likens at 16 March 2004 13:31:24:
Als Antwort auf: / In reply to: Bugs in SzG Swiss - Dorky, Nejmet, Djinn geschrieben von: / posted by: Gábor Szõts at 16 March 2004 07:34:56:

Hello Gábor,
Thanks for posting this game. Lately, I've been revamping the engine's
search, but between this game and the games posted recently by
Dr. Wael Deeb I think it's long overdue that I start revamping the
king safety code and the piece entombment code. This is *not* the
way I want the engine to become famous.
regards,
--tom

However, perhaps this was the most unfortunate game I have ever
seen. Look at Djinn's 42nd move, practically burying his own king.

Gábor

[Tom Likens]

Geschrieben von: / Posted by: Tom Likens at 16 March 2004 18:39:32:
Als Antwort auf:/In reply to: Bugs in SzG Swiss - Dorky, Nejmet, Djinn geschrieben von: / posted by: Gábor Szõts at 16 March 2004 07:34:56:

However, perhaps this was the most unfortunate game I have ever seen. Look at Djinn's 42nd move, practically burying his own king.
[Event "SzG Swiss 3rd division 7th season"]
[Site "Szentendre"]
[Date "2004.03.13"]
[Round "8"]
[White "Djinn 0.848"]
[Black "Queen_242a"]
[Result "0-1"]
[TimeControl "2700+15"]
1. e4 c5 2. Nf3 e6 3. d4 cxd4 4. Nxd4 Nf6 5. Nc3 Bb4 6. e5 Nd5 7. Bd2 Bxc3
8. bxc3 d6 9. Qg4 g6 10. Bb5+ Bd7 11. exd6 Bxb5 12. Nxb5 a6 13. Qd4 Rg8 14.
Nc7+ Nxc7 15. dxc7 Qxc7 16. O-O Nd7 17. Rad1 f6 18. Rfe1 e5 19. Qb4 O-O-O
20. Be3 Rge8 21. Rd6 h5 22. Rb1 Rf8 23. a4 g5 24. a5 Rfe8 25. c4 h4 26. c5
h3 27. c6 bxc6 28. Ba7 Nf8 29. Rxf6 Ne6 30. Rf7 Rd7 31. Qb8+ Qxb8 32. Rxb8+
Kc7 33. Rxe8 Rxf7 34. Rxe6 g4 35. Rxe5 Kd6 36. Bb8+ Kd7 37. Rh5 hxg2 38. c4
Rf3 39. Rh7+ Ke6 40. Rc7 Kd6 41. Rf7+ Ke6 42. Rxf3 gxf3 43. Bf4 Kf5 44. Be3
Kg4 45. Bd4 Kg5 46. Be5 Kg4 47. Bf6 Kf5 48. Bc3 Ke4 49. c5 Kf5 50. Bb4 Ke4
51. Bd2 Kf5 52. Bc3 Ke4 53. Bg7 Kf4 54. Bb2 Kg5 55. Bd4 Kg6 56. h4 Kf5 57.
Bc3 Kg6 58. Be5 Kf5 59. Bg7 Ke4 60. Kh2 Kf5 61. Bc3 Kg6 62. Kg1 Kf5 63. Bd4
Kg6 64. Kh2 Kf5 65. Bg7 Kg4 66. Bf6 Kf5 67. Bg5 Kg6 68. Bd8 Kh5 69. Bf6 Kg6
70. Bg5 Kf5 71. Bd8 Kg6 72. Kg1 Kh6 73. Be7 Kg6 74. Bg5 Kh5 75. Bf6 Kg6 76.
Be7 Kh5 77. Bg5 Kg6 78. Bf4 Kh5 79. Bg3 Kg4 80. Kh2 Kh5 81. Be5 Kxh4 82.
Bg3+ Kg5 83. Kg1 Kf5 84. Bd6 Ke4 85. Bc7 Kd4 86. Bf4 Kxc5 87. Kh2 Kc4 88.
Bc7 Kb5 89. Bf4 c5 90. Kg1 c4 91. Bc7 Ka4 92. Be5 Kb3 93. Bd4 Kc2 94. Kh2
c3 95. Kg1 Kd2 96. Be3+
{White resigns} 0-1
Gábor

OK, I played over this at lunch and it was *extremely* painful. Initially, I was quite pleased with
Djinn's performance early on, since it won a piece and seemed to be headed to a well-
deserved victory. But 41. Rf7+ preparing to trade the rooks and move 42. Rxf3 entombing the
king were howlers. I'm at work without access to the program, but this *has* to be a result of
the "tradedown when ahead in material" code kicking in without regard to the resultant
position. IMHO trading the rooks was a good thing, but only if the white king gets involved
in the game afterwards *and* only if it doesn't result in a protected passed pawn for black.
Still, overall (until the postional blunder) the game wasn't all that bad. And it is games like
this that highlight holes in the program's understanding so it's not all gloom and doom.
regards,
--tom
P.S. Gábor keep 'em coming!!

[Gábor Szõts]

Geschrieben von: / Posted by: Gábor Szõts at 17 March 2004 07:34:54:
Als Antwort auf:/In reply to: A couple of comments on the Djinn game geschrieben von: / posted by: Tom Likens at 16 March 2004 18:39:32:

OK, I played over this at lunch and it was *extremely* painful. Initially, I was quite pleased with
Djinn's performance early on, since it won a piece and seemed to be headed to a well-
deserved victory. But 41. Rf7+ preparing to trade the rooks and move 42. Rxf3 entombing the
king were howlers. I'm at work without access to the program, but this *has* to be a result of
the "tradedown when ahead in material" code kicking in without regard to the resultant
position. IMHO trading the rooks was a good thing, but only if the white king gets involved
in the game afterwards *and* only if it doesn't result in a protected passed pawn for black.
Still, overall (until the postional blunder) the game wasn't all that bad. And it is games like
this that highlight holes in the program's understanding so it's not all gloom and doom.
regards,
--tom
P.S. Gábor keep 'em coming!!

I have the bad habit of pinpointing bugs and blunders without mentioning the bright side. I was watching this game 'live' and I enjoyed it very much until the bitter end. Games like this are really a taste of joy for the spectator. Thank you, Tom.
Gábor

[Tom Likens]

Geschrieben von: / Posted by: Tom Likens at 17 March 2004 14:40:06:
Als Antwort auf:/In reply to: Re: A couple of comments on the Djinn game geschrieben von: / posted by: Gábor Szõts at 17 March 2004 07:34:54:

OK, I played over this at lunch and it was *extremely* painful. Initially, I was quite pleased with
Djinn's performance early on, since it won a piece and seemed to be headed to a well-
deserved victory. But 41. Rf7+ preparing to trade the rooks and move 42. Rxf3 entombing the
king were howlers. I'm at work without access to the program, but this *has* to be a result of
the "tradedown when ahead in material" code kicking in without regard to the resultant
position. IMHO trading the rooks was a good thing, but only if the white king gets involved
in the game afterwards *and* only if it doesn't result in a protected passed pawn for black.
Still, overall (until the postional blunder) the game wasn't all that bad. And it is games like
this that highlight holes in the program's understanding so it's not all gloom and doom.
regards,
--tom
P.S. Gábor keep 'em coming!!

I have the bad habit of pinpointing bugs and blunders without mentioning the bright side. I was watching this game 'live' and I enjoyed it very much until the bitter end. Games like this are really a taste of joy for the spectator. Thank you, Tom.
Gábor

As are emails, such as this!!
--tom