Main
Date: 08 Sep 2006 23:39:16
From: Dave (from the UK)
Subject: Was sacrifice sound? Engines think not.
I've been having a bit of a tough time on ICC lately, loosing virtually
all my games, although I won the one below as black.

I made a sacrifice at 24...Rxd4, loosing a rook, but winning a pawn and
knight. I thought I'd be better in the endgame with the extra pawn - I
has slightly more material before this exchange, but after it I had just
one extra pawn.

I've analaysed this game with two chess engines:

'crafty' version 20.14
'scidlet' 3.6.1 (part of the open-source scid database)

Scores from 'crafty' and 'scidlet' *before* Black's 25th move are:

score = -3.21, depth 17. (crafty)
score = -3.44, depth 16. (scidlet)

So crafty and scidlet seem to agree reasonably well about score before
my move and they both think 25...e5 should have been played.

But I did not play 25...e5, but instead 25...Rxd4. The next two moves
were almost forced (26.exd4 Rxd4)

Now the engines seem to think this was a bad move. I know it was a loss
of material, but I personally think it was justified.

score = -1.41, depth 18. (crafty)
score = -1.26, depth 14. (scidlet)

So both think the games gets worst for black by around 2, but I fail to
see how it can be. Comments?

Here's the PGN file.

[Event "ICC 10 12"]
[Site "Internet Chess Club"]
[Date "2006.09.08"]
[Round "-"]
[White "mgm"]
[Black "g8wrb"]
[Result "0-1"]
[ICCResult "White resigns"]
[WhiteElo "1247"]
[BlackElo "1116"]
[Opening "Benko's opening"]
[ECO "A00"]
[NIC "VO.09"]
[Time "16:26:27"]
[GameType "ICCStandard"]
[TimeControl "600+12"]

1. g3 g6 2. Bg2 Bg7 3. e3 Nf6 4. Ne2 O-O 5. d4 d5 6. Nd2 Bf5 7. c4 c6 8. b3
Qa5 9. O-O c5 10. cxd5 cxd4 11. Nxd4 Nxd5 12. Bxd5 Qxd5 13. Qf3 Rd8 14. Qxd5
Rxd5 15. Bb2 Nc6 16. N2f3 Nxd4 17. Bxd4 Be4 18. Nd2 Bd3 19. Bxg7 Bxf1 20.
Nxf1 Kxg7 21. h4 h5 22. Nh2 Rad8 23. Kg2 R8d7 24. Nf3 Kf6 25. Nd4 Rxd4 26.
exd4 Rxd4 27. Rc1 Rd7 28. Rc4 Ke5 29. Ra4 a6 30. f4+ Kf5 31. Kh3 f6 32. Ra5+
Ke4 33. Ra4+ Rd4 34. Ra5 Rd5 35. Ra4+ Rd4 36. Ra5 e5 37. fxe5 fxe5 38. Rc5
Rd7 39. a4 Re7 40. g4 Ke3 41. gxh5 gxh5 42. Rc3+ Kf4 43. Rc4+ Kf3 44. Rc3+
Kf4 45. Rg3 e4 46. Kg2 Rf7 47. Kf2 Re7 48. Re3 Kg4 49. Kg2 Kxh4 50. Rh3+ Kg4
51. Rg3+ Kf4 52. Kf2 Re5 53. Re3 h4 54. Ke2 Kg4 55. Rc3 Rf5 56. Re3 Rf4 57.
b4 h3 58. b5 h2 59. Rh3 Kxh3 {White resigns} 0-1


--
Dave (from the UK)

Please note my email address changes periodically to avoid spam.
It is always of the form: [email protected]
Hitting reply will work for a few months only - later set it manually.

http://witm.sourceforge.net/ (Web based Mathematica front end)




 
Date: 11 Sep 2006 13:29:12
From:
Subject: Re: Was sacrifice sound? Engines think not.
[email protected] wrote:
> In the case of your example position, clearly the best move is Qd4+. It
> REALLY "simplifies the win", because it wins the bishop AND keeps the
> rook. Also, your reference to Crafty's score needs some clarification.
> -32758 means a Mate in 6 for Black. After playing Rxd7 Kxd7, you would
> soon see Crafty say Mate in 8.

You make good points (on both counts). While both paths lead to a win
with best play, forcing the rook exchange with 25...e5 leads to a much
easier victory. I wish I had seen that move!



 
Date: 08 Sep 2006 18:21:20
From:
Subject: Re: Was sacrifice sound? Engines think not.

[email protected] wrote:

> In the case of your example position, clearly the best move is Qd4+.

Follow-up -- Qd4+ is actually a Mate in 5. Qd1+ accomplishes the same
thing, but allows the White King to move towards the center, making
mate take one move longer.

jm



 
Date: 08 Sep 2006 18:19:04
From:
Subject: Re: Was sacrifice sound? Engines think not.

[email protected] wrote:
> [email protected] wrote:
> > Chess Tiger (Palm version 15.1) sees the position after white's 25th as
> > in Black's favour. the following best line for Black its sees Black
> > ahead by a score of +2.6
> >
> > On black's 25th move it suggests 25...Rc7 (it looked long and hard at
> > e5 and also Rc5) 26.a4, Rc3, 27.Rb1,Rd3 28.b4, Rd2 29.Kf3
> >
> > After Black sacrifices the rook ChessTiger see's Black's advantage has
> > fallen (to now only 1.26). I have to say it was a gutsy sacrifice.
> > Black was lucky to hold White out.
>
> Computers aren't great judges of sacrifices. An example:
>
> FEN: 8/3B2r1/3K4/8/8/8/8/2k3q1 b - - 0 1
> Diagram: http://tinyurl.com/mqy28
>
> If Black sacrifices the rook Rxd7, Crafty sees Black's advantage fall
> from -32758 to -1005. Is this a gutsy sacrifice, or simplifying the win?

It could certainly be called the latter, but it really isn't either,
nor does it mean that computers can't judge MATERIAL sacrifices. It's
POSITIONAL sacrifices that computers have a hard time with.

In the case of your example position, clearly the best move is Qd4+. It
REALLY "simplifies the win", because it wins the bishop AND keeps the
rook. Also, your reference to Crafty's score needs some clarification.
-32758 means a Mate in 6 for Black. After playing Rxd7 Kxd7, you would
soon see Crafty say Mate in 8.

So Qd4+ is a Mate in 6 and Rxd7 is a Mate in 9.

jm



 
Date: 08 Sep 2006 18:06:08
From: [email protected]
Subject: Re: Was sacrifice sound? Engines think not.
[email protected] wrote:
> Chess Tiger (Palm version 15.1) sees the position after white's 25th as
> in Black's favour. the following best line for Black its sees Black
> ahead by a score of +2.6
>
> On black's 25th move it suggests 25...Rc7 (it looked long and hard at
> e5 and also Rc5) 26.a4, Rc3, 27.Rb1,Rd3 28.b4, Rd2 29.Kf3
>
> After Black sacrifices the rook ChessTiger see's Black's advantage has
> fallen (to now only 1.26). I have to say it was a gutsy sacrifice.
> Black was lucky to hold White out.

Computers aren't great judges of sacrifices. An example:

FEN: 8/3B2r1/3K4/8/8/8/8/2k3q1 b - - 0 1
Diagram: http://tinyurl.com/mqy28

If Black sacrifices the rook Rxd7, Crafty sees Black's advantage fall
from -32758 to -1005. Is this a gutsy sacrifice, or simplifying the win?



 
Date: 08 Sep 2006 17:12:06
From: [email protected]
Subject: Re: Was sacrifice sound? Engines think not.
Chess Tiger (Palm version 15.1) sees the position after white's 25th as
in Black's favour. the following best line for Black its sees Black
ahead by a score of +2.6

On black's 25th move it suggests 25...Rc7 (it looked long and hard at
e5 and also Rc5) 26.a4, Rc3, 27.Rb1,Rd3 28.b4, Rd2 29.Kf3

After Black sacrifices the rook ChessTiger see's Black's advantage has
fallen (to now only 1.26). I have to say it was a gutsy sacrifice.
Black was lucky to hold White out.
[email protected] wrote:
> Hi Dave,
>
> > I made a sacrifice at 24...Rxd4, loosing a rook, but winning a pawn and
> > knight. I thought I'd be better in the endgame with the extra pawn
>
> I believe this is a won endgame, so your sacrifice was sound. In
> general, pawns on both wings predicts a win. Don't be afraid of
> exchanging rooks, but at all costs maintain pawns on both wings!
>
> White later blundered with 48.Re3, and you win his h4-pawn for free.



 
Date: 08 Sep 2006 16:44:58
From:
Subject: Re: Was sacrifice sound? Engines think not.
Dave (from the UK) wrote:
> I've been having a bit of a tough time on ICC lately, loosing virtually
> all my games, although I won the one below as black.
>
> I made a sacrifice at 24...Rxd4, loosing a rook, but winning a pawn and
> knight. I thought I'd be better in the endgame with the extra pawn - I
> has slightly more material before this exchange, but after it I had just
> one extra pawn.
>
> I've analaysed this game with two chess engines:
>
> 'crafty' version 20.14
> 'scidlet' 3.6.1 (part of the open-source scid database)
>
> Scores from 'crafty' and 'scidlet' *before* Black's 25th move are:
>
> score = -3.21, depth 17. (crafty)
> score = -3.44, depth 16. (scidlet)
>
> So crafty and scidlet seem to agree reasonably well about score before
> my move and they both think 25...e5 should have been played.
>
> But I did not play 25...e5, but instead 25...Rxd4. The next two moves
> were almost forced (26.exd4 Rxd4)
>
> Now the engines seem to think this was a bad move. I know it was a loss
> of material, but I personally think it was justified.
>
> score = -1.41, depth 18. (crafty)
> score = -1.26, depth 14. (scidlet)
>
> So both think the games gets worst for black by around 2, but I fail to
> see how it can be. Comments?
>
> Here's the PGN file.
>
> [Event "ICC 10 12"]
> [Site "Internet Chess Club"]
> [Date "2006.09.08"]
> [Round "-"]
> [White "mgm"]
> [Black "g8wrb"]
> [Result "0-1"]
> [ICCResult "White resigns"]
> [WhiteElo "1247"]
> [BlackElo "1116"]
> [Opening "Benko's opening"]
> [ECO "A00"]
> [NIC "VO.09"]
> [Time "16:26:27"]
> [GameType "ICCStandard"]
> [TimeControl "600+12"]
>
> 1. g3 g6 2. Bg2 Bg7 3. e3 Nf6 4. Ne2 O-O 5. d4 d5 6. Nd2 Bf5 7. c4 c6 8. b3
> Qa5 9. O-O c5 10. cxd5 cxd4 11. Nxd4 Nxd5 12. Bxd5 Qxd5 13. Qf3 Rd8 14. Qxd5
> Rxd5 15. Bb2 Nc6 16. N2f3 Nxd4 17. Bxd4 Be4 18. Nd2 Bd3 19. Bxg7 Bxf1 20.
> Nxf1 Kxg7 21. h4 h5 22. Nh2 Rad8 23. Kg2 R8d7 24. Nf3 Kf6 25. Nd4 Rxd4 26.
> exd4 Rxd4 27. Rc1 Rd7 28. Rc4 Ke5 29. Ra4 a6 30. f4+ Kf5 31. Kh3 f6 32. Ra5+
> Ke4 33. Ra4+ Rd4 34. Ra5 Rd5 35. Ra4+ Rd4 36. Ra5 e5 37. fxe5 fxe5 38. Rc5
> Rd7 39. a4 Re7 40. g4 Ke3 41. gxh5 gxh5 42. Rc3+ Kf4 43. Rc4+ Kf3 44. Rc3+
> Kf4 45. Rg3 e4 46. Kg2 Rf7 47. Kf2 Re7 48. Re3 Kg4 49. Kg2 Kxh4 50. Rh3+ Kg4
> 51. Rg3+ Kf4 52. Kf2 Re5 53. Re3 h4 54. Ke2 Kg4 55. Rc3 Rf5 56. Re3 Rf4 57.
> b4 h3 58. b5 h2 59. Rh3 Kxh3 {White resigns} 0-1
>
>
> --
> Dave (from the UK)
>
> Please note my email address changes periodically to avoid spam.
> It is always of the form: [email protected]
> Hitting reply will work for a few months only - later set it manually.
>
> http://witm.sourceforge.net/ (Web based Mathematica front end)

Engines are (typically) materialistic -- even more so when there is a
lot of wood on the board. They can't see as far down the horizon as a
skilled player, or as somebody who understands basic rules, such as the
other poster's "pawns on both wings" theory.

Basically, with 25...e5, you force a trade of rooks when you are
already up two points in material. Instead of a 16-14 advantage it
becomes an 11-9 advantage. Much better odds.

Actually, if you move the game along several (forced) moves after
25...e5 (26.Nf3 Rd1 27.Rxd1 Rxd1) you will likely see the score jump
dramatically, perhaps even above 5.0. The less material on the board,
the farther an engine can see.

So, I think what can be said here is that 25...e5 is "the best move"
because it is the most likely to turn into a win. 25...Rxd4 is also a
"winning move" because it leads to a win with best play, although I
think the win will be harder to achieve, all else being equal.

jm



 
Date: 08 Sep 2006 23:45:45
From: Dave (from the UK)
Subject: Re: Was sacrifice sound? Engines think not.
The following was meant to go to rec.games.chess.analysis, not
rec.games.chess.computer. I'm reposting.

--------


I've been having a bit of a tough time on ICC lately, loosing virtually
all my games, although I won the one below as black.

I made a sacrifice at 24...Rxd4, loosing a rook, but winning a pawn and
knight. I thought I'd be better in the endgame with the extra pawn - I
has slightly more material before this exchange, but after it I had just
one extra pawn.

I've analaysed this game with two chess engines:

'crafty' version 20.14
'scidlet' 3.6.1 (part of the open-source scid database)

Scores from 'crafty' and 'scidlet' *before* Black's 25th move are:

score = -3.21, depth 17. (crafty)
score = -3.44, depth 16. (scidlet)

So crafty and scidlet seem to agree reasonably well about score before
my move and they both think 25...e5 should have been played.

But I did not play 25...e5, but instead 25...Rxd4. The next two moves
were almost forced (26.exd4 Rxd4)

Now the engines seem to think this was a bad move. I know it was a loss
of material, but I personally think it was justified.

score = -1.41, depth 18. (crafty)
score = -1.26, depth 14. (scidlet)

So both think the games gets worst for black by around 2, but I fail to
see how it can be. Comments?

Here's the PGN file.

[Event "ICC 10 12"]
[Site "Internet Chess Club"]
[Date "2006.09.08"]
[Round "-"]
[White "mgm"]
[Black "g8wrb"]
[Result "0-1"]
[ICCResult "White resigns"]
[WhiteElo "1247"]
[BlackElo "1116"]
[Opening "Benko's opening"]
[ECO "A00"]
[NIC "VO.09"]
[Time "16:26:27"]
[GameType "ICCStandard"]
[TimeControl "600+12"]

1. g3 g6 2. Bg2 Bg7 3. e3 Nf6 4. Ne2 O-O 5. d4 d5 6. Nd2 Bf5 7. c4 c6 8. b3
Qa5 9. O-O c5 10. cxd5 cxd4 11. Nxd4 Nxd5 12. Bxd5 Qxd5 13. Qf3 Rd8 14. Qxd5
Rxd5 15. Bb2 Nc6 16. N2f3 Nxd4 17. Bxd4 Be4 18. Nd2 Bd3 19. Bxg7 Bxf1 20.
Nxf1 Kxg7 21. h4 h5 22. Nh2 Rad8 23. Kg2 R8d7 24. Nf3 Kf6 25. Nd4 Rxd4 26.
exd4 Rxd4 27. Rc1 Rd7 28. Rc4 Ke5 29. Ra4 a6 30. f4+ Kf5 31. Kh3 f6 32. Ra5+
Ke4 33. Ra4+ Rd4 34. Ra5 Rd5 35. Ra4+ Rd4 36. Ra5 e5 37. fxe5 fxe5 38. Rc5
Rd7 39. a4 Re7 40. g4 Ke3 41. gxh5 gxh5 42. Rc3+ Kf4 43. Rc4+ Kf3 44. Rc3+
Kf4 45. Rg3 e4 46. Kg2 Rf7 47. Kf2 Re7 48. Re3 Kg4 49. Kg2 Kxh4 50. Rh3+ Kg4
51. Rg3+ Kf4 52. Kf2 Re5 53. Re3 h4 54. Ke2 Kg4 55. Rc3 Rf5 56. Re3 Rf4 57.
b4 h3 58. b5 h2 59. Rh3 Kxh3 {White resigns} 0-1




--
Dave (from the UK)

Please note my email address changes periodically to avoid spam.
It is always of the form: [email protected]
Hitting reply will work for a few months only - later set it manually.

http://witm.sourceforge.net/ (Web based Mathematica front end)


  
Date: 11 Sep 2006 22:30:26
From: Antonio Torrecillas
Subject: Re: Was sacrifice sound? Engines think not.
Hello

I think 24...Rxd4 is a mistake.

There is a simple win after 24...e5 25.Nf3 Rd1! and after exchanging the
rook white can not avoid more material loses. It's easy to check that!

24...Rxd4 lead to a difficult rook ending. Knowing that rook endings
(being a pawn up) are the most difficult endings after different color
bishop endings, I think you made the wrong exchange.

There are some books about what pieces to exchange. It's a difficult
matter in chess.

Antonio

En/na Dave (from the UK) ha escrit:
> The following was meant to go to rec.games.chess.analysis, not
> rec.games.chess.computer. I'm reposting.
>
> --------
>
> I've been having a bit of a tough time on ICC lately, loosing virtually
> all my games, although I won the one below as black.
>
> I made a sacrifice at 24...Rxd4, loosing a rook, but winning a pawn and
> knight. I thought I'd be better in the endgame with the extra pawn - I
> has slightly more material before this exchange, but after it I had just
> one extra pawn.
>
> I've analaysed this game with two chess engines:
>
> 'crafty' version 20.14
> 'scidlet' 3.6.1 (part of the open-source scid database)
>
> Scores from 'crafty' and 'scidlet' *before* Black's 25th move are:
>
> score = -3.21, depth 17. (crafty)
> score = -3.44, depth 16. (scidlet)
>
> So crafty and scidlet seem to agree reasonably well about score before
> my move and they both think 25...e5 should have been played.
>
> But I did not play 25...e5, but instead 25...Rxd4. The next two moves
> were almost forced (26.exd4 Rxd4)
>
> Now the engines seem to think this was a bad move. I know it was a loss
> of material, but I personally think it was justified.
>
> score = -1.41, depth 18. (crafty)
> score = -1.26, depth 14. (scidlet)
>
> So both think the games gets worst for black by around 2, but I fail to
> see how it can be. Comments?
>
> Here's the PGN file.
>
> [Event "ICC 10 12"]
> [Site "Internet Chess Club"]
> [Date "2006.09.08"]
> [Round "-"]
> [White "mgm"]
> [Black "g8wrb"]
> [Result "0-1"]
> [ICCResult "White resigns"]
> [WhiteElo "1247"]
> [BlackElo "1116"]
> [Opening "Benko's opening"]
> [ECO "A00"]
> [NIC "VO.09"]
> [Time "16:26:27"]
> [GameType "ICCStandard"]
> [TimeControl "600+12"]
>
> 1. g3 g6 2. Bg2 Bg7 3. e3 Nf6 4. Ne2 O-O 5. d4 d5 6. Nd2 Bf5 7. c4 c6 8. b3
> Qa5 9. O-O c5 10. cxd5 cxd4 11. Nxd4 Nxd5 12. Bxd5 Qxd5 13. Qf3 Rd8 14.
> Qxd5
> Rxd5 15. Bb2 Nc6 16. N2f3 Nxd4 17. Bxd4 Be4 18. Nd2 Bd3 19. Bxg7 Bxf1 20.
> Nxf1 Kxg7 21. h4 h5 22. Nh2 Rad8 23. Kg2 R8d7 24. Nf3 Kf6 25. Nd4 Rxd4 26.
> exd4 Rxd4 27. Rc1 Rd7 28. Rc4 Ke5 29. Ra4 a6 30. f4+ Kf5 31. Kh3 f6 32.
> Ra5+
> Ke4 33. Ra4+ Rd4 34. Ra5 Rd5 35. Ra4+ Rd4 36. Ra5 e5 37. fxe5 fxe5 38. Rc5
> Rd7 39. a4 Re7 40. g4 Ke3 41. gxh5 gxh5 42. Rc3+ Kf4 43. Rc4+ Kf3 44. Rc3+
> Kf4 45. Rg3 e4 46. Kg2 Rf7 47. Kf2 Re7 48. Re3 Kg4 49. Kg2 Kxh4 50. Rh3+
> Kg4
> 51. Rg3+ Kf4 52. Kf2 Re5 53. Re3 h4 54. Ke2 Kg4 55. Rc3 Rf5 56. Re3 Rf4 57.
> b4 h3 58. b5 h2 59. Rh3 Kxh3 {White resigns} 0-1



   
Date: 20 Sep 2006 13:46:19
From: David Richerby
Subject: Re: Was sacrifice sound? Engines think not.
Antonio Torrecillas <[email protected] > wrote:
> En/na Dave (from the UK) ha escrit:
>> 1.g3 g6 2.Bg2 Bg7 3.e3 Nf6 4.Ne2 O-O 5.d4 d5 6.Nd2 Bf5 7.c4 c6 8.b3
>> Qa5 9.O-O c5 10.cxd5 cxd4 11.Nxd4 Nxd5 12.Bxd5 Qxd5 13.Qf3 Rd8
>> 14.Qxd5 Rxd5 15.Bb2 Nc6 16.N2f3 Nxd4 17.Bxd4 Be4 18.Nd2 Bd3 19.Bxg7
>> Bxf1 20.Nxf1 Kxg7 21.h4 h5 22.Nh2 Rad8 23.Kg2 R8d7 24.Nf3 Kf6
>> 25.Nd4 Rxd4 26. exd4 Rxd4 {...} 0-1
>
> I think 24...Rxd4 is a mistake.
>
> 24...Rxd4 lead to a difficult rook ending. Knowing that rook endings
> (being a pawn up) are the most difficult endings after different color
> bishop endings, I think you made the wrong exchange.

``All rook and pawn endings are drawn.''
-- Savielly Tartakower

``Rook endgames a pawn up are drawn. Rook endgames a pawn down are
lost.''
-- Chris Walsh


Dave.

--
David Richerby Carnivorous Dangerous Dictator (TM):
www.chiark.greenend.org.uk/~davidr/ it's like a totalitarian leader but
it could explode at any minute and
it's full of teeth!