Bestia 0.90 vs Ufim 3.02 (3) 120'/40 (1-0) Commented gam

[Kurt Utzinger]

Geschrieben von: / Posted by: Kurt Utzinger at 07 March 2003 22:15:41:

And now the third game, this time the very lucky winner was Bestia 0.90. The match stands now 2-1 in favour of Bestia 0.90
Bestia 0.90 vs Ufim 3.02 (3) 1-0
Position before 50...Bf6? [correct was 50...Rf1]
 abcdefgh  8         8  7          7  6         6  5       5  4     4  3       3  2         2  1         1  abcdefgh ChessDiag V1.01 (12-OCT-2002)6k1/8/4p3/p3p1b1/P1NpP1p1/1P1R2P1/5r2/3K4 b - - 0 50
[Event "Bestia090_Ufim302 P3 650/32 120'/40"]
[Site "Wetzikon"]
[Date "2003.03.05"]
[Round "3"]
[White "Bestia 0.90"]
[Black "Ufim 3.02"]
[Result "1-0"]
[ECO "B54"]
[PlyCount "135"]
[EventDate "2003.??.??"]
1. e4 {Comments: Kurt Utzinger, Switzerland, engine/engine match under
Winboard 4.2.5, ponder=off, both engines with owns books, played on P3 650/32
MB hash at time control 120'/40} 1... c5 2. Nf3 d6 3. d4 cxd4 {
Black last book move} 4. Nxd4 {White last book move} 4... e5 5. Nf3 Nc6 6. Bc4
Be7 7. Nc3 Be6 {A surprising but not bad solution. Ufim302 is ready to get a
double pawn but this pawn controls the most important centre square d5. This
type of pawn structure is seldom a real weakness.} 8. Bxe6 fxe6 9. Be3 Nf6 10.
Bg5 O-O 11. O-O {I think that the advantage lies on Black's side who has
better perspectives in view of the half-open c- and f-files. Or in other words:
it is much easier for Black to find a plan.} 11... Rc8 {
An interesting idea would be 11...Qe8 aiming at g6/h5} 12. Qd3 h6 13. Bxf6 Rxf6
14. Rac1 {Hard to understand Bestia's intention behind this rather passive
rook maneourvre. One would rather expect 14.Rad1 to put pressure on the only
weak point in Black's position.} 14... Rf8 {And this time a somewhat mystery
move on the part of Ufim302. I see no reason for retreating the f6-rook to the
last rank. Black has two options: a) to play on the queens wing with 14...a6
later to be followed by b7-b5, Nc6-a5-c4 or b) to seek some attacking chances
on the kings wing with 14...Qe8 and the idea to bring the queen to g6 or h5
followed - depending on the circumstances - by g7-g5. There are of course
other moves like 14...Qb6 but the ideas demonstrated in a) and b) are the ones
that follow a positional sound plan.} 15. Qb5 {Not a good idea and only losing
some tempi which are useful for Black to expand on the queens side} ({
Shredder 7.04:} 15. Rfd1 a6 16. Nb1 Qb6 17. Nbd2 Qc5 18. Qc4 Qxc4 19. Nxc4 Nd4
20. Nxd4 exd4 21. Nd2 {-0.23/13}) 15... Na5 16. Rfd1 a6 17. Qd3 Nc4 {Black has
succeeded in improving his position and has now the clear better game because
White is damned for the defence.} 18. Rb1 b5 {A strong alternative was 18...Qa5
} 19. a4 b4 20. Ne2 {A critical situation: Black has the advantage but it is
by no means easy to find the best move. Only long analysis could bring a
satisfactory result in evaluating the different options like 20...Qc7, 20...
Qb6, 20...Qa5.} 20... Qc7 21. b3 Na3 {
(?!) I would hesitate to hide the good standing Nc4 at a3.} 22. Rb2 a5 23. Rc1
Qb6 24. c3 Bg5 {(!) An interesting idea. Black is driving white's rook away
from the c-file. White cannot play 25.Nxg5 on account of 25...Qxf2 with clear
advantage for Black.} 25. Rf1 bxc3 26. Nxc3 Bf6 {Are moves like 26...Bf4!? or
26...Be7!? stronger? It's hard to say. The text move protects the e5-pawn and
may sometime allow d6-d5.} 27. Nb1 Qc5 {
The line 27...Nxb1 28.Rbxb1 Rc5 29.Rfc8 Rfc8 30.Rxc5 Rxc5 looks more promising.
} 28. Nxa3 {
The variation 28.b4!? axb4 29.Nxa3 bxa3 30.Ra2 was an equal alternative.} 28...
Qxa3 {White is slightly under pressure and should try to exchange pieces and
for this reason 29.Rc2 would be the normal move for me. But Bestia 0.90 did
not believe me.} 29. Rbb1 Qb4 30. Qa6 Qc5 {A more dynamic but not stronger
solution is 30...Qxe4 31.Qxd6 Qf5 32.Qb6 e4 33.Nd4 Bxd4 34.Qxd4 Rb8=} 31. Nd2
Rc6 32. Qb5 d5 {Black does not allow the white knight to go to c4} 33. Qxc5 {
Unclear is 33.b4 Qxb5 34.axb5 Rc2 35.bxa5!? Rxd2 36.b6 Be7} 33... Rxc5 34. Rbc1
Rfc8 35. Rxc5 Rxc5 {The position is at least =+} 36. Rd1 Rc3 37. Kf1 g5 {
(?!) What is this move for? Black should activate his king by the centralizing
37...Kf7} 38. Ke2 g4 {
Now the locigal continuation with the idea to prepare Bf6-g5} 39. Nf1 d4 40.
Rd3 Rc2+ 41. Rd2 Rc1 42. Rd1 Rc2+ 43. Rd2 Rc7 44. Rd3 h5 45. Nd2 Rc2 46. h3 Bg5
47. hxg4 hxg4 48. Kd1 Ra2 49. Nc4 Rxf2 50. g3 Bf6 {(?) What a pity: After 50...
Rf1 51.Ke2 Rc1! 52.Nxa5 Rc2+ the game could have ended with perpetual check.
The text move leads to an almost winning advantage for White due to the strong
passed pawns on the queens wing.} 51. Nxa5 Rf1+ 52. Ke2 Rh1 53. Nc4 Kf8 54. Rd2
Bg5 55. Rb2 Ke7 56. b4 {The pawns on their march to a8/b8} 56... Rh3 57. Rb3
Bf6 {Also after 57...Bf4 58.a5 Bxg3 59.a6 can Black not save the game.} 58. a5
Rh2+ 59. Kf1 Rh1+ 60. Kg2 Rc1 61. Na3 Rc7 62. b5 {
Black is helpless against the two White free pawns on the queens wing.} 62...
Rd7 63. b6 d3 64. Rb1 Bg5 65. b7 Bc1 66. Nc4 d2 67. Ne3 Rd8 68. a6 {
Black resigns} 1-0

[Dimitris Poulos]

Geschrieben von: / Posted by: Dimitris Poulos at 07 March 2003 23:24:28:
Als Antwort auf: / In reply to: Bestia 0.90 vs Ufim 3.02 (3) 120'/40 (1-0) Commented game geschrieben von: / posted by: Kurt Utzinger at 07 March 2003 22:15:41:

Strange, Ruffian, Crafty19.03, List504 also sugest 50...Bf6

[Kurt Utzinger]

Geschrieben von: / Posted by: Kurt Utzinger at 07 March 2003 23:51:07:
Als Antwort auf: / In reply to: Re: Bestia 0.90 vs Ufim 3.02 (3) 120'/40 (1-0) Commented game geschrieben von: / posted by: Dimitris Poulos at 07 March 2003 23:24:28:

Strange, Ruffian, Crafty19.03, List504 also sugest 50...Bf6

Below the analysis of Ruffian 1.0.1 on P4 1.8/32 MB hash
Neue Partie
6k1/8/4p3/p3p1b1/P1NpP1p1/1P1R2P1/5r2/3K4 b - - 0 1
Analysis by Ruffian 1.0.1:
1...Ld8 2.Sxe5
µ (-1.04) Tiefe: 1/2 00:00:00
1...Tf1+ 2.Ke2 Tf7
-+ (-1.52) Tiefe: 1/5 00:00:00
= (-0.03) Tiefe: 2/6 00:00:00
1...Ld8 2.Sxe5 Tf1+ 3.Ke2
µ (-1.04) Tiefe: 2/6 00:00:00
³ (-0.31) Tiefe: 3/6 00:00:00
1...Kf8 2.Sxe5 Tf1+ 3.Ke2
³ (-0.50) Tiefe: 3/6 00:00:00
1...Lf6 2.Ke1 Tc2 3.Sxa5
µ (-0.71) Tiefe: 3/6 00:00:00
1...Tf3 2.Ke2 Ld8 3.Sxe5
µ (-0.73) Tiefe: 3/6 00:00:00
³ (-0.30) Tiefe: 4/8 00:00:00
1...Kf8 2.Sxe5 Le3 3.Sxg4
³ (-0.34) Tiefe: 4/8 00:00:00
1...Th2 2.Sxa5 Th1+ 3.Kc2 Th2+ 4.Kd1
µ (-0.74) Tiefe: 4/8 00:00:00
1...Tf1+ 2.Ke2 Th1 3.Sxe5 Th2+ 4.Ke1 Th1+ 5.Ke2
µ (-0.75) Tiefe: 4/11 00:00:00
³ (-0.40) Tiefe: 5/11 00:00:00
1...Lf6 2.Sxa5 Kf7 3.Sc6 Ke8 4.b4
³ (-0.54) Tiefe: 5/11 00:00:00
³ (-0.28) Tiefe: 6/11 00:00:00 25kN
1...Tg2 2.Sxe5 Le3 3.Sxg4 Txg3 4.Sxe3 dxe3 5.Ke2
³ (-0.44) Tiefe: 6/12 00:00:00 49kN
= (0.00) Tiefe: 7/14 00:00:00 80kN
1...Lf6 2.Sxa5 Tf1+ 3.Ke2 Tc1 4.Sc4 Kf7 5.a5 Ke8 6.Td1 Tc2+ 7.Td2 Tc1 8.Tb2 Ke7 9.Sd2 Tc3 10.a6 Txg3 11.a7 Te3+ 12.Kf2
³ (-0.30) Tiefe: 7/14 00:00:00 86kN
= (0.08) Tiefe: 18/37 00:03:38 161548kN
1...Tf1+ 2.Ke2 Tc1 3.Sxe5 Tc2+ 4.Kf1 Tc1+ 5.Ke2
= (0.00) Tiefe: 18/37 00:04:23 196076kN
= (0.00) Tiefe: 20/37 00:07:33 339117kN
(Utzinger, MyTown 07.03.2003)

[Kurt Utzinger]

Geschrieben von: / Posted by: Kurt Utzinger at 08 March 2003 00:18:22:
Als Antwort auf: / In reply to: Re: Bestia 0.90 vs Ufim 3.02 (3) 120'/40 (1-0) Commented game geschrieben von: / posted by: Dimitris Poulos at 07 March 2003 23:24:28:

Strange, Ruffian, Crafty19.03, List504 also sugest 50...Bf6

50...Rf1 after a certain thinking time.
Kurt

[Dimitris Poulos]

Geschrieben von: / Posted by: Dimitris Poulos at 08 March 2003 12:08:26:
Als Antwort auf: / In reply to: At least Ruffian 1.0.1 finds ... geschrieben von: / posted by: Kurt Utzinger at 08 March 2003 00:18:22:

Strange, Ruffian, Crafty19.03, List504 also sugest 50...Bf6

50...Rf1 after a certain thinking time.
Kurt

I didn't wait it to go that deep. Even Ruffian seems to ignore the possitional advantage of the cooperation of the two passed pawns.

[Kurt Utzinger]

Geschrieben von: / Posted by: Kurt Utzinger at 08 March 2003 13:58:18:
Als Antwort auf: / In reply to: Re: At least Ruffian 1.0.1 finds ... geschrieben von: / posted by: Dimitris Poulos at 08 March 2003 12:08:26:

Strange, Ruffian, Crafty19.03, List504 also sugest 50...Bf6

50...Rf1 after a certain thinking time.
Kurt

I didn't wait it to go that deep. Even Ruffian seems to ignore the possitional advantage of the cooperation of the two passed pawns.

Yes, you are right.
Kurt

[Uri Blass]

Geschrieben von: / Posted by: Uri Blass at 08 March 2003 15:03:54:
Als Antwort auf: / In reply to: Re: At least Ruffian 1.0.1 finds ... geschrieben von: / posted by: Dimitris Poulos at 08 March 2003 12:08:26:

Strange, Ruffian, Crafty19.03, List504 also sugest 50...Bf6

50...Rf1 after a certain thinking time.
Kurt

I didn't wait it to go that deep. Even Ruffian seems to ignore the possitional advantage of the cooperation of the two passed pawns.

I do not think that Ruffian ignore the passed pawns but black has also a protected passed pawn(d4) when white has weak pawns(e4 and g3)
I can add that black has a rook in the 7th rank when the black king is free to go to a better position.
I think that Ruffian simply evaluates black advantage as bigger than white advantage and I think that in a slightly different position a move like Bf6 may be winning.
If you criticize the positional understanding of chess programs you need to define exactly what do you think that they should evaluate.
giving bigger bonus for passed pawns is dangerous because there are cases when passed pawns are not decisive.
Uri

[Dimitris Poulos]

Geschrieben von: / Posted by: Dimitris Poulos at 08 March 2003 15:38:58:
Als Antwort auf: / In reply to: Re: At least Ruffian 1.0.1 finds ... geschrieben von: / posted by: Uri Blass at 08 March 2003 15:03:54:

If you criticize the positional understanding of chess programs you need to define exactly what do you think that they should evaluate.
giving bigger bonus for passed pawns is dangerous because there are cases when passed pawns are not decisive.
Uri

What Ruffian underestimates is that white has not one but two passed pawns that can cooperate. They are strong enough to take the game. Other engines seem to understand better the possition.

[Uri Blass]

Geschrieben von: / Posted by: Uri Blass at 08 March 2003 15:51:27:
Als Antwort auf: / In reply to: Re: At least Ruffian 1.0.1 finds ... geschrieben von: / posted by: Dimitris Poulos at 08 March 2003 15:38:58:

If you criticize the positional understanding of chess programs you need to define exactly what do you think that they should evaluate.
giving bigger bonus for passed pawns is dangerous because there are cases when passed pawns are not decisive.
Uri

What Ruffian underestimates is that white has not one but two passed pawns that can cooperate. They are strong enough to take the game. Other engines seem to understand better the possition.

other engines?
I checked only movei that gives black an advantage but you reported that crafty and List also have problems so I assumed that other engines do not understand the position better.
The opposite.
Ruffian at least can find it after search.
There are cases when 2 connected passed pawn cannot cooperate so saying that white wins because of the fact that white has 2 connected passed pawns does not seem convincing.
I suspect that if you change slightly the position of the black king
Rf1+ will not be the right move and keeping the rook in the 7th rank and going forward with the king may be better.
You need exact definitions to decide when passed pawns are good.
I doubt if Crafty List and Ruffian have less knowledge then the engines that you tried and it is possible that their knowledge is counter productive in other positions.
Uri

[Dimitris Poulos]

Geschrieben von: / Posted by: Dimitris Poulos at 08 March 2003 17:28:11:
Als Antwort auf: / In reply to: Re: At least Ruffian 1.0.1 finds ... geschrieben von: / posted by: Uri Blass at 08 March 2003 15:51:27:

At least Yace, Comet and LG find the solution and seem to recognise a small advantage for white after Bf6. I don't know if they have extra knowledge about connected passed pawns that helps them better understand the possition. Or any knowledge about the safety of those pawns that arises from the position. Or any knowledge of their mobilability. At least they find the best move.
I doubt if extra knowledge where exists is a dissadvantage for other possitions. As long as it is well implemented.

[Uri Blass]

Geschrieben von: / Posted by: Uri Blass at 08 March 2003 17:38:26:
Als Antwort auf: / In reply to: Re: At least Ruffian 1.0.1 finds ... geschrieben von: / posted by: Dimitris Poulos at 08 March 2003 17:28:11:

At least Yace, Comet and LG find the solution and seem to recognise a small advantage for white after Bf6. I don't know if they have extra knowledge about connected passed pawns that helps them better understand the possition. Or any knowledge about the safety of those pawns that arises from the position. Or any knowledge of their mobilability. At least they find the best move.
I doubt if extra knowledge where exists is a dissadvantage for other possitions. As long as it is well implemented.

The question is if it is extra knowledge or different weights for positional factors.
different weights may be productive in one position and counter productive in another position.
Uri

[Dimitris Poulos]

Geschrieben von: / Posted by: Dimitris Poulos at 08 March 2003 18:26:49:
Als Antwort auf: / In reply to: Re: At least Ruffian 1.0.1 finds ... geschrieben von: / posted by: Uri Blass at 08 March 2003 17:38:26:

The question is if it is extra knowledge or different weights for positional factors.
different weights may be productive in one position and counter productive in another position.
Uri

Well I thought of extra knowledge. But even changing weights can be productive.
Some days ago I posted a mail that proposed that we check the possition, before evaluation, for passed pawns, mobility, king safety etc. and make little changes to weights according to the special needs of each possition. For example if king safety is diminuished then use a weight greater than 1 for it.
Perhaps this way the gameplay becomes more stable and we andicipate complications.