Not logged inRybka Chess Community Forum
Up Topic Rybka Support & Discussion / Rybka Discussion / Is rook pawn less important then other pawns?
- - By Uri Blass (*****) Date 2007-07-12 21:49
http://mywebpages.comcast.net/danheisman/Articles/evaluation_of_material_imbalance.htm

"I teach my students to regard the rook's pawn as a different piece type, a crippled pawn. Database statistics indicate that it is on average worth about 15% less than a normal pawn"

I do not know if it is correct but at least rybka does not believe it
I tested many positions when both cxb6 and axb6 were possible and there were no positional reason to prefer axb6 like a file for the rook.
I do not see a big difference in evaluation between axb6 and cxb6 and rybka in part of the cases prefer cxb6 that means having pawn in the a file inspite of the fact that this pawn is not passed pawn.

Here are some examples

New game - Rybka 2.3.2a 32-bit
3k4/ppppppp1/1Q6/8/8/8/PPPPPPP1/3K4 b - - 0 1


Analysis by Rybka 2.3.2a 32-bit :

1. =  (0.18): 1...c7xb6 2.c2-c4 Kd8-c7 3.Kd1-c2 Kc7-c6 4.Kc2-d3 b6-b5 5.c4xb5+ Kc6xb5 6.Kd3-d4 Kb5-c6 7.f2-f4 Kc6-d6 8.b2-b4
2. ²  (0.30): 1...a7xb6 2.e2-e4 c7-c6 3.Kd1-e2 Kd8-c7 4.a2-a4 Kc7-d6 5.d2-d4 g7-g5 6.Ke2-e3 Kd6-e6 7.c2-c4 c6-c5

(,  13.07.2007)

New game - Rybka 2.3.2a 32-bit
1nbk4/ppppppp1/1Q6/8/8/8/PPPPPPP1/1NBK4 b - - 0 1


Analysis by Rybka 2.3.2a 32-bit :

1. =  (0.20): 1...a7xb6 2.d2-d3 d7-d6 3.c2-c4 e7-e5 4.Nb1-c3 Bc8-e6 5.Bc1-e3 Nb8-c6 6.Kd1-d2 Kd8-d7 7.f2-f3
2. =  (0.23): 1...c7xb6 2.d2-d4 d7-d6 3.Nb1-c3 Bc8-d7 4.Bc1-e3 f7-f6 5.d4-d5 Nb8-a6 6.Kd1-d2 g7-g5 7.a2-a3

(,  13.07.2007)

New game - Rybka 2.3.2a 32-bit
1nbk1b2/ppppppp1/1Q6/8/8/8/PPPPPPP1/1NBK1B2 b - - 0 1


Analysis by Rybka 2.3.2a 32-bit :

1. =  (0.16): 1...c7xb6 2.d2-d4 d7-d5 3.Bc1-f4 Nb8-c6 4.e2-e3 Bc8-f5 5.Nb1-c3 e7-e6 6.Bf1-e2 Bf8-e7
2. =  (0.24): 1...a7xb6 2.Nb1-c3 Nb8-c6 3.e2-e3 d7-d6 4.d2-d4 e7-e6 5.Kd1-d2 d6-d5 6.Bf1-e2

(,  13.07.2007)
Parent - - By lkaufman (*****) Date 2007-07-12 22:26
Rybka believes that rook pawns are worth significantly less than the others on a full board, but are worth more than others on an empty board. I'm not sure that I agree with the second statement, but I have not (yet) attempted to see whether Rybka would play better without this code, which was there before I joined the team. Your examples are all endgame examples. Add some major pieces, and I think you'll see that Rybka starts to dislike rook pawns, as I do. This principle is a major part of the justification for the Benko Gambit, for example. A pawn that can only capture one way is clearly inferior, at least in the part of the game where square control matters.
Parent - By Felix Kling (Gold) Date 2007-07-12 23:52
Another reason is that you damage your pawn structure by taking with the e.g. f-pawn (if there's an e-pawn)... "pawn islands" as capablanca said.
in the endgame some things like distant passed pawns (is this translation correct? I only know the German expression "entfernter Freibauer") become more important, but in the positions above I'm not sure if taking with the c-pawn is the best move... I would still take the a-pawn
Parent - - By Uri Blass (*****) Date 2007-07-13 03:46 Edited 2007-07-13 03:54
I added more pieces
Rybka still does not prefer the a pawn (after enough analysis it starts to prefer not having the a pawn but not by a big margin
and certainly not something close to your 0.15 pawns difference)

Edit:Note that I know that the position is not legal but I did not want to put the queen at d8 because in that case you will have more mobility for the queen by c7xb6

New game - Rybka 2.3.2a 32-bit
1nb1kbqr/pppppppp/1N6/8/8/8/PPPPPPPP/1NB1KBQR b Kk - 0 1


Analysis by Rybka 2.3.2a 32-bit :

1. =  (0.07): 1...c7xb6 2.d2-d3 Nb8-c6 3.c2-c4 d7-d6 4.Nb1-c3 g7-g6 5.g2-g3
2. =  (0.07): 1...a7xb6 2.c2-c4 Nb8-c6 3.Nb1-c3 e7-e6 4.Nc3-b5 Ke8-d8 5.g2-g4

(,  13.07.2007)

New game - Rybka 2.3.2a 32-bit
1nb1kbqr/pppppppp/1N6/8/8/8/PPPPPPPP/1NB1KBQR b Kk - 0 1


Analysis by Rybka 2.3.2a 32-bit :

1. =  (0.04): 1...a7xb6 2.Nb1-c3 Nb8-c6 3.e2-e3 d7-d6 4.Bf1-c4 Ke8-d8 5.Qg1-f1 Bc8-f5 6.Bc4-d3 Bf5-e6 7.b2-b3 g7-g5
2. =  (0.06): 1...c7xb6 2.Nb1-c3 Nb8-c6 3.e2-e3 d7-d6 4.Bf1-d3 g7-g6 5.Nc3-d5 Bf8-g7 6.Qg1-f1 e7-e6

(,  13.07.2007)
Parent - By lkaufman (*****) Date 2007-07-13 05:05
     The reason for this behavior is very complicated. Basically, Black wants to capture with the "c" pawn so that she can develop her knight to c6 and use the "a" pawn to keep the White knight from setting up permanent residence on b5. I can tell this sort of thing just by inspection; I don't even need to use Rybka for this. I haven't tried this, but I'll bet that if you remove the knights from b1 and b8, so this won't be an issue, you will see a clear preference for taking with the "a" pawn.
     I can suggest that you use very short searches, like 1 or 2 ply, for tests like this. When you use deep searches, all sorts of complications cloud the picture.
     Since you reminded me about this issue, I ran a test of 70,000 games (!) in which I set the deep endgame value of the rook's pawn to be identical to the knight's pawn, instead of slightly higher. The result was a virtual tie.
Parent - - By Gr00vy (**) Date 2007-07-13 01:44
These positions are not legal.   I thought that all illegal positions are simply ignored for testing purposes.
Parent - - By Roland Rösler (****) Date 2007-07-13 01:58
I´m sure, that position one is legal and this is a real endgame (seven pawn bothsides). It may be unlikely, but this doesn´t matter. Larry´s answer is dubious.
Parent - By lkaufman (*****) Date 2007-07-13 05:07
My answer was correct, but I can see why you would think otherwise. See above explanation.
Parent - - By turbojuice1122 (Gold) Date 2007-07-13 03:49
The only illegal position is the last one.  The others are simply "improbable".  You can get them, for example, by having the knights hopping all over the board, picking off the necessary pieces, and then getting captured by the king.  The missing pawns allow the king to move out of the queen's way.  In the third position, of course, the queen has no way of getting out because she's ultimately blocked by the bishops.
Parent - - By Gr00vy (**) Date 2007-07-13 04:54 Edited 2007-07-13 04:57
Yes, I was completely wrong.  And the last is improbable as well.   The white h pawn could have promoted to Queen as well.
Parent - - By SR (****) Date 2007-07-13 08:34
??????  And how did the queen get to g1? And how could the white h-pawn have queened when its still on h2?
Parent - By Gr00vy (**) Date 2007-07-13 19:17
I was, and I believe the replier was referring to the first three positions.
Parent - By RFK (Gold) Date 2007-07-13 21:49
Hi Soren Riis

Well, here in America we're taught not to question the magic just trust the proffered answer even if it does boggle the mind.
Parent - By turbojuice1122 (Gold) Date 2007-07-13 12:39
Oh, if you're referring to the last position that Uri put up in a later post, that one is definitely impossible--no way for the queen to get to g2 because the other pieces, i.e. king and bishop, cannot move off the first rank.  However, I don't really think that affects how Rybka evaluates positions--there isn't anything in the evaluation that relies on the position being legal--and it's a perfectly legal position in Fischer-Random chess.
Parent - - By Gaмßito (****) Date 2007-07-13 05:37
In these king+pawns endgames, to take with the rook pawn is a bit better.  Why to break and divide the black pawns chain with cxb6? 

It creates a very small weakness in the black position, that could be exploited by White more easily if Black does not play a good endgame. But Rybka cannot notice this, being based only in their calculations 
 
Regards, 
Gambito. 
Parent - By Hetman (*****) Date 2007-07-13 19:09
Hi,

the old great master Wilhelm Steinitz has told that 'a' pawn is the weakest one in the middlegame and the strongest one in endgame. Might be Rybka has read 'Modern Chess Instructor' and knows that.:-).
Regards
Hetman
Parent - By Vasik Rajlich (Silver) Date 2007-07-14 07:03
To add to Larry's comments - randomizer results have also very clearly shown that the rook pawns are the smallest of the pawn handicaps in the starting position.

Vas
Up Topic Rybka Support & Discussion / Rybka Discussion / Is rook pawn less important then other pawns?

Powered by mwForum 2.27.4 © 1999-2012 Markus Wichitill