18 Dec 09
1 edit
Originally posted by trev33if it's from the starting position the side with the pieces should have a decisive advantage, as rooks only come into play later. Besides, he'll also have the two bishops.
start of the game - do you do it?
if you go by points (which i usually do) 5 for the rook, 1 for the pawn and 3 each for the knight and bishop. equal. but is it?
18 Dec 09
2 edits
Originally posted by AThousandYoungI agree with this, I would take that work very seriously. That article is text-book material for computer chess history either. It's the basis for Rybka's evaluation (some parts of it at least, and in a very crude and simple way), and I bet for some others too. He later announced that he made slight changes on it, I don't remember if he publicized an updated version or not.
He did a very thorough statistical analysis. This impresses me. That's a way to find patterns you might miss other ways.
18 Dec 09
2 edits
Originally posted by trev33The text doesn't say anything against that. It says such battery would be great, but usually there's not enough time to pile up your major pieces like that.
i stopped reading it after that, my best attacks come when there's two (sometimes 3) major pieces on an open file.
It says that the reason his statistical analysis resulted in an advantage towards the pieces may be that in majority of the positions, there are not enough open files to be taken advantage of both rooks at the same time, so one of the rooks' function is limited.
20 Dec 09
If we are just talking about relative strength only, then exchange your two minor pieces for a rook + two pawns for break even. A rook and a pawn is hardly break even.
According to standard relative strength comparison between pieces, two minor pieces shall be equal to rook and a pawn. Why is that so? Let me try to reason. The strengths of two minor pieces are almost well distributed. Both pieces independent. On the other hand, the strengths of a rook and a pawn are vastly uneven. The pawn is defenseless without the support from the rook. Thus, the strength of the rook reduces since it has to play additional task to protect the pawn.
20 Dec 09
Originally posted by Bahariin the endgame, R+p can win, there's winning potential left.
If we are just talking about relative strength only, then exchange your two minor pieces for a rook + two pawns for break even. A rook and a pawn is hardly break even.
According to standard relative strength comparison between pieces, two minor pieces shall be equal to rook and a pawn. Why is that so? Let me try to reason. The strengths of two minor pi ...[text shortened]... Thus, the strength of the rook reduces since it has to play additional task to protect the pawn.
with B+N, even winning against the lone king is nontrivial. give the defending king anything, let alone R+p, and it's quickly almost impossible to mate him.
the numerical value of pieces has little to do with the question.
21 Dec 09
1 edit
Originally posted by wormwoodThe basic principle is we have to make the exchange decision now. We are not completely sure about how the game will progress says 20 moves from now. There are billions of possible positions. If we know them then we can use the knowledge to come up with the best decision. In this case we may want to give up a queen for a pawn and still winning. Of course this scenario is not the scenario that demands relative strenght to be known since there is no element of uncertainty.
in the endgame, R+p can win, there's winning potential left.
with B+N, even winning against the lone king is nontrivial. give the defending king anything, let alone R+p, and it's quickly almost impossible to mate him.
the numerical value of pieces has little to do with the question.
The relative strenght is important to be used if there exists the element of uncertainty in future. We decide using balance of probability. The most important thing about our decision is it shall be correct at least during the time it is made. We can only hope our decision is correct in future too.
21 Dec 09
Originally posted by BahariGood insight!
The basic principle is we have to make the exchange decision now. We are not completely sure about how the game will progress says 20 moves from now. There are billions of possible positions. If we know them then we can use the knowledge to come up with the best decision. In this case we may want to give up a queen for a pawn and still winning. Of course thi ...[text shortened]... ct at least during the time it is made. We can only hope our decision is correct in future too.
21 Dec 09
1 edit
Originally posted by Bahariwhat I've been trying to say, is that I'd put close to zero weight on principles like this in these kinds of decision points. the reason why you go with one choice or another, should be something very concrete and specific to the position at hand. if you see something you can do with B+N, and nothing for R+p, you choose B+N. but if you see something concrete you can do with R+p, and nothing for B+N, you choose R+p.
The basic principle is we have to make the exchange decision now. We are not completely sure about how the game will progress says 20 moves from now. There are billions of possible positions. If we know them then we can use the knowledge to come up with the best decision. In this case we may want to give up a queen for a pawn and still winning. Of course thi ...[text shortened]... ct at least during the time it is made. We can only hope our decision is correct in future too.
if you go by statistical general principles, that's like throwing a dart blindfolded. you don't want to do anything blindfolded at chess, unless you're absolutely forced to. avoid hope chess as the plague it is.
lets say you got 60% (I made the number up, it's not important) statistical winning chance with B+N, and 40% with R+p over all possible (realistic) positions. if you go by the statistics, you're gonna lose 40% of such games simply because of one single move. that's not good, that's catastrophic. - what you really want, is to get the choice 100% right, not 60% or 40%.
misunderstandings like these are the reason why low rated players always double opponent's pawns if they can, or snatch costly pawns in the opening. they're basing their decisions in general statistical results, which may or may NOT apply to the position. that's why the correct answer to this kind of questions is almost invariably:
"it depends on the position."
22 Dec 09
It is important to consider material balance before executing the exchange unless we know for sure that we have "clear advantage" dispite of losing the exchange. We do the same thing over and over again.
GM Karpov suggests to simplify the game even if we are only a pawn up. GM Karpov knows for sure that absolute truth is anybody can still lose the game even he or she is a queen up. However it does not stop him from making such statement.
One of the high ranking GM says this (sorry I can't remember his name-Probably GM Kasparov)- If you find a good move that wins you a knight, continue to look further. Probably there is another move that wins you a rook.
These two statements show how important to have an upper hand in material balance over the chessboard.
22 Dec 09
Originally posted by wormwoodyeah, the same hapens with the eternal question "bishop or knight?", it depends
what I've been trying to say, is that I'd put close to zero weight on principles like this in these kinds of decision points. the reason why you go with one choice or another, should be something very concrete and specific to the position at hand. if you see something you can do with B+N, and nothing for R+p, you choose B+N. but if you see something concret ...[text shortened]... r to this kind of questions is almost invariably:
"it depends on the position."