bob wrote:
But MY guess, based on looking at lots of GM games, is that a knight won't be enough against perfect play assuming that a GM is playing the stronger side. Perfect play might well show a knight wins. But can a human complete the game with zero mistakes? Seems beyond implausible based on existing games...
Based on what games? Bullet and blitz games don't count as humans can easily make very stupid mistakes when short of time. We need a minimum of one minute per move to avoid the worst thoughts we can imagine. A minus knight removes so many resources for the handicapper. It frees the human mind of troubles.
At TOURNAMENT games. I watched Topolov, for example, blunder a queen to a move I saw instantly. After making the blunder, the next move he had to give up his queen or else be mated on the next move. Use a computer to go through GM games, give them an hour per move, which is hardly up to the standard we call "perfect" and see how many mistakes are made. Usually mistakes that the opponent fail to take advantage of because they didn't see the punch line either...
As it stands, I have no idea whether a knight is a forced win or not given perfect play by both sides. But even without knowing that, I know how prone humans are to make mistakes so that I consider the chances of a GM winning with a knight handicap to be far worse than for a perfect player winning. How much worse is "far worse"? No way to know at the moment since there is nothing even remotely approaching perfect play if we go beyond 7 pieces left on the board, total.
I guess humans would blunder now and then but it wouldn't change the score completely. The more difficult the position is, the easier it is to make a mistake. A knight off helps us an extreme amount. From the start humans could play in a totally different way than normal. No need to fight for the initiative or to defend accurately. Give a pawn to trade some pieces and maybe another one too to get an endgame. Then just play for a win with a basically guaranteed draw in the worst case.
You are forgetting that you are playing against an absolutely perfect player. A minor mistake extends the game at the very least, a bigger mistake can lose the advantage completely. But the minor mistakes are a problem because extending the game extends the number of moves where the human has a chance to make another mistake.
You REALLY think you can give a computer two pawns, leaving a one pawn advantage, and still win? A perfect player? GMs have to work long and hard to win with a pawn handicap today against an anything but perfect computer program..
You mean the smart,clever,astute intelligent, sensible,crafty wily player don't you Uri? What do you mean by perfect for goodness sake?As I understand it perfection is broadly a state of completeness and flawlessness [Wikipedia] and as yet I have not seen that demonstrated in the chess arena.
Regards Allan
rabbits23 wrote:You mean the smart,clever,astute intelligent, sensible,crafty wily player don't you Uri? What do you mean by perfect for goodness sake?As I understand it perfection is broadly a state of completeness and flawlessness [Wikipedia] and as yet I have not seen that demonstrated in the chess arena.
Regards Allan
Perfect has a quite definite and precise definition. But quite a few here have never bothered looking it up it seems. GM play != perfect.
rabbits23 wrote:You mean the smart,clever,astute intelligent, sensible,crafty wily player don't you Uri? What do you mean by perfect for goodness sake?As I understand it perfection is broadly a state of completeness and flawlessness [Wikipedia] and as yet I have not seen that demonstrated in the chess arena.
Regards Allan
perfect is not playing mistakes.
always find the fastest possible mate when it is winning and always find the longest way to lose when it is losing
I feel sure that GM can beat it when the perfect player is without a knight.
I think that
the best way for the GM to win against perfect may be to go for simplifications and knight advantage allow him to force simlifications or get a bigger advantage.
The best way for the perfect player to win against everybody with the smallest number of moves is not always going for simplifications because there is no risk that the perfect player is going to do mistakes in complex position for humans.
Uri I am unable to follow your reasoning with regards to a perfect player.
Is there any player who can play a whole game of chess under time pressure[computer or GM] without making a mistake? And given the complexities and the variety of choices that can arise in a game how do you know what the so-called perfect move is? This is why I question so deeply the idea of there being a perfect player. It is simply too complicated. I agree that in certain positions a clever player will find the perfect moves to get the fastest mate and find the longest way to lose when it is losing -unless it resigns before-hand.
As for whether a GM can beat this fictitious perfect player with a knight advantage I can't comment because the concept of perfect has yet to be
defined satisfactorily to me.Additionally no parameters or guidelines have been provided either.
Cheers Allan
bob wrote:
But MY guess, based on looking at lots of GM games, is that a knight won't be enough against perfect play assuming that a GM is playing the stronger side. Perfect play might well show a knight wins. But can a human complete the game with zero mistakes? Seems beyond implausible based on existing games...
Based on what games? Bullet and blitz games don't count as humans can easily make very stupid mistakes when short of time. We need a minimum of one minute per move to avoid the worst thoughts we can imagine. A minus knight removes so many resources for the handicapper. It frees the human mind of troubles.
At TOURNAMENT games. I watched Topolov, for example, blunder a queen to a move I saw instantly. After making the blunder, the next move he had to give up his queen or else be mated on the next move. Use a computer to go through GM games, give them an hour per move, which is hardly up to the standard we call "perfect" and see how many mistakes are made. Usually mistakes that the opponent fail to take advantage of because they didn't see the punch line either...
As it stands, I have no idea whether a knight is a forced win or not given perfect play by both sides. But even without knowing that, I know how prone humans are to make mistakes so that I consider the chances of a GM winning with a knight handicap to be far worse than for a perfect player winning. How much worse is "far worse"? No way to know at the moment since there is nothing even remotely approaching perfect play if we go beyond 7 pieces left on the board, total.
I guess humans would blunder now and then but it wouldn't change the score completely. The more difficult the position is, the easier it is to make a mistake. A knight off helps us an extreme amount. From the start humans could play in a totally different way than normal. No need to fight for the initiative or to defend accurately. Give a pawn to trade some pieces and maybe another one too to get an endgame. Then just play for a win with a basically guaranteed draw in the worst case.
You are forgetting that you are playing against an absolutely perfect player. A minor mistake extends the game at the very least, a bigger mistake can lose the advantage completely. But the minor mistakes are a problem because extending the game extends the number of moves where the human has a chance to make another mistake.
You REALLY think you can give a computer two pawns, leaving a one pawn advantage, and still win? A perfect player? GMs have to work long and hard to win with a pawn handicap today against an anything but perfect computer program..
Yep without queens humans have much better chances.
Edit. Why not challenge a few gms already to a match without the queens and the b1 knight plus two black pawns.
rabbits23 wrote:Uri I am unable to follow your reasoning with regards to a perfect player.
Is there any player who can play a whole game of chess under time pressure[computer or GM] without making a mistake? And given the complexities and the variety of choices that can arise in a game how do you know what the so-called perfect move is? This is why I question so deeply the idea of there being a perfect player. It is simply too complicated. I agree that in certain positions a clever player will find the perfect moves to get the fastest mate and find the longest way to lose when it is losing -unless it resigns before-hand.
As for whether a GM can beat this fictitious perfect player with a knight advantage I can't comment because the concept of perfect has yet to be
defined satisfactorily to me.Additionally no parameters or guidelines have been provided either.
Cheers Allan
The perfect player is an egtb with 32 pieces. OR in the case of knight odds, all EGTBs with 31 or fewer pieces. That leaves no doubt as to what the best move is in any position. Won't ever happen, but we still seem to have those that argue that such a player can't win with knight odds. Always assuming that a GM can't squander a knight advantage. But that is only an assumption. My take: this can't be answered. Other's take: A GM will always win. How they know that to state it as a fact I have no idea...
bob wrote:
But MY guess, based on looking at lots of GM games, is that a knight won't be enough against perfect play assuming that a GM is playing the stronger side. Perfect play might well show a knight wins. But can a human complete the game with zero mistakes? Seems beyond implausible based on existing games...
Based on what games? Bullet and blitz games don't count as humans can easily make very stupid mistakes when short of time. We need a minimum of one minute per move to avoid the worst thoughts we can imagine. A minus knight removes so many resources for the handicapper. It frees the human mind of troubles.
At TOURNAMENT games. I watched Topolov, for example, blunder a queen to a move I saw instantly. After making the blunder, the next move he had to give up his queen or else be mated on the next move. Use a computer to go through GM games, give them an hour per move, which is hardly up to the standard we call "perfect" and see how many mistakes are made. Usually mistakes that the opponent fail to take advantage of because they didn't see the punch line either...
As it stands, I have no idea whether a knight is a forced win or not given perfect play by both sides. But even without knowing that, I know how prone humans are to make mistakes so that I consider the chances of a GM winning with a knight handicap to be far worse than for a perfect player winning. How much worse is "far worse"? No way to know at the moment since there is nothing even remotely approaching perfect play if we go beyond 7 pieces left on the board, total.
I guess humans would blunder now and then but it wouldn't change the score completely. The more difficult the position is, the easier it is to make a mistake. A knight off helps us an extreme amount. From the start humans could play in a totally different way than normal. No need to fight for the initiative or to defend accurately. Give a pawn to trade some pieces and maybe another one too to get an endgame. Then just play for a win with a basically guaranteed draw in the worst case.
You are forgetting that you are playing against an absolutely perfect player. A minor mistake extends the game at the very least, a bigger mistake can lose the advantage completely. But the minor mistakes are a problem because extending the game extends the number of moves where the human has a chance to make another mistake.
You REALLY think you can give a computer two pawns, leaving a one pawn advantage, and still win? A perfect player? GMs have to work long and hard to win with a pawn handicap today against an anything but perfect computer program..
Yep without queens humans have much better chances.
Edit. Why not challenge a few gms already to a match without the queens and the b1 knight plus two black pawns.
That could be done, but what does it prove? Computers still don't play anywhere near perfectly, even with queens removed.