It will be a proof when freezer is run against the problem and verifies the result. Until then, it is only a very good hypothesis.Terry McCracken wrote:Uri, George posted a link above, the utility you need for your final proof isUri Blass wrote:Robin's post is convincing but I do not consider it as a proof.Terry McCracken wrote:Huh??Uri Blass wrote:I agree that this is more convincing and can be used for what I consider as a real proof that I am lazy to construct(I still do not consider it as a proof inspite of the fact that it convinced me).
I think that black even does not need the quare a6 and can do fine even on 7*7 board with only c6 and b6
In order to construct a proof that I am too lazy to do we pracitcally need simply to divide the possible pair of squares of white king and white queen to 12 classes and give the corresponding squares of black king and rook in every class.
After this we need to prove that after white does a move in one of the classes black can capture the queen or make a move that goes to one of the classes.
Uri
Robin gave proof, and it's quite simple
Page One.... http://www.talkchess.com/forum/viewtopi ... 72&start=0
I simply disagree with you about the meaning of the word proof.
Robin gave a strategy to draw the game.
A proof should prove that the strategy is really possible and that white cannot do something different that I did not imagine.
I am not sure if we practically need to do what I suggested and it is only one way to prove the draw.
Thinking about again it may be more simpler to prove that the strategy that Robin suggested is possible but it needs proof.
Here is part of the proof that you need to write in order to prove it based on my standards and not only in order to convince me(I added comments to Robin comments inside the ().
2) If the king is in check, move out of check to any one of the squares h7, g8, h8(possible but need proof because the queen may control all 3 squares from g7 and if the white king is at f8 white is winning
the white king can never go to f8 but in order to prove that black draw we need to mention the possible squares of the white king and prove that white cannot escape them in every one of the cases)
3) If the rook is not on f6 move the rook to f6(possible assuming the rook is not pinned but the rook cannot be pinned when the black king is at h7 g8 h8 and the rook is at f6 f2 c6 b6 a6).
Uri
Freezer.
If this can't give a proof, then you'll have to run a hundred plys full width!
In ten years that might not be a problem
Terry
Is this positions drawn ??
Moderators: hgm, Rebel, chrisw
-
- Posts: 12540
- Joined: Wed Mar 08, 2006 8:57 pm
- Location: Redmond, WA USA
Re: Is this positions drawn ??
-
- Posts: 2307
- Joined: Wed Mar 08, 2006 8:41 pm
- Location: Brownsville Texas USA
Re: Is this positions drawn ??
Hi Dan,Dann Corbit wrote:It will be a proof when freezer is run against the problem and verifies the result. Until then, it is only a very good hypothesis.Terry McCracken wrote:Uri, George posted a link above, the utility you need for your final proof isUri Blass wrote:Robin's post is convincing but I do not consider it as a proof.Terry McCracken wrote:Huh??Uri Blass wrote:I agree that this is more convincing and can be used for what I consider as a real proof that I am lazy to construct(I still do not consider it as a proof inspite of the fact that it convinced me).
I think that black even does not need the quare a6 and can do fine even on 7*7 board with only c6 and b6
In order to construct a proof that I am too lazy to do we pracitcally need simply to divide the possible pair of squares of white king and white queen to 12 classes and give the corresponding squares of black king and rook in every class.
After this we need to prove that after white does a move in one of the classes black can capture the queen or make a move that goes to one of the classes.
Uri
Robin gave proof, and it's quite simple
Page One.... http://www.talkchess.com/forum/viewtopi ... 72&start=0
I simply disagree with you about the meaning of the word proof.
Robin gave a strategy to draw the game.
A proof should prove that the strategy is really possible and that white cannot do something different that I did not imagine.
I am not sure if we practically need to do what I suggested and it is only one way to prove the draw.
Thinking about again it may be more simpler to prove that the strategy that Robin suggested is possible but it needs proof.
Here is part of the proof that you need to write in order to prove it based on my standards and not only in order to convince me(I added comments to Robin comments inside the ().
2) If the king is in check, move out of check to any one of the squares h7, g8, h8(possible but need proof because the queen may control all 3 squares from g7 and if the white king is at f8 white is winning
the white king can never go to f8 but in order to prove that black draw we need to mention the possible squares of the white king and prove that white cannot escape them in every one of the cases)
3) If the rook is not on f6 move the rook to f6(possible assuming the rook is not pinned but the rook cannot be pinned when the black king is at h7 g8 h8 and the rook is at f6 f2 c6 b6 a6).
Uri
Freezer.
If this can't give a proof, then you'll have to run a hundred plys full width!
In ten years that might not be a problem
Terry
Even Freezer will not generate a true proof, in the strict mathematical sense, which seems to be what you and Uri seek. This is not only because Freezer's source is not open, as Uri mentioned, but also because Freezer does not generate a full database of all possible relevant positions. In order to keep the computation size and eventual database size manageable, Freezer allows and often even requires users to define rules that will reduce the database computation to levels manageable by Freezer. In the specific position in question at least some rules are required by Freezer. Depending on what rules a user defines the _exact_ same position might result in an evaluation of "White wins", "Draw" or "Black wins", so clearly specifying these rules properly is essential. A savvy Freezer user will be good at defining rules which will result in an accurate conclusion, but this hardly constitutes a true proof in the mathematical sense, any more than my move generating algorithm did.
P.S. I ran Freezer on the position in question. The conclusion Freezer/I reached was "Draw". However I do not have a complete set of 6 man tablebases and I had to define several rules for Freezer to use in its database generation. The first of the rules I defined was "If White's queen is captured" = draw.
- Robin Smith
-
- Posts: 12540
- Joined: Wed Mar 08, 2006 8:57 pm
- Location: Redmond, WA USA
Re: Is this positions drawn ??
I read the Freezer thesis paper. If all the rules supplied are correct, then I believe that the outcome is correct. I also agree with your supposition (except to say draw is the best possible outcome -- it is also possible to lose the queen and lose the game).smirobth wrote:Hi Dan,Dann Corbit wrote:It will be a proof when freezer is run against the problem and verifies the result. Until then, it is only a very good hypothesis.Terry McCracken wrote:Uri, George posted a link above, the utility you need for your final proof isUri Blass wrote:Robin's post is convincing but I do not consider it as a proof.Terry McCracken wrote:Huh??Uri Blass wrote:I agree that this is more convincing and can be used for what I consider as a real proof that I am lazy to construct(I still do not consider it as a proof inspite of the fact that it convinced me).
I think that black even does not need the quare a6 and can do fine even on 7*7 board with only c6 and b6
In order to construct a proof that I am too lazy to do we pracitcally need simply to divide the possible pair of squares of white king and white queen to 12 classes and give the corresponding squares of black king and rook in every class.
After this we need to prove that after white does a move in one of the classes black can capture the queen or make a move that goes to one of the classes.
Uri
Robin gave proof, and it's quite simple
Page One.... http://www.talkchess.com/forum/viewtopi ... 72&start=0
I simply disagree with you about the meaning of the word proof.
Robin gave a strategy to draw the game.
A proof should prove that the strategy is really possible and that white cannot do something different that I did not imagine.
I am not sure if we practically need to do what I suggested and it is only one way to prove the draw.
Thinking about again it may be more simpler to prove that the strategy that Robin suggested is possible but it needs proof.
Here is part of the proof that you need to write in order to prove it based on my standards and not only in order to convince me(I added comments to Robin comments inside the ().
2) If the king is in check, move out of check to any one of the squares h7, g8, h8(possible but need proof because the queen may control all 3 squares from g7 and if the white king is at f8 white is winning
the white king can never go to f8 but in order to prove that black draw we need to mention the possible squares of the white king and prove that white cannot escape them in every one of the cases)
3) If the rook is not on f6 move the rook to f6(possible assuming the rook is not pinned but the rook cannot be pinned when the black king is at h7 g8 h8 and the rook is at f6 f2 c6 b6 a6).
Uri
Freezer.
If this can't give a proof, then you'll have to run a hundred plys full width!
In ten years that might not be a problem
Terry
Even Freezer will not generate a true proof, in the strict mathematical sense, which seems to be what you and Uri seek. This is not only because Freezer's source is not open, as Uri mentioned, but also because Freezer does not generate a full database of all possible relevant positions. In order to keep the computation size and eventual database size manageable, Freezer allows and often even requires users to define rules that will reduce the database computation to levels manageable by Freezer. In the specific position in question at least some rules are required by Freezer. Depending on what rules a user defines the _exact_ same position might result in an evaluation of "White wins", "Draw" or "Black wins", so clearly specifying these rules properly is essential. A savvy Freezer user will be good at defining rules which will result in an accurate conclusion, but this hardly constitutes a true proof in the mathematical sense, any more than my move generating algorithm did.
P.S. I ran Freezer on the position in question. The conclusion Freezer/I reached was "Draw". However I do not have a complete set of 6 man tablebases and I had to define several rules for Freezer to use in its database generation. The first of the rules I defined was "If White's queen is captured" = draw.
Re: Is this positions drawn ??
You have no confidence in players far better than yourselfDann Corbit wrote:It will be a proof when freezer is run against the problem and verifies the result. Until then, it is only a very good hypothesis.Terry McCracken wrote:Uri, George posted a link above, the utility you need for your final proof isUri Blass wrote:Robin's post is convincing but I do not consider it as a proof.Terry McCracken wrote:Huh??Uri Blass wrote:I agree that this is more convincing and can be used for what I consider as a real proof that I am lazy to construct(I still do not consider it as a proof inspite of the fact that it convinced me).
I think that black even does not need the quare a6 and can do fine even on 7*7 board with only c6 and b6
In order to construct a proof that I am too lazy to do we pracitcally need simply to divide the possible pair of squares of white king and white queen to 12 classes and give the corresponding squares of black king and rook in every class.
After this we need to prove that after white does a move in one of the classes black can capture the queen or make a move that goes to one of the classes.
Uri
Robin gave proof, and it's quite simple
Page One.... http://www.talkchess.com/forum/viewtopi ... 72&start=0
I simply disagree with you about the meaning of the word proof.
Robin gave a strategy to draw the game.
A proof should prove that the strategy is really possible and that white cannot do something different that I did not imagine.
I am not sure if we practically need to do what I suggested and it is only one way to prove the draw.
Thinking about again it may be more simpler to prove that the strategy that Robin suggested is possible but it needs proof.
Here is part of the proof that you need to write in order to prove it based on my standards and not only in order to convince me(I added comments to Robin comments inside the ().
2) If the king is in check, move out of check to any one of the squares h7, g8, h8(possible but need proof because the queen may control all 3 squares from g7 and if the white king is at f8 white is winning
the white king can never go to f8 but in order to prove that black draw we need to mention the possible squares of the white king and prove that white cannot escape them in every one of the cases)
3) If the rook is not on f6 move the rook to f6(possible assuming the rook is not pinned but the rook cannot be pinned when the black king is at h7 g8 h8 and the rook is at f6 f2 c6 b6 a6).
Uri
Freezer.
If this can't give a proof, then you'll have to run a hundred plys full width!
In ten years that might not be a problem
Terry
It's a draw Dan, and a well known and understood draw.
Re: Is this positions drawn ??
Dann Corbit wrote:I read the Freezer thesis paper. If all the rules supplied are correct, then I believe that the outcome is correct. I also agree with your supposition (except to say draw is the best possible outcome -- it is also possible to lose the queen and lose the game).smirobth wrote:Hi Dan,Dann Corbit wrote:It will be a proof when freezer is run against the problem and verifies the result. Until then, it is only a very good hypothesis.Terry McCracken wrote:Uri, George posted a link above, the utility you need for your final proof isUri Blass wrote:Robin's post is convincing but I do not consider it as a proof.Terry McCracken wrote:Huh??Uri Blass wrote:I agree that this is more convincing and can be used for what I consider as a real proof that I am lazy to construct(I still do not consider it as a proof inspite of the fact that it convinced me).
I think that black even does not need the quare a6 and can do fine even on 7*7 board with only c6 and b6
In order to construct a proof that I am too lazy to do we pracitcally need simply to divide the possible pair of squares of white king and white queen to 12 classes and give the corresponding squares of black king and rook in every class.
After this we need to prove that after white does a move in one of the classes black can capture the queen or make a move that goes to one of the classes.
Uri
Robin gave proof, and it's quite simple
Page One.... http://www.talkchess.com/forum/viewtopi ... 72&start=0
I simply disagree with you about the meaning of the word proof.
Robin gave a strategy to draw the game.
A proof should prove that the strategy is really possible and that white cannot do something different that I did not imagine.
I am not sure if we practically need to do what I suggested and it is only one way to prove the draw.
Thinking about again it may be more simpler to prove that the strategy that Robin suggested is possible but it needs proof.
Here is part of the proof that you need to write in order to prove it based on my standards and not only in order to convince me(I added comments to Robin comments inside the ().
2) If the king is in check, move out of check to any one of the squares h7, g8, h8(possible but need proof because the queen may control all 3 squares from g7 and if the white king is at f8 white is winning
the white king can never go to f8 but in order to prove that black draw we need to mention the possible squares of the white king and prove that white cannot escape them in every one of the cases)
3) If the rook is not on f6 move the rook to f6(possible assuming the rook is not pinned but the rook cannot be pinned when the black king is at h7 g8 h8 and the rook is at f6 f2 c6 b6 a6).
Uri
Freezer.
If this can't give a proof, then you'll have to run a hundred plys full width!
In ten years that might not be a problem
Terry
Even Freezer will not generate a true proof, in the strict mathematical sense, which seems to be what you and Uri seek. This is not only because Freezer's source is not open, as Uri mentioned, but also because Freezer does not generate a full database of all possible relevant positions. In order to keep the computation size and eventual database size manageable, Freezer allows and often even requires users to define rules that will reduce the database computation to levels manageable by Freezer. In the specific position in question at least some rules are required by Freezer. Depending on what rules a user defines the _exact_ same position might result in an evaluation of "White wins", "Draw" or "Black wins", so clearly specifying these rules properly is essential. A savvy Freezer user will be good at defining rules which will result in an accurate conclusion, but this hardly constitutes a true proof in the mathematical sense, any more than my move generating algorithm did.
P.S. I ran Freezer on the position in question. The conclusion Freezer/I reached was "Draw". However I do not have a complete set of 6 man tablebases and I had to define several rules for Freezer to use in its database generation. The first of the rules I defined was "If White's queen is captured" = draw.
Re: Is this positions drawn ??
Robin, you improvised and even removed the Black h6 pawn and EGTBs returned a draw. What , more proof do Dan and Uri need??smirobth wrote:Hi Dan,Dann Corbit wrote:It will be a proof when freezer is run against the problem and verifies the result. Until then, it is only a very good hypothesis.Terry McCracken wrote:Uri, George posted a link above, the utility you need for your final proof isUri Blass wrote:Robin's post is convincing but I do not consider it as a proof.Terry McCracken wrote:Huh??Uri Blass wrote:I agree that this is more convincing and can be used for what I consider as a real proof that I am lazy to construct(I still do not consider it as a proof inspite of the fact that it convinced me).
I think that black even does not need the quare a6 and can do fine even on 7*7 board with only c6 and b6
In order to construct a proof that I am too lazy to do we pracitcally need simply to divide the possible pair of squares of white king and white queen to 12 classes and give the corresponding squares of black king and rook in every class.
After this we need to prove that after white does a move in one of the classes black can capture the queen or make a move that goes to one of the classes.
Uri
Robin gave proof, and it's quite simple
Page One.... http://www.talkchess.com/forum/viewtopi ... 72&start=0
I simply disagree with you about the meaning of the word proof.
Robin gave a strategy to draw the game.
A proof should prove that the strategy is really possible and that white cannot do something different that I did not imagine.
I am not sure if we practically need to do what I suggested and it is only one way to prove the draw.
Thinking about again it may be more simpler to prove that the strategy that Robin suggested is possible but it needs proof.
Here is part of the proof that you need to write in order to prove it based on my standards and not only in order to convince me(I added comments to Robin comments inside the ().
2) If the king is in check, move out of check to any one of the squares h7, g8, h8(possible but need proof because the queen may control all 3 squares from g7 and if the white king is at f8 white is winning
the white king can never go to f8 but in order to prove that black draw we need to mention the possible squares of the white king and prove that white cannot escape them in every one of the cases)
3) If the rook is not on f6 move the rook to f6(possible assuming the rook is not pinned but the rook cannot be pinned when the black king is at h7 g8 h8 and the rook is at f6 f2 c6 b6 a6).
Uri
Freezer.
If this can't give a proof, then you'll have to run a hundred plys full width!
In ten years that might not be a problem
Terry
Even Freezer will not generate a true proof, in the strict mathematical sense, which seems to be what you and Uri seek. This is not only because Freezer's source is not open, as Uri mentioned, but also because Freezer does not generate a full database of all possible relevant positions. In order to keep the computation size and eventual database size manageable, Freezer allows and often even requires users to define rules that will reduce the database computation to levels manageable by Freezer. In the specific position in question at least some rules are required by Freezer. Depending on what rules a user defines the _exact_ same position might result in an evaluation of "White wins", "Draw" or "Black wins", so clearly specifying these rules properly is essential. A savvy Freezer user will be good at defining rules which will result in an accurate conclusion, but this hardly constitutes a true proof in the mathematical sense, any more than my move generating algorithm did.
P.S. I ran Freezer on the position in question. The conclusion Freezer/I reached was "Draw". However I do not have a complete set of 6 man tablebases and I had to define several rules for Freezer to use in its database generation. The first of the rules I defined was "If White's queen is captured" = draw.
http://www.shredderchess.com/online-che ... abase.html
-
- Posts: 2307
- Joined: Wed Mar 08, 2006 8:41 pm
- Location: Brownsville Texas USA
Re: Is this positions drawn ??
Yes of course, but just as with tablebases, Freezer must identify every possible position with only one of three possible specific outcomes; "won", "drawn", or "lost" ... there is no way to flag a position as "drawn or lost".Dann Corbit wrote: I also agree with your supposition (except to say draw is the best possible outcome -- it is also possible to lose the queen and lose the game).
- Robin Smith
-
- Posts: 12540
- Joined: Wed Mar 08, 2006 8:57 pm
- Location: Redmond, WA USA
Re: ComputerChess is bad for you !!
Neither Uri nor I have ever said we did not think it was a draw. Both of us agree that (almost certainly) it is. With Robin's Freezer analysis, the uncertainty is very, very low.Terry McCracken wrote:This is just over the top....ok then, only God can know...sigh...Uri Blass wrote:It should be an easy task to write a program that proves that it is a drawTerry McCracken wrote:George, if you have the program, please demonstrate to the skeptics, that this is indeed a drawn position.George Tsavdaris wrote:Have you heard about Freezer?Dann Corbit wrote:Sorry, but there are no 7 man tablebase files with pawns constructed. There are no chess engines I know of that have resolved it either, by elimation of chessmen. There is no proof for this position, only highly probable speculation.Terry McCracken wrote: There is nothing important in this position, and endgame table bases have already proved that.
It can solve such positions! In fact it is ideal for such positions. It can generate solutions to 7-8 piece problems even if you don't have (obviously!) the 7-8 piece tablebases and only have 6 or even 5 piece tablebases.....
Terry
but a program that comes with no source code proves nothing for people who do not see the source because they cannot be 100% sure that it has no bugs.
Uri
The distinction that is made here is between proof and acceptance.
Almost all mathematicians accept the Riemann hypothesis:
http://mathworld.wolfram.com/RiemannHypothesis.html
All known evidence and measurements support the hypothesis.
However, despite the fact that we believe it, it has never been proven.
-
- Posts: 3707
- Joined: Thu Mar 16, 2006 7:10 pm
Re: ComputerChess is bad for you !!
Yes I think both Uri and Dan have made it clear they realize it is a draw ... and ofcourse it is a theoretical draw. The idea is to setup something that can prove it without any doubt similar to TB's. Main point being that in some other less clear position such a system to prove it would be useful. I have a feeling that there are many positions that humans for centuries have evaluated as losing or winning or drawing ... to be wrong. Already I see this in many computer games that are played ... doubled pawns are not always bad ... putting a rook on an open corridor is not always the best move ... 2 bishops don't ususally mean a superior position ... many many ideas will change and be corrected with the advent of computers.
Re: Is this positions drawn ??
Note the words "general rule" Uri. It means in that there often are exceptions. Anyway, in my experience, this general rule is right a lot more often than its wrong. And in general ( ) thats all you can ask of a general rule!Uri Blass wrote:This is not correct.Cubeman wrote:Also as a general rule.When engines see a big plus from the start of evalution to the end of evalution but it has not grown, even decresed after some time you can be pretty sure no progress can be made.
It is possible that they simply do not see the right plan that is too long for them to see but it is possible to make a progress by the right plan.
Uri