Don wrote:
Computer outplay humans in endgames too. There are specific weaknesses computer have which sometimes show up, but it's not enough to prevent the large ELO superiority they have over humans.
I very much doubt that first part of your statement. I am confident that top humans will outplay top engines in pretty much all 8 piece endgames, especially if these are queenless. The engine weaknesses are just too obvious in the endgame.
Ron Langeveld
P.S. Just out of curiosity, I was wondering whether you received my emails. I haven't received a single reply since February 12th.
Don wrote:
Computer outplay humans in endgames too. There are specific weaknesses computer have which sometimes show up, but it's not enough to prevent the large ELO superiority they have over humans.
I very much doubt that first part of your statement. I am confident that top humans will outplay top engines in pretty much all 8 piece endgames, especially if these are queenless. The engine weaknesses are just too obvious in the endgame.
I think you are wrong about this. Yes, it's possible to find positions which embarrass computers and conclude that they are really weak in the endgame, but what matters is how they play in general. I am talking about the point where both queens come off, you are probably focused on a few specialized cases.
Ron Langeveld
P.S. Just out of curiosity, I was wondering whether you received my emails. I haven't received a single reply since February 12th.
About the email:
I am well past critical mass on my emails, the point where I have to let some go or I will be answering email all day long. I have been accused of not being able to write a short email so that is part of the problem. Sorry about missing your emails. I will look more carefully in the future.
Capital punishment would be more effective as a preventive measure if it were administered prior to the crime.
Don wrote:
Computer outplay humans in endgames too. There are specific weaknesses computer have which sometimes show up, but it's not enough to prevent the large ELO superiority they have over humans.
I very much doubt that first part of your statement. I am confident that top humans will outplay top engines in pretty much all 8 piece endgames, especially if these are queenless. The engine weaknesses are just too obvious in the endgame.
I think you are wrong about this. Yes, it's possible to find positions which embarrass computers and conclude that they are really weak in the endgame, but what matters is how they play in general. I am talking about the point where both queens come off, you are probably focused on a few specialized cases.
Ron Langeveld
P.S. Just out of curiosity, I was wondering whether you received my emails. I haven't received a single reply since February 12th.
About the email:
I am well past critical mass on my emails, the point where I have to let some go or I will be answering email all day long. I have been accused of not being able to write a short email so that is part of the problem. Sorry about missing your emails. I will look more carefully in the future.
Don this is just a case of some people who know you are right, but they will scratch and claw to hang on to the past. Not being critical here, as I have done the same thing in other aspects of my life. But you just have to let it go. Kasparov was in his prime when he played Deep Blue. And I would be surprised if Deep Blue today could win a match ag. Houdini, Critter, Komodo or Rybka on a quad laptop from Office Depot. (Yes, you could most likely win also on 1 core).
Don wrote:
Computer outplay humans in endgames too. There are specific weaknesses computer have which sometimes show up, but it's not enough to prevent the large ELO superiority they have over humans.
I very much doubt that first part of your statement. I am confident that top humans will outplay top engines in pretty much all 8 piece endgames, especially if these are queenless. The engine weaknesses are just too obvious in the endgame.
I think you are wrong about this. Yes, it's possible to find positions which embarrass computers and conclude that they are really weak in the endgame, but what matters is how they play in general. I am talking about the point where both queens come off, you are probably focused on a few specialized cases.
Ron Langeveld
P.S. Just out of curiosity, I was wondering whether you received my emails. I haven't received a single reply since February 12th.
About the email:
I am well past critical mass on my emails, the point where I have to let some go or I will be answering email all day long. I have been accused of not being able to write a short email so that is part of the problem. Sorry about missing your emails. I will look more carefully in the future.
Don this is just a case of some people who know you are right, but they will scratch and claw to hang on to the past. Not being critical here, as I have done the same thing in other aspects of my life. But you just have to let it go. Kasparov was in his prime when he played Deep Blue. And I would be surprised if Deep Blue today could win a match ag. Houdini, Critter, Komodo or Rybka on a quad laptop from Office Depot. (Yes, you could most likely win also on 1 core).
Best,
george
Deep Blue did not give Kasparov a chance to play the endgame, he was outplayed by some opening preparations prepared by Joel Benjamin, and in other openings that were equal Deep Blue outplayed him tactically in the middlegame. But I am not referring to the opening nor Middlegame, but simply the endgame where the engine play without a tablebase.
PS: Kasparov is not an endgame specialist or as good as GM Kramnik or GM Aronian in the endgame
Don wrote:
Computer outplay humans in endgames too. There are specific weaknesses computer have which sometimes show up, but it's not enough to prevent the large ELO superiority they have over humans.
I very much doubt that first part of your statement. I am confident that top humans will outplay top engines in pretty much all 8 piece endgames, especially if these are queenless. The engine weaknesses are just too obvious in the endgame.
I think you are wrong about this. Yes, it's possible to find positions which embarrass computers and conclude that they are really weak in the endgame, but what matters is how they play in general. I am talking about the point where both queens come off, you are probably focused on a few specialized cases.
Ron Langeveld
P.S. Just out of curiosity, I was wondering whether you received my emails. I haven't received a single reply since February 12th.
About the email:
I am well past critical mass on my emails, the point where I have to let some go or I will be answering email all day long. I have been accused of not being able to write a short email so that is part of the problem. Sorry about missing your emails. I will look more carefully in the future.
Don this is just a case of some people who know you are right, but they will scratch and claw to hang on to the past. Not being critical here, as I have done the same thing in other aspects of my life. But you just have to let it go. Kasparov was in his prime when he played Deep Blue. And I would be surprised if Deep Blue today could win a match ag. Houdini, Critter, Komodo or Rybka on a quad laptop from Office Depot. (Yes, you could most likely win also on 1 core).
Best,
george
Hi George!
Yes, I have noticed a lag period of a few years (for accepting a change of thinking) for many people on many issues. I think every human being has a natural resistance to that. And there are many misconception floating around too. Jonathon Shaeffer wrote the book "One Jump Ahead" and observes that people still think Arthur Samuels solved checkers several decades ago!
A friend of mine who is a master could not accept the fact that one of his peers was now playing 2300+ ELO strength even though he was able to hold that rating. And it was because for many years this player couldn't break 1500 strength. He was sure it was some kind of crazy fluke and couldn't be true. It was fun watching him gradually accept this but his brain just couldn't make the adjustment quickly. What actually happened was the former C player suddenly started taking chess seriously and dedicated a lot of time to serious study, where previously he loved chess but never studied and did not know how (or was unwilling) to focus or concentrate.
Capital punishment would be more effective as a preventive measure if it were administered prior to the crime.
pichy wrote:
Deep Blue did not give Kasparov a chance to play the endgame, he was outplayed by some opening preparations prepared by Joel Benjamin, and in other openings that were equal Deep Blue outplayed him tactically in the middlegame. But I am not referring to the opening nor Middlegame, but simply the endgame where the engine play without a tablebase.
PS: Kasparov is not an endgame specialist or as good as GM Kramnik or GM Aronian in the endgame
I don't believe Deep Blue really deserved to win that match. I don't want to take anything away from them, they were talented and their accomplishment was awesome but I think the odds were heavily stacked in their favor in almost every way and I think that Kasparov beat himself.
Computers are stronger in the middle-game than in the end-game, I'm not disputing that. Kasparov would have done well in endgames against Deep Blue but not playing against modern programs on fast hardware. Today's programs are still better than human players even in the endgame, it's just that their superiority is less than in the middle-game.
I will agree that there are certain types of positions (even in the middle-game) where humans are going to play as well or better than computers but that doesn't make them superior. You cannot use that as a "proof" than humans are still superior. You have to also consider all the positions where the computers are better than humans, which is MOST positions.
Don
Capital punishment would be more effective as a preventive measure if it were administered prior to the crime.
Don wrote:
Computer outplay humans in endgames too. There are specific weaknesses computer have which sometimes show up, but it's not enough to prevent the large ELO superiority they have over humans.
I very much doubt that first part of your statement. I am confident that top humans will outplay top engines in pretty much all 8 piece endgames, especially if these are queenless. The engine weaknesses are just too obvious in the endgame.
I think you are wrong about this. Yes, it's possible to find positions which embarrass computers and conclude that they are really weak in the endgame, but what matters is how they play in general. I am talking about the point where both queens come off, you are probably focused on a few specialized cases.
Ron Langeveld
P.S. Just out of curiosity, I was wondering whether you received my emails. I haven't received a single reply since February 12th.
About the email:
I am well past critical mass on my emails, the point where I have to let some go or I will be answering email all day long. I have been accused of not being able to write a short email so that is part of the problem. Sorry about missing your emails. I will look more carefully in the future.
Don this is just a case of some people who know you are right, but they will scratch and claw to hang on to the past. Not being critical here, as I have done the same thing in other aspects of my life. But you just have to let it go. Kasparov was in his prime when he played Deep Blue. And I would be surprised if Deep Blue today could win a match ag. Houdini, Critter, Komodo or Rybka on a quad laptop from Office Depot. (Yes, you could most likely win also on 1 core).
Best,
george
Hi George!
Yes, I have noticed a lag period of a few years (for accepting a change of thinking) for many people on many issues. I think every human being has a natural resistance to that. And there are many misconception floating around too. Jonathon Shaeffer wrote the book "One Jump Ahead" and observes that people still think Arthur Samuels solved checkers several decades ago!
A friend of mine who is a master could not accept the fact that one of his peers was now playing 2300+ ELO strength even though he was able to hold that rating. And it was because for many years this player couldn't break 1500 strength. He was sure it was some kind of crazy fluke and couldn't be true. It was fun watching him gradually accept this but his brain just couldn't make the adjustment quickly. What actually happened was the former C player suddenly started taking chess seriously and dedicated a lot of time to serious study, where previously he loved chess but never studied and did not know how (or was unwilling) to focus or concentrate.
I hope you noticed that I kept my promise to you. I can now test Komodo 64-bit. I am just running a match right now with Texel 64-bit ag. Zap Paderborn. I didn't load and run Komodo 64 because that would have been a bit lopsided ag. Texel. But I will be getting you more 64bit matches than I imagine you would want.
Don wrote:
Computer outplay humans in endgames too. There are specific weaknesses computer have which sometimes show up, but it's not enough to prevent the large ELO superiority they have over humans.
I very much doubt that first part of your statement. I am confident that top humans will outplay top engines in pretty much all 8 piece endgames, especially if these are queenless. The engine weaknesses are just too obvious in the endgame.
I think you are wrong about this. Yes, it's possible to find positions which embarrass computers and conclude that they are really weak in the endgame, but what matters is how they play in general. I am talking about the point where both queens come off, you are probably focused on a few specialized cases.
Ron Langeveld
P.S. Just out of curiosity, I was wondering whether you received my emails. I haven't received a single reply since February 12th.
About the email:
I am well past critical mass on my emails, the point where I have to let some go or I will be answering email all day long. I have been accused of not being able to write a short email so that is part of the problem. Sorry about missing your emails. I will look more carefully in the future.
Don this is just a case of some people who know you are right, but they will scratch and claw to hang on to the past. Not being critical here, as I have done the same thing in other aspects of my life. But you just have to let it go. Kasparov was in his prime when he played Deep Blue. And I would be surprised if Deep Blue today could win a match ag. Houdini, Critter, Komodo or Rybka on a quad laptop from Office Depot. (Yes, you could most likely win also on 1 core).
Best,
george
Hi George!
Yes, I have noticed a lag period of a few years (for accepting a change of thinking) for many people on many issues. I think every human being has a natural resistance to that. And there are many misconception floating around too. Jonathon Shaeffer wrote the book "One Jump Ahead" and observes that people still think Arthur Samuels solved checkers several decades ago!
A friend of mine who is a master could not accept the fact that one of his peers was now playing 2300+ ELO strength even though he was able to hold that rating. And it was because for many years this player couldn't break 1500 strength. He was sure it was some kind of crazy fluke and couldn't be true. It was fun watching him gradually accept this but his brain just couldn't make the adjustment quickly. What actually happened was the former C player suddenly started taking chess seriously and dedicated a lot of time to serious study, where previously he loved chess but never studied and did not know how (or was unwilling) to focus or concentrate.
I hope you noticed that I kept my promise to you. I can now test Komodo 64-bit. I am just running a match right now with Texel 64-bit ag. Zap Paderborn. I didn't load and run Komodo 64 because that would have been a bit lopsided ag. Texel. But I will be getting you more 64bit matches than I imagine you would want.
Best,
george
Now that you have Komodo 64-bit. Test it with this endgame position without tablebase to see if it can find the best move Be8 sacrificing the bishop
[d]2b2k1K/6p1/p1p2p1p/P2p1P1B/1P1P2P1/7P/8/8 w - - 0 1
pichy wrote:
Deep Blue did not give Kasparov a chance to play the endgame, he was outplayed by some opening preparations prepared by Joel Benjamin, and in other openings that were equal Deep Blue outplayed him tactically in the middlegame. But I am not referring to the opening nor Middlegame, but simply the endgame where the engine play without a tablebase.
PS: Kasparov is not an endgame specialist or as good as GM Kramnik or GM Aronian in the endgame
I don't believe Deep Blue really deserved to win that match. I don't want to take anything away from them, they were talented and their accomplishment was awesome but I think the odds were heavily stacked in their favor in almost every way and I think that Kasparov beat himself.
Computers are stronger in the middle-game than in the end-game, I'm not disputing that. Kasparov would have done well in endgames against Deep Blue but not playing against modern programs on fast hardware. Today's programs are still better than human players even in the endgame, it's just that their superiority is less than in the middle-game.
I will agree that there are certain types of positions (even in the middle-game) where humans are going to play as well or better than computers but that doesn't make them superior. You cannot use that as a "proof" than humans are still superior. You have to also consider all the positions where the computers are better than humans, which is MOST positions.
Don
P.S.
I can tell you why endgame databases have very little impact on the strength of chess programs. Years ago I implemented my own KPvsK database so that it would analyze perfectly when reaching these endings or deciding whether to go into these endings. I ran tens of thousands of games to try to prove that the program was better. I could not prove it. I started looking at the game and in the first 100 or more than I looked at I could not find a single game that was even likely to get into this ending.
Most of the database are trivial, the position is either mostly won or mostly draw except in unusual circumstances which computers can usually see tactically. So there are very few databases that are actually interesting or have some hope of improving the programs play in a noticeable way. Probably the most important practical ending handle by these databases is KRP vs KR.
This ending is practical because it is relatively common. I say "relatively" because there is not single ending that occurs a very high percentage of the time. A lot of games are over before you get to the endings, even if the ending is played out. Only a fraction end up getting to the point where KRPvsKR is going to be useful. For this ending to be worth a "bunch of ELO" it has to actually change the outcome of the game. In other words, if the position is a draw and your engine does not play it correctly, you could lose. In practice that only happens SOME of the time.
So to summarize your database is not going to help the program much because it only covers practical endings that your program does not see very often and does not necessary mis-play it the few times it does occur.
Working against this is the fact that there is SOME overhead for using these database which take a toll on the strength of the program. Many clever things have been done in the design and implementation of database to minimize this and it helps to run these from a SSD drive, but it's still not free. So the value of the database has to first overcome this deficit before it can start being an improvement. When you consider that even without the overhead the improvement is tiny, you don't have much to work with.
Endgame databases will start to make a big difference in playing strength once you get past 7 or 8 pieces and can cover every reasonable configuration, not just a few that are easy to compute. This assumes you can do it with a very small overhead. If you can get past 10-12 pieces you are talking about really super-charging the strength of the engines. However when even 6 piece databases challenge the limits of technology and what can be expected from a home computing system we have a very long way to go.
I do plan support for Gaviota as I know it's a good thing for analysis and it's a popular feature - I would even like it for myself in Komodo. But I'm not expecting any ELO gain from it.
Capital punishment would be more effective as a preventive measure if it were administered prior to the crime.
Don wrote:
Computer outplay humans in endgames too. There are specific weaknesses computer have which sometimes show up, but it's not enough to prevent the large ELO superiority they have over humans.
I very much doubt that first part of your statement. I am confident that top humans will outplay top engines in pretty much all 8 piece endgames, especially if these are queenless. The engine weaknesses are just too obvious in the endgame.
I think you are wrong about this. Yes, it's possible to find positions which embarrass computers and conclude that they are really weak in the endgame, but what matters is how they play in general. I am talking about the point where both queens come off, you are probably focused on a few specialized cases.
Ron Langeveld
P.S. Just out of curiosity, I was wondering whether you received my emails. I haven't received a single reply since February 12th.
About the email:
I am well past critical mass on my emails, the point where I have to let some go or I will be answering email all day long. I have been accused of not being able to write a short email so that is part of the problem. Sorry about missing your emails. I will look more carefully in the future.
Don this is just a case of some people who know you are right, but they will scratch and claw to hang on to the past. Not being critical here, as I have done the same thing in other aspects of my life. But you just have to let it go. Kasparov was in his prime when he played Deep Blue. And I would be surprised if Deep Blue today could win a match ag. Houdini, Critter, Komodo or Rybka on a quad laptop from Office Depot. (Yes, you could most likely win also on 1 core).
Best,
george
Hi George!
Yes, I have noticed a lag period of a few years (for accepting a change of thinking) for many people on many issues. I think every human being has a natural resistance to that. And there are many misconception floating around too. Jonathon Shaeffer wrote the book "One Jump Ahead" and observes that people still think Arthur Samuels solved checkers several decades ago!
A friend of mine who is a master could not accept the fact that one of his peers was now playing 2300+ ELO strength even though he was able to hold that rating. And it was because for many years this player couldn't break 1500 strength. He was sure it was some kind of crazy fluke and couldn't be true. It was fun watching him gradually accept this but his brain just couldn't make the adjustment quickly. What actually happened was the former C player suddenly started taking chess seriously and dedicated a lot of time to serious study, where previously he loved chess but never studied and did not know how (or was unwilling) to focus or concentrate.
I hope you noticed that I kept my promise to you. I can now test Komodo 64-bit. I am just running a match right now with Texel 64-bit ag. Zap Paderborn. I didn't load and run Komodo 64 because that would have been a bit lopsided ag. Texel. But I will be getting you more 64bit matches than I imagine you would want.
I think you will find that 64 bit Komodo is a different animal than 32 bit Komodo. Komodo suffers more than other top programs going to 32 bit. I want to look into some things to improve that situation, but right now my focus is on MP and many other things. I did make a change over 2 years ago that made it not as bad as it would have been, one of the low level routines was not optimal for 32 bits and that made a big difference, but I can do more of this kind of thing I believe.
Best,
george
Capital punishment would be more effective as a preventive measure if it were administered prior to the crime.
