Will computers ever solve chess?

Sort:
Elroch
ponz111 wrote:
Elroch wrote:

You claimed to be virtually certain about the result of the starting position but I would think it very likely you would agree there are positions about which you could not reach 99.99% certainty. Extremely difficult endings would easily qualify.

I agree with this. But methinks there are zillions of ways to force a draw and such endgame positions can be avoided. [or don't happen] 

That's essentially "methinking the conclusion". Not what one would call reasoning.

ponz111
Elroch wrote:
ponz111 wrote:
Elroch wrote:

You claimed to be virtually certain about the result of the starting position but I would think it very likely you would agree there are positions about which you could not reach 99.99% certainty. Extremely difficult endings would easily qualify.

I agree with this. But methinks there are zillions of ways to force a draw and such endgame positions can be avoided. [or don't happen] 

That's essentially "methinking the conclusion". Not what one would call reasoning.

Disagree. I can see how any such position can be avoided. One does not have to play into the Kings Indian or any such closed systems.

[you did not  respond to my post #6098] 

ponz111

Lots of laughs--i think i am outnumbered here!Laughing

DiogenesDue
ponz111 wrote:

The last time i played USCF otb was in 1973--16 games against a lot of strong players including a GM [who i beat]. My performance rating was approximately 2438 for those 16 games.

However, my preference is and was correspondence chess where I won the US Championship with a score in the Finals of 13 wins, 1 draw, and no losses.  My last rating was above 2520.  However if you were to add in my games in the preliminary rounds my rating would be approximately 2575. 

And i was playing as an amateur.   [so my point is, i am not as low as you think]

If your memory were better, you'd recall that Smyslovfan and I defended your playing strength and credentials when somebody would not believe them (and on this very thread, I believe).  I know your playing strength.  Now it's time to pony up and admit that are not anywhere near 2500 anymore, due to your own admitted diagnosis (was it Alzheimer's or dementia?  Not sure, but something affecting cognitive reasoning).  This is one reason why you and your buddy in Ponziani Power had been all but accused of using outside assistance in your Votechess games.  You all but admitted it yourself in 150+ pages Cheating Forum thread about the Ponziani Power vs. Bobby Fischer Group game.

You have a hard time accepting that you are not playing at your peak anymore.  I think 2200-2300 was generous enough here.  I have great respect for your accomplishments, but you need to "keep it real".

However chess engines are now very advanced. Alpha Zero is getting close to the ultimate playing strength [this is my opinion].

This is as unfounded as your 99.9% claim.  Statements like this make it clear you are just guessing and are not really sure of anything.

But there is something STRONGER than the very best chess engines. and that is what i can now use to analyze games and spot errors in games.

Yes, yes...player plus engine beats engine alone.  This does not mean you can recognize and interpret engine's "errors" in positional play. 

One reason Carlsen cannot win a game vs Stockfish is that chess is a draw when neither side makes a mistake. Chess engines have a big advantage vs  the top human players. That advantage is that chess engines can calculate much faster than the best humans. The advantage would be somewhat taken away if humans were given more time to calculate. As it is--the humans are not given this time.

And the win ratio for engines which are playing another engine the same strength is not going up. 

Draw rates between players/engines playing themselves are, quite logically, higher than normal.  As for even strengths for different engines, TCEC results this last time around not only had a relatively high "decisive" rate...even black won some games in the championship round.

Of course if an engine is stronger than another engine--the win ratio could go up.

 Look at history take all the World Championships for example--the draw ratio keeps going up and up. 

Human play is far behind and all but irrelevant to this discussion.

Also in the venue which is stronger than the best chess engines--the win ratio also keeps going down.

...I doubt that the win ratio for centaur Chess is actually going down in recent times (when engines were 3150+ rating), it's already rock bottom.  Please give us some stats for this claim.  

There is a ton of evidence that chess is a draw.  i cannot give it all here but i have given some of the evidence before and it took several pages...

Comments above in blue.

ponz111
btickler wrote:  ponz in red
ponz111 wrote:

The last time i played USCF otb was in 1973--16 games against a lot of strong players including a GM [who i beat]. My performance rating was approximately 2438 for those 16 games.

However, my preference is and was correspondence chess where I won the US Championship with a score in the Finals of 13 wins, 1 draw, and no losses.  My last rating was above 2520.  However if you were to add in my games in the preliminary rounds my rating would be approximately 2575. 

And i was playing as an amateur.   [so my point is, i am not as low as you think]

If your memory were better, you'd recall that Smyslovfan and I defended your playing strength and credentials when somebody would not believe them (and on this very thread, I believe).  I know your playing strength.  Now it's time to pony up and admit that are not anywhere near 2500 anymore, due to your own admitted diagnosis (was it Alzheimer's or dementia? There was a diagnosis of Alzheimer's but that was clearly wrong. [i was depressed at the time] I do have some memory loss but it does not affect so much my chess playing. I am mostly out of vote chess now. [2 partial games playing and not from the first move] as my playing strength deteriorated fairly recently due to health problems. Before the recent health problems i was probably playing at about the 2275 USCF level {down about 300 points} I freely admit that my chess playing level has gone down considerably. One does not need a high level of playing strength to do well at vote chess if you have other vote chess  strong players helping. The vote chess teams are relatively weak so any team which has a few master rated players who are willing to do the hard work--can do very well.  When I played on Ponziani Power we received White every game and every game  was the Ponziani Opening which i know backwards and forwards. I think we lost 2 or 3 games and when i started playing we had a record of something like 12 wins and 1 loss? 

   

Not sure, but something affecting cognitive reasoning).  This is one reason why you and your buddy in Ponziani Power had been all but accused of using outside assistance in your Votechess games.  You all but admitted it yourself in 150+ pages Cheating Forum thread about the Ponziani Power vs. Bobby Fischer Group game. This is not true!  Where do you get this?  It had to be one particular member of our team who was upset with me. That particular game had a long series of forced moves  which i and most of the team could follow which led to a forced win at the end.  I am very good at following forced lines to the end--especially if there is 3 days for one move.

You have a hard time accepting that you are not playing at your peak anymore.  I think 2200-2300 was generous enough here.  I have great respect for your accomplishments, but you need to "keep it real". I have no trouble at all admitting i am not playing at my peak anymore. Just ask members of the current vote chess team i am playing on.

However chess engines are now very advanced. Alpha Zero is getting close to the ultimate playing strength [this is my opinion].

This is as unfounded as your 99.9% claim.  Statements like this make it clear you are just guessing and are not really sure of anything. I said it was a guess--never said it was a fact.

But there is something STRONGER than the very best chess engines. and that is what i can now use to analyze games and spot errors in games.

Yes, yes...player plus engine beats engine alone.  This does not mean you can recognize and interpret engine's "errors" in positional play. Actually my chess knowledge helps me with this. Most of my wins throughout the years were based on positional play--usually followed with a winning tactic. 

One reason Carlsen cannot win a game vs Stockfish is that chess is a draw when neither side makes a mistake. Chess engines have a big advantage vs  the top human players. That advantage is that chess engines can calculate much faster than the best humans. The advantage would be somewhat taken away if humans were given more time to calculate. As it is--the humans are not given this time.

And the win ratio for engines which are playing another engine the same strength is not going up. 

Draw rates between players/engines playing themselves are, quite logically, higher than normal.  As for even strengths for different engines, TCEC results this last time around not only had a high "decisive" rate...black won some games in the championship round.  That hasn't happened for a long time. It can happen. The  results i have seen all show something like 90% plus of all games being drawn. Why don't you give some cross tables to enlighten us?  

Of course if an engine is stronger than another engine--the win ratio could go up.

 Look at history take all the World Championships for example--the draw ratio keeps going up and up. 

Human play is far behind and all but irrelevant to this discussion. it shows a pattern--the stronger the players--the more draws. thus "relevent"

Also in the venue which is stronger than the best chess engines--the win ratio also keeps going down.

...I doubt that the win ratio for centaur Chess is actually going down in recent times (when engines were 3150+ rating), it's already rock bottom.  Please give us some stats for this claim.  i got this just from looking at cross tables for many years.  If i could import some of the cross tables--i would. 

There is a ton of evidence that chess is a draw.  i cannot give it all here but i have given some of the evidence before and it took several pages...

Comments above in blue.

ponz111

One of my main points is that as chess players become stronger [and this includes chess engines] there are more and more draws. This can be seen easily by looking at chess history.

This, in itself, is very telling evidence that chess is a draw with best play.

Elroch

The ratio of draws to wins is not good evidence that chess is a draw. Whatever the true result of chess, as play gets more accurate, you would expect lower variance in results because the errors are smaller. This can increase the number of draws regardless of the true perfect result.

The more relevant statistic is white's score. This seems to rise with strength, right up to the level of the current very strongest computer players. Certainly, it is still a lot nearer 0.5 than 1.0, but any interpretation of this depends on how close to perfection current players are. I think it is possible for perfection to be over 4000 Elo: I am very unsure about the actual level.

If the fairly high ratio of draws to wins at the very highest level (eg 3:1) makes anyone feel chess is a win, imagine an ending where white has a forced win but it requires great precision and where black can secure a draw if white goes astray at one of many stages. It is not inconceivable that chess itself is similar.

ponz111

Also, i have been noticed that even when a player makes a noted inferior move in an opening [inferior from a practical point of view] there is often not a way to force a win.  White sometimes gets up more than half a pawn up but the position becomes obviously drawn.  

[this is some evidence but not as strong as some of my other evidence]

By the way it is not just me who notices this. A lot  of very strong players notice this.

Elroch

Obviously if white has a practical advantage, white giving up a small practical advantage will often leave the game drawn. The player at a practical disadvantage may have more opportunities to make what appear to be modest errors that are crucial and lose.

Note that even thinking like this reveals very hazy knowledge of the truth: there are in truth no small advantages and the only mistakes are blunders that change the score. All the rest is just practical considerations for players who can't tell which positions are winning (i..e. all of us).

ponz111
Elroch wrote:  ponz in red

The ratio of draws to wins is not good evidence that chess is a draw. Simply disagree with this.  

 

Whatever the true result of chess, as play gets more accurate, you would expect lower variance in results because the errors are smaller. Yes, you would expect lower variance in the results and if [with the lower variance] the results head for one result--this is very telling. 

This can increase the number of draws regardless of the true perfect result. If chess was a win--why wouldn't  that increase the number of wins?  There is an increase in draws precisely because, with best play, chess is a draw.

The more relevant statistic is white's score. This seems to rise with strength, right up to the level of the current very strongest computer players. Certainly, it is still a lot nearer 0.5 than 1.0, but any interpretation of this depends on how close to perfection current players are. I think it is possible for perfection to be over 4000 Elo: I am very unsure about the actual level. My guess is around 4100 Elo. I am unsure also. 

If the fairly high ratio of draws to wins at the very highest level (eg 3:1) makes anyone feel chess is a win, first of all at the very highest levels--the cross tables i have seen put the draws about 8 or 9 to 1 over wins.  That does not make me feel chess is a win at all!

 

imagine an ending where white has a forced win but it requires great precision and where black can secure a draw if white goes astray at one of many stages. It is not inconceivable that chess itself is similar.  This is possible. not at all likely from the evidence i gave, but possible, and you refer to an ending not about the position from the start of the game. 

ponz111
Elroch wrote:  ponz in red

Obviously if white has a practical advantage, white giving up a small practical advantage will often leave the game drawn. very often not giving up the small practical advantage leaves the game drawn. And sometimes when the practical advantage is increased--the game is still drawn!

 

The player at a practical disadvantage may have more opportunities to make what appear to be modest errors that are crucial and lose.  This is obviously true...

Note that even thinking like this reveals very hazy knowledge of the truth:Thinking about what? Do you think i do not know what a practical advantage is? 

 

there are in truth no small advantages and the only mistakes are blunders that change the score. this sentence is rather ambiguous. I think you mean there is a difference between a practical advantage and a real, forcing, advantage. I agree that to lose a game one must make an error/you can call it "blunder" which changes the theoretical score of the game.  However in practical play, one can make a blunder and still win--happens all the time... My discussions are always about errors which change the theoretical result of the game.  I could go on and on about practical advantages. There are more to practical advantages than the positions on the chess board. I always play for a win even though i know that chess is a draw with best play.

 

All the rest is just practical considerations for players who can't tell which positions are winning (i..e. all of us).  Actually if you were to place 100 random positions at random on a chess board I would average over 97% knowing if the position was a win, loss, or a draw.  You could do this also.   

Elroch
[COMMENT DELETED]
Elroch

I am sure you realise that all discussion of practical advantages and imprecise play is irrelevant to the question. Only precise play and theoretical values matter. We don't have a precise grasp on these for all but a small fraction of positions.

USArmyParatrooper
Elroch wrote:

I am sure you realise that all discussion of practical advantages and imprecise play is irrelevant to the question. Only precise play and theoretical values matter. We don't have a precise grasp on these for all but a small fraction of positions.

Exactly! Perfect play is easy to establish with, say, mate in 3 on the board.

 

But after 1. e4 e5 2. Nf3 Nc6  - we have NO IDEA if any of the moves were perfect, nor the perfect response(s). We would have to exhaust an astronomically large number of game variations, far beyond the ability of any human or computer.

zborg
ponz111 wrote:

Lots of laughs--i think i am outnumbered here!

After 6000+ posts, only the zealots, or totally committed are still contributing.

"Healthy OCD" is another possible explanation for the continuing scrum.

 

At first blush, why focus of the "Perfect Game ??"  No one knows perfection, not even Quantum Mechanics, which is the closest we / science have gotten to it.  Unless, you're into String Theory, which many physicists  believe is all tied up in (unproveable) mathematical knots.

 

"Neither side has enough of an advantage to force a win" -- This is a much weaker (and pragmatic) proposition that makes it easier to sleep at night.  By extension,"perfect play" is a mirror image of String Theory, inside chess theory.  Hence, round and round we go.

Not that's there's anything wrong with that.  grin.png 

ponz111
USArmyParatrooper wrote:
Elroch wrote:

I am sure you realise that all discussion of practical advantages and imprecise play is irrelevant to the question. Only precise play and theoretical values matter. We don't have a precise grasp on these for all but a small fraction of positions.

Exactly! Perfect play is easy to establish with, say, mate in 3 on the board.

 

But after 1. e4 e5 2. Nf3 Nc6  - we have NO IDEA if any of the moves were perfect, nor the perfect response(s). We would have to exhaust an astronomically large number of game variations, far beyond the ability of any human or computer.

So you think 1.e4  e5  2. Nf3  Nc6  are perfect moves?

You forgot to include 3. c3!

Do you really think we must solve chess to know some perfect moves?

How did you know 1. e4  e5  2. Nf3  Nc6 are perfect moves?

Do you think because you do not know something that it is not possible for somebody else to know something?

[i am just kidding] [did you know that?]

USArmyParatrooper

ME: “Perfect play is easy to establish with, say, mate in 3 on the board.”

 

YOU: “Do you really think we must solve chess to know some perfect moves?”

🤪

At least try to have a legit conversation. 

 

ponz111
Elroch wrote:

I am sure you realise that all discussion of practical advantages and imprecise play is irrelevant to the question. Only precise play and theoretical values matter. We don't have a precise grasp on these for all but a small fraction of positions.

Of course we don't because there are so many positions in chess. However we do know a lot about chess even though we do not consider 99.99999999999999999999999999999999999% of the possible positions as most are not relevant anyway.

There are many grains of sand on this planet. We cannot view each grain of sand on this planet. However people who study sand know a heck of a lot about sand. 

I agree that in this discussion practical play does not matter to an enormous degree-it does to a slight degree as it tells us something about chess. It tells us as players become stronger--there are more and more draws. It also tells us that as players become stronger--there are more and more players who believe chess is a draw.

We don't have a precise grip on the properties of every grain of sand--some may not fit the common modes. But there are still people who study sand and know a heck of a lot about sand.

In the past there were not many sand experts. And of those sand experts--they did not know near as much about sand as the sand experts know now. As the sand experts progress and get stronger and more knowledgable--they know more and more about sand. There are many things sand experts can tell us about sand even though they do not have the ability to view every grain of sand.

This, by the way, is how much of our knowledge about things progresses and eventually gets to most people world wide.

Sand Experts can say they "know" certain things about sand even though some detractors will tell them "How can you "know" anything about sand when you have not even seen every grain of sand on earth???Undecided"

In past times people who claimed to know things which were not part of the general knowledge were severly chastized.  

Elroch

Have you an explanation for the fact that as players get stronger, there is a subtle tendency for white to get a higher score? Surely strong players should become better at achieving the theoretical result, at least on average, rather than being worse at it than weaker players?

You don't claim to know, you claim to have a very strong belief that you have suggested would justify odds of 1 to 9999 on.

USArmyParatrooper
Elroch wrote:

Have you an explanation for the fact that as players get stronger, there is a subtle tendency for white to get a higher score? Surely strong players should become better at achieving the theoretical result, at least on average, rather than being worse at it than weaker players?

You don't claim to know, you claim to have a very strong belief that you have suggested would justify odds of 1 to 9999 on.

 And in reality he does claim to know.  That’s just him hedging, on one hand claiming to know but on the other hand claiming he is not “absolutely certain“ about anything, giving himself both ways. But as a baseline, he says he is AS certain chess is a draw as he is the earth revolves around the sun, and AS certain 1+1=2.

 

 So setting aside him hedging with philosophical absolute certainty, in the colloquial sense he is claiming to be absolutely certain perfect chess is a draw.