As I said before. Post your 10/10 wins against SF15. You should have no difficulty if there's no uncertainty.
Chess will never be solved, here's why
Sort:
Dec 10, 2022
0
#4964
Another example of a very small probability that is not zero is the probability of winning against a much better chess player. Say one of us against the latest Stockfish.
This can be intuitively seen to be true by considering a deeper evaluation of the moves played by the two players and observing that there is a finite probability on two consecutive half moves that the weaker player plays a better move than the stronger player. With a reasonable assumption about some degree of independence, this means there is a finite probability of this happening on enough consecutive moves to achieve the victory.
The finite probability involved for this extreme reverse domination is so absurdly small as to be safe to believe it would not happen in practice, but no-one with an understanding of probability could seriously deny it is not a strictly positive probability. (It is of course not the only way a win could be secured, but emulates a typical theme of mathematical proofs of finding an easy route to the conclusion rather than dealing with all the inessential details).
Dec 10, 2022
0
#4965
@Optimissed
You've no doubt misunderstood something.
People generally mean different things by words according to context. You use the word "know" when "guess" would be the appropriate word in the context of solving chess, so in common with @tygxc fall into @Elroch's "fallacy of proof by (repeated) assertion".
SF doesn't claim to know the result after 1.e4 e5 2.Ba6; it gives an evaluation.
That could be loosely interpreted as a degree of confidence.
In that case it is more confident of winning this (drawn) position.
So obviously SF's "degree of confidence" is not knowledge.
You claim absolute knowledge of the correct evaluation of positions such as the first (by assertion). If your use of the word "knowledge" is in any way useful in this context then you should be able to crap all over SF (or indeed anything). But you always seem reluctant to demonstrate.
Dec 11, 2022
0
#4967
Anyone truly smart would not have spent over 300 forum pages arguing about this topic. That would make almost everyone on chess.com MORE than your intellectual equals.
Dec 11, 2022
0
#4968
@Optimissed And yet you are still here and adding to the pile. It seems you love arguing about pointless things more than chess. Perhaps a debate team reject? Also you are apparently not so good at math as you think, as you joined this pile of c-r-a-p thread on page 2. Since this is page 328, you have been a contributor to this nonsense for 327 pages. You must be single, possibly divorced, as no wife would put up with you hyper-narcissism. You are so in love with yourself and your own opinions, there could be no room for anyone else. Then there's this gem:
0
#6552I'm not going to argue with you; but to explain something. Stockfish is designed to "explain" analysis wrt degrees of confidence and is completely incapable of making judgements unless they're programmed ones. That makes it different from humans. It does not think: it calculates.
It's your job to interpret the calculations. If you are not capable of making a judgement where the judgement ought to be clear, that's your failing, however good you believe you may be at chess. There are different positions, where a definitive judgement may not be possible but this is not one of them. Another type of judgement is used to discern between what we can be sure of and what we cannot. That doesn't function well in very many people.
You say you are not going to argue, then in the same breath start arguing. You can't even be honest with yourself. My advice is to seek psychiatric help, and I'm pretty sure it's not the first time that has been said to you.
Dec 11, 2022
0
#4969
DesperateKingWalk wrote:
...
Stockfish would know this position is a 100% win if Stockfish could show a mate score.
Actually not even then.
It can show a mate score then change its mind to +/-152.xx on occasion.
Dec 11, 2022
0
#4970
Interesting idea. I've only seen it happen without tablebase access (but then I don't have Syzygy tablebases so I wouldn't see it any other way).
So far as I know SF uses only Syzygy tablebases and they don't give a mate length.
Why don't they announce "M?" instead of "152.xx"?
Dec 11, 2022
0
#4971
Some engines come to the conclusion there is a mate based on incomplete analysis. Sometimes the incompleteness turns out to be significant later, as the mate gets at least pushed beyond the horizon.
Dec 11, 2022
0
#4972
Optimissed wrote:
Put it this way. It isn't a sacrifice because it doesn't achieve anything in return for the bishop.
To be accurate:
(a) It's a sacrifice because Black can take the bishop.
(b) You can't see anything that is achieved in return for the bishop so you're willing to use proof by assertion to say it's fact.
(c) You can't prove the sacrifice is not perfect whether or not it achieves anything in return. You can't prove that the position is not, for example, drawn before and after the move (except by assertion, of course).
@6599
"Post your 10/10 wins against SF15. You should have no difficulty if there's no uncertainty."
++ This shows a complete lack of understanding. A position being analytically drawn, won, or lost has nothing to do with that position being empirically drawn, won, or lost in a contest between two human or engine players.
1 e4 e5 2 Ba6? is analytically lost for white. It is a checkmate in 53. That does not mean a human cannot draw or even lose it as black against an engine, just like a human may even lose a game with queen odds against a grandmaster.
1 e4 e5 2 Ba6? loses. That is 100% certain and not 99.99% certain. Probalility is inappropriate for deterministic events. You can ask about the probability that player A wins position X against player B. That is empirical, not analytical. Probability is always linked to an experiment, i.e. empirical. The experiment can be coin tosses, or detecting radioactive decays, or playing chess games.
A player may know that KBN vs. K is a forced checkmate in less than 50 moves. That is 100% certain. He may however fail to do so in a game against an engine or human. His inability to checkmate does not make it less than 100% certain that the position is a forced checkmate in less than 50 moves.
Chess
will never be solved
@6630
is five years away from being weakly solved.
Dec 12, 2022
0
#4978
tygxc wrote:
-
@6599
"Post your 10/10 wins against SF15. You should have no difficulty if there's no uncertainty."
++ This shows a complete lack of understanding. A position being analytically drawn, won, or lost has nothing to do with that position being empirically drawn, won, or lost in a contest between two human or engine players.Totally agree with the second sentence. That's why neither you nor @Optimissed will venture to demonstrate against SF - you don't have an analysis.
1 e4 e5 2 Ba6? is analytically lost for white. It is a checkmate in 53.
Yes - a joke.
That does not mean a human cannot draw or even lose it as black against an engine, just like a human may even lose a game with queen odds against a grandmaster.
Another one you haven't analysed.
1 e4 e5 2 Ba6? loses. That is 100% certain and not 99.99% certain.
Or 100% certain that it doesn't lose. You don't know which. Like @Optimissed you guess. So do I, but I don't claim to know on that basis. Neither have I attempted to assign a probability to any outcome.
Probalility is inappropriate for deterministic events.
Not if the calculation of the outcome is impracticable owing to lack of sufficiently accurate measurement of the factors that determine the event or insufficient resources or lack of a suitable method to calculate the outcome. It's widely used and useful .
But what is the relevance? Both you and @Optimissed are claiming certainty (without anything to back up the claim). Which point of view are you trying to argue?
You can ask about the probability that player A wins position X against player B. That is empirical, not analytical. Probability is always linked to an experiment, i.e. empirical. The experiment can be coin tosses, or detecting radioactive decays, or playing chess games.
A player may know that KBN vs. K is a forced checkmate in less than 50 moves. That is 100% certain. He may however fail to do so in a game against an engine or human. His inability to checkmate does not make it less than 100% certain that the position is a forced checkmate in less than 50 moves.
More of what passes for analysis in your eyes.
Players don't know that KBN vs. K is a forced checkmate in less than 50 moves. They may believe it because they've been told in the same way that they "know" there was a battle in the vicinity of Hastings, England in 1066 A.D. That's not analytical knowledge and useless as a guide to playing the endgame.
A correct analysis shows it isn't necessarily so. Mostly false if the 50 move and triple repetition rules are in force. E.g.
White to play, any ply count > 39Black ro play
Black to play
Black to play
Black to play
Either side to play
Either side to play
Nobody has ever published a complete analysis of this endgame for positions in which there have been prior repeated positions considered the same for the purposes of FIDE art.9.2 with the same material and I don't expect any soon. (KRvK might be feasible.)
As for 1.e4 e5 2.Ba6 you have got just as far with solving that as you have with solving the starting position, Nowhere.
Dec 12, 2022
0
#4979
Optimissed wrote:
It should be obvious that tygxc is his own worst enemy, posting things like "mate in 52", which is very difficult to support. However, that doesn't alter the real thrust of the comment, which is that a competent person would know when analysis is necessary and when it isn't.
There is currently no way of judging the absolute competence of players, only competence relative to other players. Except (partially) when players play positions with material covered by tablebases or admit a practicable analysis. The exceptional case would suggest there are no competent players and never have been.
Actually he said 53, but I think he's also said 82.
Dec 12, 2022
0
#4980
I agree that computers are pretty irrelevant apart from speeding things up a bit (a bit in this context). If chess ever gets a practical solution (it's been solved according to @tygxc's definition probably since it was first thought of) I think human intelligence will play a far larger rôle than Superficial Intelligence.
The difference between you and @Elroch and I is not about computers. It's about what would constitute a solution and the nature of any human thought that would arrive at one (though computers could be necessary to effect the arrival). In simple terms we don't count guessing as a solution.
Let me introduce you to the well-known fallacy of "proof by assertion" (also "proof by repeated assertion"). You are a keen user of this so I am sure you will be pleased to now know the correct term for what you do.