Chess Will Never Be Solved. Why?

Yes - correct.  No reason.
Very neat.

#275
Sveshnikov > you

The way to know with absolute certainty is brute force solving, so your argument is circular...you cannot ever establish that a "chess AI" is 100% perfect until you solve the game.

This whole thread was done better the first time around...this is just redundant rehashing.

https://www.chess.com/forum/view/general/chess-will-never-be-solved-heres-why

#279
"me definitely > Shveshnikov" ++ That is hubris.
#278
"it's necessary to evaluate all positions reached with algorithms"
++ No not at all. You cannot evaluate a position with an algorithm. To evaluate a position calculation is the way.

Why would algorithms have to be exclusive of calculations?
But now another concession seems to have been made.
A 'node' is hardly 'calculations'.  
Actually - that looks like another contradiction too.
But illogic is often self-contradicting anyway.

Did Sveshnikov claim that a figure of 10^44 positions could be pre-reduced to 10^17 by arbitrary means ?
If he did that was not 'great'.
In other words an argument that only a 1000 trillion trillionth of positions would be 'sensible' ?  
Sveshnikov made this claim or one of his 'disciples' ?

I imagine if I check back in a week there'll be more spam that contradicts itself and spam about nodes and arbitrarily produced cosmetic numbers of positions.   The spam will be packaged differently though.

#281

"Did Sveshnikov claim that a figure of 10^44 positions could be pre-reduced to 10^17 by arbitrary means ?"
++ This is what Sveshnikov claimed:
"Chess is an exact mathematical problem. The solution comes from two sides: the opening and the endgame. The middlegame does not exist. The middlegame is a well-studied opening. An opening should result in an endgame.... Give me five years, good assistants and modern computers, and I will trace all variations from the opening towards tablebases and 'close' chess. I feel that power."
So he claimed 5 years, more than one computer and human assistants to prune irrelevant lines.
Per my calculations he was right and there are only 10^17 legal, sensible, reachable, and relevant positions.
It is pathetic, that weak players say: "Sveshnikov was wrong because I say so".

A simple "no" would have done.

#283
So a GM claimed something, calculations confirm it and a simple "no" by weak players suffices.

A GM is a weak player in terms of perfect play, as is an engine.

Calculations confirm it only if they're ludicrously incompetent calculations.

And no, Sveshnikov did not claim that a figure of 10^44 positions could be pre-reduced to 10^17 by arbitrary means.

#285
Sveshnikov was close to perfect an analyst.
On 'his' B33 variation he wrote: "By publishing a monograph on the 5...e5 system in 1988, I practically exhausted this variation."
Sveshnikov did not advocate arbitrary means: reducing the irrelevant lines is the task of his 'good assistants'.

     It seems to me there is a weakness in Sveshnikov's proposal: that we can just start where opening theory leaves off and work toward known endings. "Settled" opening evaluations are often later overturned and the search for new wrinkles is unending. Not to mention the many hardly-explored side lines. 

     Do we leave all these possibilities unexamined? Do we take today's top GMs opinions, backed by today's top engines, as gospel?  If so, why ask the experts which lines to analyze and just accept their opinion on the answer to the final question as fact?

     I posed it as a question as I don't know what exactly Sveshnikov means by "opening". It does seem that it must mean present-day theory, or else it wouldn't trim much to trim down the analysis tree.

There are more possibilities than there are atoms in the known universe. Maybe a quantum computer will analyse them all some 300 years from now. If it can do so to a depth of 150, I think we can safely say everything is solved.

Not at all.

We already know of positions that need a depth of well over 1000 with only seven men on the board (depth is usually expressed in plies). If Haworth's law were to hold up to 32 men that would suggest forced mates of depth between 10⁹ and 10¹⁰ with 32 men on the board.

It might be possible to weakly solve both basic or competition rules chess already if the starting position is say a mate in 20 moves we've all missed. But if the starting position is a draw, a depth of 150 would be negligible even for the competition rules game.

#293
"if the starting position is a draw, a depth of 150 would be negligible"
++ No, not at all. The starting position is a draw and ICCF WC correspondence games of 5 days / move with engines allowed last on average 40 moves before ending in a draw.

The long checkmates in the endgame table bases are in positions without pawns or with just one blocked pawn. At the start of the game there are 16 pawns and 16 other pieces.

#295
"Why would you think any of that has any relevance to solving chess? The majority of those games aren't even finished they're cut short by agreed draws, but even if they were finished and even if they used the same rules it would be irrelevant to the search depth you'd need."
++ 99% of ICCF WC draws are perfect play and thus relevant to solving chess.
74% are agreed draws, e.g. in endgames with opposite colored bishops known to be draws.
16% are draws by 3-fold repetition e.g. perpetual check.
10% are draws claimed by the 7-men endgame table base.
ICCF has the same rules, with the exception that it allows table base win claims that exceed 50 moves without pawn move or capture. In theory ICCF thus is more decisive. In practice such table base win claims never happen, while table base draw claims happen in 10% of games.
This is one more argument why the long checkmates play no role and the so called Haworth's law is no law at all. The long checkmates exist, but are never reached.

"Of those constructed so far that leaves only the six man tablebases where the longest mate is pawnless (and some reasons to think it may be the one and only)."

++ The longest mate in the 8 men table base under construction:

So for 6, 8 and presumably 10, 12... men the longest checkmates are pawnless.
Positions with an even number of men are more relevant than with an odd number of men.
The initial position has 32 men, that is even.
The approximate piece values N = B = 3P, R = 5P, Q = 9P are all odd in pawn units.
So only positions with an even number of men can be balanced in material.
All positions with an odd number of men are unbalanced in material by at least 1 pawn.

The prospect of solving chess in a brute force manner hinges on a breakthrough in scalability in quantum computing. This is an area where the physics itself isn't yet fully understood. My opinion is think we're more likely to uncover new physics which places a limit on how far quantum computing can practically scale than be able to access the computing power to be able to be completely certain that we have a solution to chess.

"The way to know with absolute certainty is brute force solving, so your argument is circular...you cannot ever establish that a "chess AI" is 100% perfect until you solve the game. This whole thread was done better the first time around...this is just redundant rehashing." To see why the reasoning I've given isn't circular, consider Fermat's Last Theorem. It was correctly assumed for a long time Fermat was correct however it took a very long time to prove that he was correct.