Chess will never be solved, here's why

You also still haven’t addressed the fact that I have objectively proven that the strategy stealing method cannot work. 

 This is sort of what I was talking about with the idea of the chess engines themselves not necessarily being programmed correctly. 

Some things are counterintuitive. 

@8759

"how come schaeffer didn’t evaluate the square root of positions?"
He evaluated 10^14 relevant positions of the 5*10^20 legal positions.
Checkers is no Chess.
A Checkers board is more crowded: 24 men on 32 squares as opposed to 32 men on 64 squares.
Checkers has only 2 kinds of men, Chess 6.
A Checkers board has more edge effects: 16 of 32 squares on the edge as opposed to 30 of 64.

Never, as storage sits currently (or with any serious/reasonable predicted advances).  There's not enough matter in our solar system to do the job.  So unless you are going to invent FTL travel before you solve chess...

@8761

"How do you determine which move by white to do?"
++ At first sight all legal white moves and 1 black response.
On closer inspection some clearly wrong moves like 1 e4 e5 2 Ba6? need no inspection.
The idea is to take the top w white engine moves, e.g. w = 4.
As proven the table base exact move is among the top 4 engine moves when running 17 s on a 10^9 nodes/s engine with 1 error in 10^20 positions.

@87660

"10^34 / 44 is the search space"
++ How do you arrive at that?

It’s short hand for 10^34 (your claimed number) OR the established 10^44

Wdym “as proven” 

you LITERALLY ASSUMED A FALSE ERROR DISTRIBUTION

also, for that 17s running, it would take over a million years to calculate 10^17 positions.

your nodes/sec is what is required to calculate a position with that accuracy every 17 seconds

If you are running 17s then BY DEFINITION it would take 

Your 10^17 assumes w= 1.  In fact, if w = 4, you would have to calculate at least 10^29 positions (assume game length of 40 moves.

@8768
"It’s short hand for 10^34 (your claimed number) OR the established 10^44"
++ No. the 10^44 or 10^38 or 10^34 is the total number of positions: the outer boundary in Schaeffer's figure number of positions (logarithmic).
ICCF WC draws can serve as seeded lines.
The stored boundary and the relevant search space shrink with each pawn move and each capture. For example after 1 e4 e5 all positions with a black pawn on e7 are no longer reachable.

@8769

"Your 10^17 assumes w= 1"
++ No not at all. E.g w = 4. w is the number of white moves that do not transpose.
If we look at all white moves and all black moves then N = w^(2d).
If we look at all white moves and 1 black response then w^d = Sqrt (N)

Ofc the initial search space is 10^3/4 4, that’s the collection of path trees for the game of chess.  That’s how it is by definition.

“If we look at all white moves and 1 black response then w^d = Sqrt (N)”

But you still haven’t addresssed how you find that black move.  You claim that it’s within the top 4 engine moves.  That means nothing.  

@8769

"if w = 4, you would have to calculate at least 10^29 positions (assume game length of 40 moves"
4^40 = 10^24.
4^80 = 10^48.
This proves that the 4 or the 40 are too high.
There can be no more positions than there are legal positions.

@8773

"But you still haven’t addresssed how you find that black move."
++ For black take the top 1 engine move. For white consider the top 4 engine moves.
For black do not worry if the move is exact or not: once the 7-men endgame table base draw is reached, that retrospectively validates all black moves.
For white the concern is that the table base exact move must be among  the top w engine moves considered.
For w = 4, 10^19 nodes/s, 17 s/move the table base exact move is among the top 4 engine moves with 1 error in 10^20 positions, i.e. no error for 10^17 relevant positions.

@8772

"the initial search space is 10^3/4 4"
++ No. The 10^44 or 10^38 or 10^34 is the total number of positions, corresponding to the 5*10^20 of Schaeffer, the outer boundary of his Figure of Number of Positions (logarithmic).
Within that outer boundary lies his relevant rearch space.
Within that relevant search space lies his stored boundary along his seeded line.

“For black take the top 1 engine move”

That engine move is often wrong. 
“For black do not worry if the move is exact or not: once the 7-men endgame table base draw is reached, that retrospectively validates all black moves“

SO YOU LITERALLY ASSUME THAT THE BLACK MOVE IS GOING TO LEAD TO A DRAW AS PROOF THAT THE BLACK MOVE IS GOING TO LEAD TO A DRAW.

the top engine move has a very real possibility of leading to a table base loss.

secondly “For white consider the top 4 engine moves.”

By definition this is no longer a proof.

you have to consider EVERY move.  Regardless of whether it’s probable or not.

“ 17 s/move the table base exact move is among the top 4 engine moves with 1 error in 10^20 positions,“

this has not been proven.  You made that claim off of 2 data points that DID NOT MEASURE ERRORS SO IT WAS PHYSICALLY IMPOSSIBLE TO ACCURATELY CLAIM A NUMBER OF ERRORS.

You also completely ignored where I pointed out that that alpha zero game database used human-set openings.

@8777

"That engine move is often wrong."
++ At 17 s/move on 10^9 nodes/s the top 1 black move is wrong in 1 position out of 100,000. However, do not worry about that. If the calculation ends in a 7-men endgame table base draw against all white opposition, then that retroactively validates all black moves as right.
If some black move is wrong, then no 7-men endgame table base draw can be reached against all white opposition and the error will show.

"ASSUME THAT THE BLACK MOVE IS GOING TO LEAD TO A DRAW AS PROOF THAT THE BLACK MOVE IS GOING TO LEAD TO A DRAW"
++ No. It does not matter where the black move comes from: a good engine, a bad engine or even a random generator. If some black move is wrong, then no 7-men endgame table base draw is reached and a black move needs retracting to correct the error.
A good engine with more time/move makes less errors than a bad engine with less time/move and than a random generator. A good engine does the job faster than a bad engine or a random generator as the latter need more retractions.

"the top engine move has a very real possibility of leading to a table base loss."
++ Yes 1 position in 100,000. Then a black move needs retracting.

"By definition this is no longer a proof."
++ It is a best first heuristic. If the good moves cannot win for white, then the bad moves cannot win either. If 1 e4 e5 2 Nf3 cannot win for white, then 1 e4 e5 2 Ba6? cannot win either.

"you have to consider EVERY move" ++ Every move that opposes to the draw. We can discard 1 e4 e5 2 Ba6? right away. No need to burn engine time on what we already know.

"You made that claim off of 2 data points that DID NOT MEASURE ERRORS SO IT WAS PHYSICALLY IMPOSSIBLE TO ACCURATELY CLAIM A NUMBER OF ERRORS."
++ 3 data points really, the 3rd is infinite time, 0 decisive games, making the extrapolation effectively an interpolation between 3 data points. The errors follow from the number of decisive games: each decisive game must contain an odd number of errors.
As the number of decisive games shrinks to zero with more time/move, so does the number of errors/game shrink to zero with more time/move.

@8778
"alpha zero game database used human-set openings"
++ Where in Figure 2 did you read that?