Chess will never be solved, here's why

Sort:
tygxc

#3516
Losing Chess has been rigorously solved considering 10^9 positions.
http://magma.maths.usyd.edu.au/~watkins/LOSING_CHESS/LCsolved.pdf 

playerafar


The person who spams 'nodes' couldn't even understand when he was shown a clearly drawing position that a strong engine evaluates as a win.
Its like showing pictures of the receding horizon to a flat earther.
Same kind of thing.  
With flat earthers - there appears to be a deep animosity to science ...
but also a contempt for those who recognize the realities of the horizon.
They are 'fools' to 'believe' that the world is round.
But in this very obscure form of 'flat earth math' concerning a chess subject ...  it doesn't seem to be about animosity to science and logic?
The contempt factor is there though.

Elroch
tygxc wrote:

#3516
Losing Chess has been rigorously solved considering 10^9 positions.
http://magma.maths.usyd.edu.au/~watkins/LOSING_CHESS/LCsolved.pdf 

Yes, it has. This was practical because losing chess is a tiny game compared with the major classic games - chess, shogi, go. Be aware that this required not ignoring possible moves against a strategy.

The same is true of the solution of checkers. The same would need to be true of a hypothetical solution of chess.

tygxc

#3520
So 5 years to convince a chess player and 500 years to convince a non chess player?

DiogenesDue
tygxc wrote:

#3520
So 5 years to convince a chess player and 500 years to convince a non chess player?

No, zero years and no evidence to convince one fanatic, unknown number of years to convince everyone else.

snoozyman

Chess will never be solved in our lifetime. 

Elroch

@Optimissed may well be right. Exponential growth may still give you an impractically large number even if you ignore all but the top 4 engine candidates (woefully inadequate as such an approach might be).

playerafar


"Yes, it has. This was practical because losing chess is a tiny game compared with the major classic games - chess, shogi, go."

Something wrong there because the word 'chess' appears in both parts of the sentence with the word 'compared' in between.

tygxc

#3525
The fallacy of exponential growth has been debunked before.
Chess has many transpositions, different move sequences leading to the same position, that vastly reduces the number.

playerafar


It doesn't reduce the number of positions At All.
Number of move sequences is different from number of positions.
But perhaps there would be 'cleverness' in confusing the two?
Magic?

Elroch
playerafar wrote:


"Yes, it has. This was practical because losing chess is a tiny game compared with the major classic games - chess, shogi, go."

Something wrong there because the word 'chess' appears in both parts of the sentence with the word 'compared' in between.

No.

"Losing chess" is a different game. Think of it as a compound word.

DiogenesDue
tygxc wrote:

#3525
The fallacy of exponential growth has been debunked before.
Chess has many transpositions, different move sequences leading to the same position, that vastly reduces the number.

No, it hasn't been debunked.  You have made unsuccessful claims about dropping 20-30 orders of magnitude, which nobody has ever agreed with other than yourself.  This would seem to be the opposite of the definition.

Vertwitch

then what to do after it is solved lol 

tygxc

#3533
"what to do after it is solved"
++ It will make no big difference.
People will avoid known losing lines which will narrow down defences for black,
but people will have a wider choice of equally drawish lines for white.
It is already partially the case now.
Carlsen, Nepo and their teams of grandmasters and cloud engines probably have largely solved the Petrov Defence, so Carlsen tried all different lines as white against them.
Carlsen and his team of grandmasters and cloud engines probably have largely solved the Sveshnikov and the Marshall, so Caruana and Nepo avoided the main line and tried early deviations.

tygxc

#3532
Of course it is fully refuted.
If chess had no transpositions then the number of positions with width w = 4 and depth d = 40 would be: 1 + w + w² + w³ + ... + w^d = (w^(d+1) - 1) / (w - 1)
If all move orders could be permuted then that  would be
1 + w + w²/2 + w³/3! + ... + w^d/d! ~= e^w
where d! = 1 * 2 * 3*...*d
and e= 2.718281812
The truth lies between both: chess has many transpositions, but no full permutations.

Elroch

w = 4??

Chess positions typically have 40 legal moves.

[Here I can save you blundering by saying some of them can be ignored. They can't unless based on an evaluation function that is so vast and sophisticated that it essentially includes a solution of chess].

tygxc

#3536
It is not about legal moves, but about reasonable moves that accomplish something.
The initial position has 20 legal moves, but only 4 stand out: 1 e4, 1 d4, 1 c4, 1 Nf3.
That is not a subjective evaluation, but it has been derived independently by AlphaZero with no other input but the Laws of Chess.
On 1 e4 there are 4 reasonable replies: 1...e5, 1...c5, 1...e6, 1...c6. Of those only 2 were played in the last Candidates', for good reason.
on 1 d4 there are 2 main replies: 1...d5, and 1...Nf6.
So there are far less reasonable moves that accomplish something than legal moves.
Carlsen said in an interview that he considers 3 candidate moves.

haiaku

A0 is a engine, not God. It didn't solve chess.

tygxc

#3538
AlphaZero acquired chess knowledge from no other input but the Laws of Chess.
This chess knowledge confirms the human knowledge accumulated over centuries.
https://arxiv.org/pdf/2111.09259.pdf 
That chess knowledge is beneficial to solve chess with the brute force method.
https://reader.elsevier.com/reader/sd/pii/S0004370201001527?token=7C9C654FA40A1A1AB270731F1FFE2C05BD558C23799135CC6B76E6FC6E5C383AE0A3FC09F492AF9B7C422E0E9D923A3E&originRegion=eu-west-1&originCreation=20220705135736 

haiaku

Humans are not God and they didn't solve chess yet.