Chess Will Never Be Solved. Why?

But we're not talking about this in practical terms. The question is will chess be 100% analyzed? Chess is a finite game, so eventually it will be solved. There's only so many positions that occur on the chessboard, and eventually computers will solve chess. That's how I see it. Seems pretty logical to me...

     You might want to check out the forum "Chess will never be solved" where computer scientists have been discussing the difficulties involved in actually achieving the solution.

We need math to solve this

Chess isn't infinite.  There's a finite number of positions.
Upper bounds on that number are known.  Well known.
They start with 13 multiplied by itself 63 times.
13 to the 64th power.
Because there are 64 squares and each square can only have a maximum of 13 states.   6 types of piece - but two colors.  Plus a square can be empty.
That number can be further reduced by 32 of the squares having to be definitely empty.  The math for that not heavyweight.
But such 'cutdowns' get more difficult - the more are attempted.
And eventually become illegitimate.  

@ Shrek - you're correct that the number of positions is finite.
'Eventually' - yes - but only with provisos.  That computers improve enough (and that's a Lot of improvement) and there's enough time and there's enough interest and money for same.
It also makes the assumption that something won't get in the way too.
A lot can happen in Ten Million Years ... or whatever.

#130
"the amount of analysis required to do so is virtually infinite"
++ No not at all. The number of legal chess positions is finite: 10^44.
The number of legal, and sensible positions is less than 10^38.
The number of legal, sensible, and reachable positions is a tiny fraction of that.
The number of legal, sensible, reachable, and relevant positions is a tiny fraction of that.
Sveshnikov said: "Give me five years, good assistants and modern computers, and I will trace all variations from the opening towards tablebases and 'close' chess."
3 cloud engines of 10^9 nodes/s can indeed weakly solve chess in 5 years, provided that the human assistants prune away the non relevant positions based on chess knowledge.

Again the spam about "10^9" nodes
But now something else coming out -
its 'five years' because Sveshnikov said so ??
so its also 'rumor based' as well as mathematically invalid in the first place.
With a circular argument also in -
that the 'tiny fraction' is known in advance - without proving what its value is is or even defining it at all.
Argument by assertion:  'tiny fraction'.
But no sign of the 'lets take the square root' creature yet ...
but if one listens carefully - is it audible in the distance -
getting closer ? 
Behind the trees or maybe over that rise ?  😁

#135

"the number of positions is irrelevant because each position then has to be analysed"
++ The number of legal, sensible, reachable, and relevant positions is the only relevant number in this discussion: it is the time in nanoseconds it takes on 1 cloud engine of 10^9 nodes/s.

"the only way to attempt it is to follow every possible game"
++ No, not at all: there are enormously more games than positions. Every position can be reached by more games than there are positions. Transpositions are a major feature.

"Doing it by following games is the only efficient method, because, among other good reasons, determining whether a position may be reached legitimately becomes unnecessary."
++ No, you do not need to determine legality. Just calculate forward and use transposition tables. That is also how Losing Chess has been solved.

"by far the greatest majority of positions is irrelevant to a well-played chess game and so is irrelevant to a "solution" for chess"
++ Yes, pruning is the key to make weakly solving chess feasible.
We know that 1 e4 e5 2 Ba6 loses for white, so it needs no analysis.
We know that 1 a4 is not superior to 1 e4 or 1 d4, so 1 a4 is not relevant.

"I detest the use of the descriptors "weak" and "strong". The words are jargon which is unfit for purpose, because it fails to easily convey the intended meaning."
++ It is easy:
Ultra-weakly solved means that the game-theoretic value of the initial position has been determined,
weakly solved means that for the initial position a strategy has been determined to achieve the game-theoretic value against any opposition, and
strongly solved is being used for a game for which such a strategy has been determined for all legal positions

"It's better to be more descriptive and use  words like "complete" and "reduced""
++ No, that would only confuse.

The number of possible chess positions is of course - not irrelevant.
But people will try to sidestep it in different ways.
And with varying rationales.

And - should anybody care about what den Herik's definitions of what weakly or strongly solved means ?
Or anybody's definition of 'solved' other than their own ?
Suggestion:  each individual decides for him/herself.
On each point.
Including about generic discussion versus brandname/credentials discussion.
The first upper bound of possible positions is known.
13 to the 64th power.  Well defined mathematically.
No need to quote experts.  Its straight math.
As chess evolved - did anybody care about 'total solving' ?
Unlikely.  It wasn't designed and further adapted with such in mind.
But now there's not only computers - there's chess software too.
Its big business.  
So projects like endgame tablebases are offshoots of that.

Chess has a game-tree complexity of about 10^120 but if you only account for moves that make sense then the number goes down to 10^40. Checkers has a square root number of positions as chess and has only been weakly solved. It will take a technological breakthrough to solve chess.

'Taking the square root' isn't legitimate - to reduce the number of possible chess positions.
The number of possible games (permutations of moves) results in an even more unmanageable number.

Regarding the number of possible chess positions (which is directly relevant) -
the larger the number - then when you 'take the square root' then that's a larger ratio of the positions.
Square root of 900 is 30.  So you're left with a number that's been cut down  by over 96%.
But the square root of 10^38th power is 10^19.
In other words you're left with a number that is far less than a trillionth of the previous number.
Its ridiculous.
How about we 'take the cube root' instead?  Or the 4th root?
In other words - as part of the sales pitch just simply cut the number down by whim - and then 'solve' in three weeks ?

Lets revolutionize space travel !  
The distance to the nearest star Alpha Centauri is over four light years away.
In other words about 25 trillion miles ...
But to make the trip easier -
lets take the Square Root of the distance !!

Then its only 5 million miles !
Our spacecraft can go 25,000 mph Easy !  
That would only be 200 hours.   !!!
A little over a week.  Want to go to the nearest star ?
Credit cards accepted.
Anytime you've got a difficult task that takes a long time - just take the square root of the time.
If something costs too much money - ask the seller to take the square root of the price ...

#142
"It will take a technological breakthrough to solve chess."
++ To strongly solve chess it takes a technological breakthrough e.g. in quantum computers to generate a full 32-men table base: from 7 to 8 to 9... to 32.

Weakly solving chess is already in reach of present computers.
3 cloud engines of 10^9 nodes/s can in 5 years assess
10^9/s/engine * 3 engines * 5 a * 365.25 d/a * 24 h/d * 3600 s/h = 10^17 positions
So the question is how many legal, sensible, reachable, and relevant positions there are.

but all the possible moves?

very long to calculate it..

"but all the possible moves?"
That's a different number - and even more unmanageable.
If every position was solved (hypothetically) - then why worry about all possible moves?
All possible moves is relevant - but if that was the approach from the opening position ...
then maybe the time it would take to 'solve' chess would have to be multiplied by an 80 digit number.   

For some reason that forum went quiet.
Maybe people began to realize that forum was dominated by months-long repetition of its second post.  

maybe

but to solve chess wont u have to know every move????

If you know every position why do you have to know every move ?

Example - endgame tablebase.
King and rook against King.
If the position is not stalemate - then does it have to be analyzed further ?
Its a win.  Done deal.

Try this:  Somebody plays their knight out on move 1 of a game ...
Nf3.  Now the opponent replies Nf6.
So then white puts the knight back at g1 and Black puts his knight back at g8.
Why would we have to know every one of such moves ?
That's part of the idea of positions instead.

#154

"This fixation with "assessing positions" .... how exactly is it going to be achieved?"
++ By calculation from the opening towards the 7-men endgame table base.

"Can your "cloud engine" just quickly assess a position as definitely winning, losing or drawing, just by looking at it for a split second?"
++ No, all existing or future evaluation functions are inherently flawed. The only way to reliably assess a position is by calculation towards the 7-men endgame table base.

"No calculation needed?" ++ No, on the contrary: much calculation needed from the opening towards the 7-men endgame table base.

"No exploring endless pathways?" ++ No, not endless, but still many.

"Why do you deliberately give out false information?"
++ No, I do not, I give out correct information. You seem unable or unwilling to understand it.

"I just proved that your method is impossible" ++ No, you did not. You apparently even misunderstood what I wrote many times. I should repeat some hoping more to sink it in.

"Do you understand what I wrote?"
++ Yes, I understand but you are wrong. You apparently do not understand what I wrote.

"Will you ignore it?" ++ No, I do not , but I probably should.

"Have you similarly ignored everyone else?"
++  No, I addresses all even remotely meaningful posts, but I probably should ignore more.

"Are you right?" ++ Yes, of course.