Chess Will Never Be Solved. Why?

Sort:
playerafar
tygxc wrote:

#459
"chess is nowhere near solved because its space of possible games far outstrips what could be dealt with by current computing technology"

++ There are far less chess positions than chess games. The number of legal, sensible, reachable, and relevant chess positions is 10^17. That takes 5 years on existing cloud engines.

As I expected - the nodes/square root spam is repeated again.
And like the other forum perhaps there's now another 500 posts coming up in the next few days with further repetitions of that spam and reactions to it.
Better to Break the Power of that spam but it might take several people knowledgeable enough about computer speed units to do so.
If you mention Floating point Operations per second at all it's like Candy to the person spamming ....
because he'll probably be very anxious to then spam (again) how much he and 'we' don't care about those units of computer hardware speed.

Computers can only do so many hardware operations per second.
'nodes per second' is a kind of scam that shoots down better conversations about the real daunting nature of the task.
So is 'taking the square root' of a number whose number of digits is in the middle 30's.
That means he arbitrarily wants to reduce the size of the task to a millionth of a trillionth of its actual size.
Its like an argument that the earth is flat.

playerafar

And there are more than four 'sensible' replies to e4 too.
More spam there.
Next - we'll see more posts with double crosses in them.
By July ...  grin 

not_cl0ud
playerafar wrote:


He'll be back probably.
He's been muted before.  

If he's been muted before, then he probably already used up his warning. But please, not public shaming.

Now back on track with our topic. happy.png

playerafar
ChessFlair01 wrote:
playerafar wrote:


He'll be back probably.
He's been muted before.  

If he's been muted before, then he probably already used up his warning. But please, not public shaming.

Now back on track with our topic.

I agree with 'not public shaming' including the namecalling that gets people muted.
Regarding 'back on track' it doesn't seem the conversation has ever been on track because there has not yet been proper establishment of units of computer hardware speed in the conversation.
To mathematically connect the topic to reality 
then 'nodes per second' needs to be dismissed and substituted for with real units of computer speed.
Then - at that point the number of operations needed to present a position properly and also consider its possibilities needs to be somehow factored in.
Upper and lower bounds for those numbers of operations need to be established.
Then - using known hardware speeds and dividing the time needed into the total task (not into a millionth of a trillionth of it) then upper and lower bounds for the time needed (in billions of years) can start to be established.

not_cl0ud
playerafar wrote:
ChessFlair01 wrote:
playerafar wrote:


He'll be back probably.
He's been muted before.  

If he's been muted before, then he probably already used up his warning. But please, not public shaming.

Now back on track with our topic.

I agree with 'not public shaming' including the namecalling that gets people muted.
Regarding 'back on track' it doesn't seem the conversation has ever been on track because there has not yet been proper establishment of units of computer hardware speed in the conversation.
To mathematically connect the topic to reality 
then 'nodes per second' needs to be dismissed and substituted for with real units of computer speed.
Then - at that point the number of operations needed to present a position properly and also consider its possibilities needs to be somehow factored in.
Upper and lower bounds for those numbers of operations need to be established.
Then - using known hardware speeds and dividing the time needed into the total task (not into a millionth of a trillionth of it) then upper and lower bounds for the time needed (in billions of years) can start to be established.

i literally dont understand XD

not_cl0ud

In practice, however, chess cannot be solved because it’s beyond the capacity of any human mathematicians and beyond the capabilities of computers. There are 10120 potential variations of games in chess and around 10^43 (1E43) different potential positions on the board. To fully solve chess, every single one of these must be compared against the other.

not_cl0ud

Whereas tic-tac-toe is solved thanks to a quite small space of possible games, chess is nowhere near solved because its space of possible games far outstrips what could be dealt with by current computing technology. As noted in another answer, endgame tablebases exhibit optimal play for all positions with limited numbers of pieces.

mpaetz

     "Never" and "cannot" and the like presuppose that computer technology will never achieve the kind of capacity and speed necessary and that no one will ever come up with new technique that will simplify the task. Considering the advances made since WWII I think "never" is too pessimistic.

not_cl0ud

I didn't note chess as never. I said in future's time.

DiogenesDue

I believe the last time I ran a back of the envelope calculation of this, it was would still take something like 3^24 years to solve chess even if you spent the entire world's wealth on nothing but the fastest 2020 era supercomputers ($80 trillion for a few hundred thousand of them) and set them on the task.  So...effectively never.  None of us here will be alive when it happens.  Nor our grandchildren, or their grandchildren.  Even with the most optimistic interpretations of Moore's Law, quantum computing advances, etc.

playerafar

"i literally dont understand XD"
@ChessFlair01   
Are you familiar with simple equations like speed x time = distance
?
For example - if your car travels at 60 kmh for two hours -
then it has travelled 120km.
Yes?   But also valid is Distance/time = Speed.
And Distance/speed = Time taken.

In this case 'Distance' refers to the size of the task - 
Speed refers to the speed of the computers.
If the computers were fast enough - chess could be 'solved' in five minutes.
But they're not.
Now if the terms upper bound and lower bound are not understood -
then yes that post wouldn't be understood.
But also wouldn't be understood if the Distance Time Speed equations aren't familiar enough.

tygxc

#487
"I ran a back of the envelope calculation"
++ Your envelope calculation is wrong.
10^44 legal positions, 10^32 sensible positions, 10^19 positions reachable during the search, 10^17 relevant positions leads to 5 years, corroborating the claim by the late GM Sveshnikov

"None of us here will be alive when it happens"
++ Maybe, maybe not.
It depends on the will to allocate 3 cloud engines and 3 (ICCF) (grand)masters during 5 years
None of us may be alive when humans walk on Mars, a matter of money.

Meanwhile partial solutions probably are already there, though not public.
Caruana and Nepo and their teams of cloud engines and grandmasters presumably already have largely solved C42 Petrov
Carlsen and his team of cloud engines and grandmasters presumably already have largely solved B33 Sveshnikov and C89 Marshall.

 

tygxc

#494
"there are between 10^111 and 10^123 positions "
++ No, not at all.
There are 10^44 legal positions, but the vast majority is not sensible, i.e. cannot occur in a game with >50% accuracy and thus not in a perfect game.
There are 10^37 positions without promotions to pieces not previously captured
https://arxiv.org/pdf/2112.09386.pdf

Most of these are not sensible either: cannot occur in a game with >50% accuracy and thus not in a perfect game.
That leaves an estimated 10^32 sensible positions that can occur in a reasonable game.
Of these about 10^19 are reachable in the course of the solution. For example when working on 1 e4, no positions with a white pawn on e2 are reachable.
Of these about 10^17 are relevant to the solution. For example once 1 e4 e5 is proven a draw, then it does not matter if 1 e4 c5 draws as well or not.

playerafar
DaMrBrandon wrote:

it is estimated there are between 10^111 and 10^123 positions in the game of chess, as a comparison, the number of atoms in the known universe is about 10^80. Decide for yourself

What you're talking about is permutations of move orders.
Not positions.
Unfortunately this gave a foothold to the other person to repeat more pseudoscience.
But that's okay.  Worse things happen.  grin

MARattigan
btickler wrote:

I believe the last time I ran a back of the envelope calculation of this, it was would still take something like 3^24 years to solve chess even if you spent the entire world's wealth on nothing but the fastest 2020 era supercomputers ($80 trillion for a few hundred thousand of them) and set them on the task.  So...effectively never.  None of us here will be alive when it happens.  Nor our grandchildren, or their grandchildren.  Even with the most optimistic interpretations of Moore's Law, quantum computing advances, etc.

That depends on

(a) whether chess is a win or a draw 

(b) whether a BFI algorithm (e.g. tablebase construction or forward search of some nature) is the method of solution

and 

(c) what you mean by "chess"

If chess happens to be a mate in 12 that everyone has overlooked then an alpha beta search with a zero leaf evaluation for anything but mates could produce a result in relatively short time  (whether 50 move and 3-fold repetition rules are included in the rules or not). If it's a mate in a trillion moves it would obviously take longer and it would only be mate in a version of chess that includes no 50 move rule.

If chess is draw then you're in for the long haul with a BFI approach in any version of chess because you have to show that no lines are forced wins.

First of all you need a soluble version of chess. The FIDE laws don't define a soluble version even under basic rules, because the order of the possible yields of the game are not defined. If a player resigns simultaneously with delivering checkmate, for example, both players win according to FIDE. How does that compare with a similar situation where the player delivering checkmate doesn't simultaneously resign and he then wins, but his opponent doesn't?

Before computers a BFI solution would not have been considered, but complete solutions were produced for many positions with fewer men on the board using human intelligence. There's nothing on the horizon that would extend the same approach to the starting position, but that doesn't mean it's impossible. A timescale would be anybody's guess. 

tygxc

#494

"If chess is draw then you're in for the long haul with a BFI approach because you have to show that no lines are forced wins." ++ That is correct, but it is possible to limit the lines considered using human or engine intelligence, i.e. lemmata derived from the Laws of Chess.

"Before computers a BFI solution would not have been considered, but complete solutions were produced for many positions with fewer men on the board using human intelligence."
++ Connect Four was solved in two ways: by Allen with brute force and by Allis using 7 rules.
http://www.informatik.uni-trier.de/~fernau/DSL0607/Masterthesis-Viergewinnt.pdf 
Chess will probably be solved by a combination of brute force and knowledge.

MARattigan
Optimissed wrote:
MARattigan wrote:

If a player resigns simultaneously with delivering checkmate, for example, both players win according to FIDE.

This is a situation, regarding which, one of two assessments must prevail. Either you're mistaken or FIDE is incompetent, for obvious reasons regarding game fixing. In any case, this situation, which you persist in bringing up, is external to the game of chess and so is irrelevant. There isn't an argument here. It has nothing to do with a solution of chess.
FIDE is unfortunately incompetent. The laws could be simply fixed to take the situation into account.

It's not external to the game of chess. It's quite possible - why do you say it's "external"? It just doesn't happen very often in practice.

The games described in the FIDE laws are not soluble. The game described in the Basic Rules section would be soluble with a simple fix, e.g. excising the agreed draw and resignation articles, or, for a practical fix, e.g. making those secondary to the results determined by moves. (The games described under Competition Rules would need quite a lot of assumptions to make them soluble.)


How does that compare with a similar situation where the player delivering checkmate doesn't simultaneously resign and he then wins, but his opponent doesn't?

Before computers a BFI solution would not have been considered, but complete solutions were produced for many positions with fewer men on the board using human intelligence. There's nothing on the horizon that would extend the same approach to the starting position, but that doesn't mean it's impossible. A timescale would be anybody's guess.

It occurs to me that if there is no forced win then a solution, in the context we're discussing here, is impossible. There is no forced win.

Under FIDE laws, as I said above, a solution is impossible, whether or not there is a forced win. 

If, for example, the FIDE basic rules with a fix as suggested above were taken to define "chess" then a solution would be possible (if not necessarily practicable) whether the basic rules were taken pre or post 2017, that is whether or not the 50 move and triple repetition rules were included.

That is irrespective of whether the starting position happens to be a win or a draw. A solution should determine moves to achieve the best result for the player with the solution whether that happens to be win or draw. There is no valid distinction to be made here.

E.g. the following position is a theoretical draw. White has twenty one possible moves but only two of them draw. If the position were reachable following some solution the solution must also produce at least one of those two moves. 

White to play, ply count 0

 

There is no forced win but that doesn't render a solution unnecessary or impossible.

There is a solution here.
 

 

 

not_cl0ud

What a long reply XD

ChessDude009

L UR LITERALLY JUST tryna promote your topic so it gets views

MARattigan
Optimissed wrote:
MARattigan wrote:

There is a solution here.
 

 

 

There isn't a solution to illogical conclusions caused by incompetence, except in rejecting them.

Possibly true.

Your contention,

"It occurs to me that if there is no forced win then a solution, in the context we're discussing here, is impossible. There is no forced win."

is an illogical conclusion caused by incompetence.

The solution I tried instead was to explain your error, but it doesn't appear to have been very effective.