Chess will never be solved, here's why

Sort:
sup96_squad2

snoozyman
The answer is 42
tygxc

#1196
Of course chess is a mathematical problem. Mathematicians solved checkers, connect four, nine men's morris, antichess and published their findings in mathematical journals.
http://magma.maths.usyd.edu.au/~watkins/LOSING_CHESS/ICGA2016.pdf 

Sveshnikov had a master degree in engineering and had started to prepare a PhD in engineering, but he gave that up to pursue a career in chess: as a player, as a theoretician and analyst, first without and later with engines. So Sveshnikov was formally trained in mathematics and his job was to analyse chess positions.

The "towards" means that weakly solving chess starts from the initial position or better the tabiya and calculates towards the 7-men endgame table base.
This contrast to strongly solving chess, starting from 2-men and then backwards towards 3 men, 4 men, 5 men, 6 men, 7 men, 8 men...

Sveshnikov claimed he could weakly solve chess in 5 years if given modern computers and good assistants. He also said how: from the opening towards the endgame table base.

Calculations show he was right: it is possible indeed.

So far I am the only one who has presented analysis with facts and figures here. All the others just express opinions.

MARattigan
Optimissed wrote:

... Indeed, if it continues on its current trajectory, I might have zero brain activity by then. ...

Well look on the bright side. At least you're expecting an upward trajectory.

playerafar

A lot of time and screen space could be wasted arguing over the semantics of the word 'mathematics'.
Including various 'proclamations' 'I am right-you are wrong' stances and similiar behaviours.
Reminder - the words are there to serve us - not us the words.
And to serve individuals as well as groups.
That is how and why languages have developed. 
That is how and why usage changes and is adapted to situations at the time.
If there's to be a major investment in semantics conflicts - that would be because whoever doesn't have a contribution to the main topic and its main peripheral topics.

playerafar

'Sveshnikov' ?  'Calculations show he was right: it is possible indeed.'
Well that does indeed look like a kind of authority/credentials backed claim ...
but the illogics in these claims has been constantly exposed here in this forum - beyond just 'opinions'.
(note - I don't care if there's no formal word 'illogics' - but others can get excited about that if they choose)

tygxc

#1204
So far nobody has put forward any valid argument against it.
10^36 legal and sensible positions would be needed for strongly solving chess to a 32-men table base.
For weakly solving chess i.e. finding an ideal game with proof that the moves are optimal like for checkers about the square root of these are relevant because pawn moves and captures are irreversible that leaves 10^18 relevant positions.
On 1 cloud engine of 10^9 nodes/seconds that is 10^9 seconds for weakly solving chess.
That is 2*10^6 seconds for weakly solving 1 of the 500 ECO codes i.e. 23 days.

playerafar
tygxc wrote:

#1204
So far nobody has put forward any valid argument against it.
10^36 legal and sensible positions would be needed for strongly solving chess.
For weakly solving chess like for checkers about the square root of these are relevant because pawn moves and captures are irreversible that leaves 10^18 relevant positions.
On 1 cloud engine of 10^9 nodes/seconds that is 10^9 seconds for weakly solving chess.
That is 2*10^6 seconds for weakly solving 1 of the 500 ECO codes i.e. 23 days.

People have refuted your arguments several times.
But that's okay.  This isn't a courthouse and we don't have to start splitting hairs over 'refute' and its meanings.
You'll have your interpretations of what the words mean.

Nobody can nor will prove anything to you that you don't want nor intend to have proven to you.

But what is actually happening is that the case for it taking billions of years to 'solve' is getting stronger and stronger here.
Just adding one piece to the 7 piece tablebase - makes it much harder -
and that 7-piece has been such a struggle that they couldn't even factor in castling and en passant.  
That's not an 'opinion'.  But you can believe it is if you want.
How many years before the 8 piece?  How many millenia for the 9-piece ?

Elroch

How many times do I need to say "starting with tabiya is definitely not adequate" before @tygxc can understand the point?

To elaborate, you need to deal with the positions reached after every crappy response by the opponent as well as the ones that look reasonable to a chess player. No number of decades of experience as a chess player stops this being a mathematical fact.

Let's be even more specific. Consider the strategy for white at move 1. Let's suppose, for example, it uses the move 1. Nf3.  Then it is necessary to deal with the 20 positions that black can legally reach with his first move. For example 1. Nf3 a5.  Let's say white's strategy includes 1. Nf3 a5 2. e4 at this point.  Then it needs to deal with all 21 positions black can reach on his next move.

Almost all the positions in an opening book for solving chess are seen to be those reached by what would be considered inferior choices. If it is the case that the value of chess is a draw, then some of these choices will be bad (turning a drawable position into a losing one), but this is irrelevant: a solution needs to deal with them to determine (i.e. not guess) their value.

It would be ridiculously optimistic to guess that this merely multiplies the number of positions that need to be dealt with by 1000, and if that were so an estimate of 5 years of cloud computing would turn into 5000 years.  The truth is surely a lot more.

playerafar

@tygxc has done a good job of pushing his points in a civil way.
And his posts have looked better than they are because of 'hard guy's' posts being so personal.
The terms 'soft guy' and 'hard guy' aren't in nor of themselves insulting.
Police for example - use soft guy/hard guy when questioning a suspect.
Its good police work.
But in this forum soft guy/hard guy has happened by incidence not design.
On the other hand terms like 'fools' and 'troll' are definitely insulting.
Even moreso when put with usernames.
'Somebody' finally/apparently managed to get through some of his posts consecutively without using them.  What could have brought that on?  Somebody else got his attention likely ...

And @Elroch is correct in his points just now.
And it is surely 'mathematical'.  And chess has certain mathematical elements.  As does hypothetically 'solving' it. 

If we in theory had some way to 'prove' that 1) d4 is a better move than 1) b4  (confidently thought to be the case but hardly 'proven')
that doesn't mean 1) b4 could nor can be 'skipped' ! happy.png
But many might not accept that. 
Or wouldn't accept it and hold 1) b4 as 'skippable'. 

Elroch

Yes, a strategy for black needs to cope with 20 first moves for white. And so on.

There is one hypothetical situation where you could ignore 1. b4. That is where there is a winning strategy for white that does not use this move.

tygxc

#1209
No, if black can draw against 1 e4 and 1 d4 then it is trivial to prove that black can draw against 1 a4 as well.

playerafar
Elroch wrote:

Yes, a strategy for black needs to cope with 20 first moves for white. And so on.

There is one hypothetical situation where you could ignore 1. b4. That is where there is a winning strategy for white that does not use this move.

in other words - if you had proven a win for one of the other white first moves.
But then - semantics might come in.
Or what definitions of 'solved' are deemed appropriate.

Some would argue that 'solved' is an abbreviations for 'totally solved'.
It looks like a good argument to me.
Because of the inane nature of the alternative.
Here's something 'solved' ...    1) f4 e6 2) g4 Qh4# checkmate.  !!
Its 'solved'.
Does that now mean that 'chess is solved'?  No.

'Solved' would mean 'totally solved'.
So ...  1) b4 cannot be skipped.  Nor can the Duras gambit nor the Englund gambit be skipped..

Is there any hope at all by 'adding forward moves' to initial positions ?
It just would make it harder.  Take longer.

But adding reverse capture moves or reverse other moves to solved endings would be not nearly as hard.  But still take billions of years.

Note:  Can all 'more pieces' positions be generated by adding reverse captures?  
Maybe not.  For example if the 'more pieces' position was already stalemate or checkmate - how are you going to 'reverse capture' back to it?  happy.pnggrin.png

To generate 'more pieces' positions - simply adding any legal piece to any empty square might not be valid.
And in instances where it would be - is it going to generate all of those 'more pieces' positions ?  
What if there were 'legal factors' blocking the addition or subtraction of pieces in both directions?
'Position generation with more pieces'  has issues.  
It might be necessary to show there's proper 'move communications' between the lesser and more pieces positions.  

playerafar

I'm thinking of positions where you can't get from more to lesser nor from lesser to more - because the capture isn't available in the first instance (usually most of the moves available in a position are not captures) and many positions don't have Any captures ! -
or because you can't get back to the more position legally by reverse capture - either because the capture isn't legal - or the reverse capture arrives at an illegal position - or gets back to a position already check or stale mate.  
'Piece addition' seems just as dubious if not more dubious.

'Total solving' might have to itemize all illegal positions ?

Elroch
playerafar wrote:
Elroch wrote:

Yes, a strategy for black needs to cope with 20 first moves for white. And so on.

There is one hypothetical situation where you could ignore 1. b4. That is where there is a winning strategy for white that does not use this move.

in other words - if you had proven a win for one of the other white first moves.
But then - semantics might come in.
Or what definitions of 'solved' are deemed appropriate.

Here it has generally meant "weakly solved" (according to the standard definition). This means you require a complete optimal strategy for each side (as achieved by Chinook for 8x8 checkers).

 

MARattigan
tygxc wrote:

#1209
No, if black can draw against 1 e4 and 1 d4 then it is trivial to prove that black can draw against 1 a4 as well.

It might be trivial to prove, but, whenever you're asked for a proof, your only response is to say it again. If it's trivial to prove can you post a valid proof please?

tygxc

#1214
I have not solved chess: that takes 5 years on 3 cloud engines.
It is evident that 1 e4 and 1 d4 are the most testing moves and to a lesser extent 1 c4 and 1 Nf3.
Once it is proven that black can force a draw against 1 e4 and 1 d4, then it is indeed trivial that black can draw against 1 a4 as well.
The same method as used for 1 e4 and 1 d4 can be used for 1 a4. There is no doubt at all that if black has a proven draw against 1 e4 and 1 d4, that black then has a forced draw against 1 a4 as well. It can then be proven by the same method as used for 1 e4 and 1 d4, but it is not even worth the engine time to do that as well then.

Elroch
tygxc wrote:

#1214
I have not solved chess: that takes 5 years on 3 cloud engines.

No. Greater than 5000 years. See discussion.

 

Elroch
MARattigan wrote:
tygxc wrote:

#1209
No, if black can draw against 1 e4 and 1 d4 then it is trivial to prove that black can draw against 1 a4 as well.

It might be trivial to prove,

Very generous! We both realise it is almost as hard as solving chess.

but, whenever you're asked for a proof, your only response is to say it again. If it's trivial to prove can you post a valid proof please?

Perhaps he is using the well-known proof techique "declaration of triviality"? As in "it is trivial to prove the Riemann hypothesis, but I will need someone to provide me with coffee for a year to do it".

 

tygxc

#1216
5 years on 3 cloud engines, 5000 years on 3 desktops.

See discussion: where? Who except me has calculated anything? Which post #?