Chess will never be solved, here's why

"the inclusion of the word "never" in the original question means that we are all making"
No it doesn't.   'we are all' doesn't follow.
Isn't it nearly always a mistake to try to speak for everybody?
People do try it.
And sometimes they get away with it - or even make the mistake 'work'.

But I think some progress has now been made - on at least two fronts.
1) the concession has now been made by somebody who's been pushing '5 years' that its not going to be 'tablebased' anytime soon.
I think its too late now - for that to be withdrawn.  Its conceded.  Finally.
Progress.
2) progress has been made on another front.  A certain 'ad hominem' phrase has been edited by a moderator.  

The 'real' discussion does go on - its 'hidden' to some extent inside a lot of other typing - but its there - every now and then.
Plus @btickler is correct.  Again.  
@MARattigan is also correct.  Again.  Subjective impositions of what 'solving' is - isn't going to get it.  

Many things still to be discussed.
There's a general unwillingness of the person most pushing a website's finding - to post here the times taken for each tablebase step from 1 to 7 pieces.
"The chain is as strong as its weakest link." 
This is often extremely true in mathematics.  
Weakest link.  
Those who choose to be lawyer-minded might do anything and everything to obfuscate the weakest link - seeing it as paramount that they do so.
They must convince the jury.  Its paramount to them.
Imperative.  
Math doesn't care about must.  Is there anything more objective than math ?
Death maybe. 

@Optimissed

Re #736

It wasn't intended as an ad hominem attack; rather as friendly advice.

Your post made it crystal clear that you don't even know what the question you've been discussing at such length means.

Since I know that the impression you make on other posters is your principal concern in posting, I thought I should suggest that you shut the f*** up before you made an even bigger ars*h*le of yourself. 

I thought it might embarrass you less if I left it at that.  

But anyway.

The idea of a strong solution in that sense is nonsense.

See the sequel.

Perfect moves are those that don't relinquish the result.

Correct so far.

They might be considered to be those that stand most chance of improving a result

They might, but that doesn't correspond with the definition of perfect move in the Wikipedia article. You would have known this had you read it and understood it (or if you had read and understood my correction to your definition of perfect move earlier in the thread).

but that doesn't make sense, since the algorithm may be playing against a similar algorithm. Equelly, if a human opponent, their effect can't be predicted. So -1 for Wiki so far.

If the algorithm produces perfect play from a position that is winning for one player it will enforce a loss for the other whatever moves he makes (even if these are according to the same algorithm). If it produces perfect play from a drawn position it will not lose if the other side employs the same algorithm.  

The Wikipedia definition of strong solution makes perfect sense for competition rules chess. It is inadequate for basic rules chess because, in requiring only perfect moves, not perfect play. It would allow algorithms that don't win from winning positions as strong solutions (see earlier posts).

But in that instance it's wrong not because of the reason you give.

Ther weak solution is, in effect, no different from the strong one. 

If you look at this position 

Many sources will give you a perfect (and perfectly accurate) account of how to mate as White from that position.

If you consider a version of chess restricted to the material shown, with a starting position as shown, these constitute a weak solution.

It's a mate in 20.

The longest mates in the endgame are 33 moves. These positions can all be reached from the position shown by imperfect play, but clearly the weak solution doesn't contain these positions because all the positions reached will be mates in 20 or less. A strong solution must  produce an optimal solution (not necessarily a win) from any position that can be reached from the position shown.

The ultra-weak one isn't a solution at all.

You don't get to know the moves, but it's enough to answer @ponz111's question, "Chess is a draw with best play, true or false?", for example.

In fact it's more than enough. Ponz's question is asking for what I might call an ultra-ulltra-weak solution in that if it's a win you don't need to say for which player.

It turned out that he really wanted an ultra-ultra-ultra-weak solution where you need only say chess is a draw.

...

Would you use two toddlers who play chess but can never checkmate each other as a proof that checkmate isn't possible in the game of chess?  Because that's the same mistake you are making by trying to use Stockfish or any other engine to guarantee perfect positional evaluations all the way until the tablebase.

All you are essentially attempting to prove is that Stockfish considers chess a draw...which is irrelevant since Stockfish improves measurably with every release...an indisputable proof that Stockfish cannot be relied upon to evaluate perfect play (except for the actual tablebase lookup positions, of course).  

The difference between weak and strong solutions for games is pretty clearly defined.  If you cannot understand the difference between the ability to guarantee/prove perfect play from the starting the position vs. being able to guarantee/prove perfect play from any position, then you cannot understand the difference between standard chess and Chess960, for example. 

It's kind of ironic, because a strong solution for chess is more likely by "cracking" the rules and proving the absolute value of pieces and positional considerations, rather than by brute force calculation...and since the former is your own professed position for the best way to solve chess...

The Wikipedia article (you cannot refer to Wikipedia as just "wiki", since there are innumerable Wiki's out there) is correct, but written at a high level without diving into a lot of specifics...which is what encyclopedia entries are for .

@btickler right yet again !  
Five years from now - supercomputers will be 'faster' than they are now.
Its something that constantly progresses.
Supercomputers evolve.  Something like laptops and desktops and phones do.
And upon that progress - some will be filled with much hope and again proclaim chess will be 'solved' in five years!
the wording of the thread topic is perhaps unfortunate - 
'never' is somewhat inappropriate?
It invites backlash.  But maybe in a constructive way.
Chess 'could' be solved.  But when?  Where?
Here on earth?  Only if society survives that long.  Millions of years.
Somebody already conceded - appearing to even suggest that the tablebases will never get to '32 pieces'.
I don't think its 'never'.  But maybe millions of years.  Or even billions of years.
But that could be revised as processing speeds increase and number of computers running in parallel for the task increases.
As for 'programming'  that can get misinterpreted.
Programming shouldn't mean imposition of subjective opinions about what 'solving' is.
Programming should mean better programs for tablebases.
To proceed faster. And while maintaining objectivity - not increasing subjectivity.
Issue:  subjectivity and perceived subjectivity are usually what people care about.  Usually - rightly so.  Usually.

Not in fact, I would say. @tygxc's proposal is not tablebasing.

Tablebasing could take longer on current technology and you don't have the space anyway. 

But I think in either case the human race might have evolved enough to find chess trivial before @tygxc's computation could complete on Summit machines (if indeed it could) and even before that people might be able to ask Siri to do it on their wristwatch. 

@MARattigan
I'm not referring to his 'proposal'.
I'm referring to his concessions.

"Tablebasing could take longer on current technology and you don't have the space anyway."
Of course tablebasing takes 'longer'.  Suggestion - this is what the topic is really about.  If tablebasing didn't take 'longer' - would this forum even exist ?

Care to define what 'the space' refers to ? 

Nalimov 3-5 man DTM tables on my desktop 7GB. Lomonosov 3-7 man tables (wherever they put them) 140TB. Continue the progression. 

So am I. He conceded only that tablebasing might need an extra desktop and cleaning woman. He conceded nothing in respect of his own proposal that I've seen.

No. It's just to identify the best moves. (Assuming you mean perfect play.)

I think he conceded well beyond that.
We could argue forever about 'what was said' and get nowhere.
I'm suggesting to you he's dismissed for himself publically tablebased solving of all chess positions.
That's a Huge concession.
The only thing that could maybe reverse it - is if he deletes all his posts so conceding.
But then - that still wouldn't be over.

@playerafar

#753

But I think he conceded that tablebasing is a non-starter from the outset.

The checkers paper he quotes and the method based on it that he proposes for solving chess are not tablebasing. I haven't seen anything where he concedes that his own method is a non starter.

" I haven't seen anything where he concedes that his own method is a non starter."
which I didn't claim.
But your comments about the size of memory that would be needed to solve for many pieces on the board look like icing on the cake to me.
Reinforcement of his concessions about tablebasing.
His silence on the matter is further concession too.

If you can't solve it - you can't solve it.  Simple as that.
7 pieces 'solved'.  More - no.
No estimates yet as to how long it would take to solve for 8.
No postings yet from the 'article' as to the time taken to get from 6 to 7.
The silence is telling.  Concession.  

This can be talked about as a lawyer thing - or objectively - or in many other variants.  Idea: Mathematics isn't a lawyer thing.  Not for this.
Nor is it a political thing ...
'Solved in five years !' is running for office?  What do the polls say?  
Nah ....  

Weak and strong refer to the *outcomes* of the solutions...not to the difficulty of achieving them.

A weak solution for chess will tell you if chess is a draw from the starting position and how to do it from that position.  An application of a strong solution for chess will "solve" any chess position given to it and tell you if it's a win, loss, or draw with best possible play.  This makes a strong solution more useful (and for games of all types, not just chess).

"It's 2/2/2222, and chess has finally been solved!  Who would have known that the Ponziani Opening of all openings would be a forced win?  What will this do for the chess world, FabianoC#23?"

"Well, of course, this does mean in the short term that everyone will at first try to play the Ponziani...but that is effectively meaningless, since a human being could not possibly reproduce the forced win..."

"Indeed...what does it mean to you personally?"

Weak solution:

"Well, my 7 cycle clone ancestor Fabiano Caruana always wanted to know if Carlsen vs. Naroditsky 2029 was actually a salvageable draw for Carlsen in the pivotal game 31...but sadly we still don't know..."

Strong solution:

"Well, my 7 cycle clone ancestor Fabiano Caruana always wanted to know if Carlsen vs. Naroditsky 2029 was actually a salvageable draw for Carlsen in the pivotal game 31...and now we know that Naroditsky's unfathomable Na1 was perfect and Carlsen did not blow the world championship by making a mistake...Naroditsky took it from him with the move of the century!"

The tablebasing is the strongest standard mentioned so far.
All lesser solvings are a progressive diminishment in the standards.
Can be compared to the highest standard.  
Right down to the very lowest ones ..
"its a draw - the game is a draw because 'we' say so" or
"people wouldn't play that"  looks like another bottom-feeding standard.
But people who have learned from experience with tactics problems - know about 'counter-intuitive' moves. 
And know about the blunders that 'dismissal' leads to. 

Could be the fault of the person refusing to grasp the logic too.  
Or instead.  Rather.  

@Optimissed

It would be certainly be possible to discern the difference between between weak and strong solutions in practice. In the KBNK example I posted for you, if you consult any weak solution about how you should play from a mate in 32 position the answer is "don't know". If you consult a strong solution it will give you a move.

Also the weak solutions are for one side. If the solution is not for you and you ask it for a move, you get "don't know" anyway. A strong solution gives both sides moves.

The Wiki article is simply defining various terms that are generally agreed among game theorists. Nothing more subtle. They allow you to have a conversation where everybody understands what the others have said.

"strong and weak 'solutions' " look like bait and switch.
Neither one is 'going to get it' anyway.
It looks like there's now a general concession that chess will never be table-base solved anytime soon.  
In other words - it will not be solved in the foreseeable future.  
That foreseeable future could change - if programming improves enough to either tablebase the task - or perform solving that's close enough.

From the opening - no moves - one position.
One ply - 20 positions.  Two ply - 400 positions.  
Then it starts getting much tougher.  
Seems though - if all possible positions could be itemized - then 'solving' could come in after that.  
Moves leading to positions instead of 'games'.  

But that looks even harder than table-basing from the endgame end.
Because its still depending on moves instead of positions.
For example - pieces could move backwards - even to their original squares.
F-bishops sometimes do that - after castling and the rook moves.
I've seen it in GM games.  even the c bishop could do it.  A lot of pieces could.  
Point:  Its going to be too much.   Too many moves to handle - too much multiplication.  

Solving chess is only feasible right now as weakly solving, just like was done for checkers (8*8 draughts).

Strongly solving chess i.e. a 32-men table base would require to visit all 10^37 positions. That would take 10^28 seconds on 1 cloud engine and would require at least 10^37 bit of storage assuming only 1 bit draw/no draw per position and a 1 to 1 relationship between natural numbers 1 to 10^37 and the 10^37 positions in the sense like Tromp did. That is not feasible with present technology.

Weakly solving chess allows to prune non relevant positions and visiting only the square root of the legal and sensible positions, like was done for checkers, to account for positions rendered irrelevant by each pawn move and each capture.

Weakly solving chess allows also to prune the opening variations needing evaluation. If the Berlin is proven to draw for black against 1 e4, then it is not  necessary to check if the Marshall, the Petrov, the Sveshnikov, the Najdorf, the French, the Caro-Kann draw as well or not. Hence only 19 ECO codes suffice instead of 200 ECO codes B00 to C99.

As said the Tromp count is way too high as it contains almost all insensible positions with multiple excess underpromotions that play no role in solving chess. Take a data base with 4 million games, that gives about 320 million sensible positions. Take a sample of 320 million positions as counted by Tromp, that leaves 95% illegal positions and at best 1 sensible position.

A legal position is a position that can result from the initial position by a proof game of legal moves. I tentatively propose a sensible position as a legal position where the proof game has e.g. an accuracy > 90% or an average centipawn loss < 10. None of the random samples by Tromp is sensible.

Even the Gourion paper gives too high an estimate: a random sample of 200 positions also contains many non sensible positions, mainly because of non sensible pawn structures like quadrupled pawns.

The 50-moves rule plays no role, forget it.

It is not necessary to double the number of Gourion positions. The side having the move can be inferred by counting the moves for both sides. Moreover any position with black to move can be converted to a diagram with white to move by switching colors, as endgame books conventionally do.

It is possible to cut the Gourion number in 2. As soon as both sides have lost castling rights, most often because both sides have castled, as is most often in 26-men tabiya, there is left/right symmetry. All positions with a white king on the left half of the board can be converted to an equivalent position with the white king on the right side of the board by switching wings, as endgame books conventionally do.

For all Gourion positions without any pawns, the number can be cut in almost 8 by applying symmetry. The position can be converted to an equivalent one with the white king confined to the triangle e1-e4-h1. The syzygy table base does that.

As to Stockfish against Stockfish at 60 h / move is like letting 2 toddlers play: If the calculation reaches a table base draw, then that means all moves of the black toddler were in retrospect good enough to hold the draw. As for the white toddler: we grant him takebacks starting with his last move and working backwards. That ascertains all his reasonable moves lead to no more than a table base draw.

The table base decides draw or not, not the evaluation function of Stockfish. The evaluation function of Stockfish merely serves to guide the search in a meaningful direction.