Chess will never be solved, here's why

Nice point about application to chess960, although this hasn't been a major priority. The point about the difference wrt to the 50 move rule is important. (Also means that application to chess960 relies on a similar 50 move rule!)

i agree

"(with the same caveats as the Lomonosov tablebases)."
So now we know - from the informative post by @MARattigan
that the Szygy tablebases are compromised in the way the Lomonosov tablebases are.
Regarding the 'impossible position' with a pawn unable to get to a square  
- I should point out that en passant can't occur anyway - when a pawn has been moved to its fifth rank ...  its the fourth rank - not fifth.
So perhaps the pieces concerned were moved up a rank by mistake somehow?

Wrong - because 'ceteris paribus' - the pawn so often does Not win.
Whether because its an a-pawn or h-pawn - or the other King gets the opposition or the pawn is 'caught' ....
The other King cannot 'catch up to' the rook.  The rook 'outruns' it. 
Opposition is going to mean something in K + R against K ?
A rook on an edge file can't win ?

Main point:  'ceteris paribus' - the plus of a rook wins whereas the plus of a knight or bishop doesn't.
Lasker didn't know what he was talking about ?
I think the term 'ceteris paribus' could be improved on though.

The logic isn't hard to follow about the nature of material advantage of a rook as opposed to of a minor piece.

Again - with e4 e5 Ba6 axb ....  its not 'solved'.

But maybe we could argue that a4 e6 Ra1 BxR b3 is 'solved' ...  because by b3 black could withdraw and save his bishop by putting it back on e7 or whatever and now black is a full rook up.  

Standards are needed to break the back of the task.
Arguing that a pawn up is enough to win - doesn't work for so many reasons.
The Benko Gambit 'loses' ?   

But in the opening rook-dropping example - 'white loses' seems very reasonable - because he has no 'compensation' for his rook.
But there - you run into a snag. 
'Compensation' could be far too subjective or poorly defined to argue 'solved'.

Even at the Grandmaster level. Black can still win. So it's not only white that has happy end games in chess. 

#1605
"the position after 1. e4 e5 2. f4 exf4 -  a true gambit - scores 49% for white"
In rapid or blitz games?
"It loses by force" - Fischer
"I found no way to equalise for white" - Kramnik
Gambits do occur in reasonable games with reasonable moves, but except Queen's Gambit which is no gambit not in ideal games with optimal moves.
"An endgame with an extra pawn is won, the plan is to queen the pawn;
an endgame with an extra knight is won, the plan is to trade the knight for a pawn"-Capablanca
"I felt honour-bound to take the pawn" - Capablanca on the very first Marshall Attack in history

#1610

"The Benko Gambit 'loses' ?" ++ It does. I would not play it in ICCF correspondence.    

"'Compensation' could be far too subjective or poorly defined to argue 'solved'."
++ Compensation = winning back the material within x moves.

#1611
"Even at the Grandmaster level. Black can still win."
Black can win if and only if white makes a mistake.
White can win if and only if black makes a mistake.
Humans make mistakes.
"To err is human." - Seneca

@tygxc
No comment there from you about the Benko gambit.

@PLAYERIII
You may want to factor in to your posts that each square can have up to 13 states.  Six piece types of each color - plus empty square.
But at the end - you're still ending up with a number of positions much too daunting for today's computers.

@Elroch - you keep missing the points made to you over and over again.
You've already conceded that chess is unlikely to be solved soon.
You might do better to understand your own arguments better.
You've now conceded that there's at least a 100 fold increase in difficulty each time a piece is added to the tablebases.
(I think its more than that ... with at least ten possible piece types to be added - and at least 32 square choices on which to add - but then there's the increased number of then-available options - plus the increased depth too.  At least Four major factors.) 
Also it continues to be that the tablebases are compromised by castling/en passant factors.
But you've missed another point too - you want to link positions of different material by moves ...   but moves lead to 'games' and games lead to powers of ten over 100 ...   numbers even more unmanageable - I think you already stated (and therefore conceded) this earlier in the conversation.
Simply marking posts in red with approval and disapproval and as a teacher would mark students' examinations - isn't proving.
Nor are the 'sly retorts' 
But go ahead.  Its harmless  !   The discussion can survive that.
If you want to do that and after I 'pass' and you arrange a PhD for me at Cambridge University - with 'accelerated tenure' - then I'll kick back 30% of my first year 400,000 pounds salary on their research team back to you !

Again - you also seem to not grasp the point that allowing positions that must be illegal by logic - in advance - isn't helping the computers do their job faster.
You also haven't addressed the point about number of promotions being tied to number of pieces onboard.
I believe you have the capacity to discuss that objectively.
Whether 'looking it up' or not.
If humanity had always been accepting somebody saying "this way - the way I accepted after 'looked up' - the 'doctrine' way .... is always the best and only good way"
then we'd all still be back in the Stone Age. 

@Elroch - looking things up is good.  Good for you.
You deserve credit for that.
But there's also going to websites and looking things 'over' as opposed to looking them 'up'.  
I go to websites too.  I pay attention to evidence - especially when its crushingly conclusive instead of just argumentative.
And to logical evidence.  But to use logical evidence properly - one must consider objectively.  Many people refuse to do that though.
Why?  Because its natural to 'pick a side'.  Its what people do. 
But that leads to situations where logical evidence is either rejected - or even worse - seen as not even existing.  Neither in the context at hand nor even existing at all in the universe even conceptually !  
Its dismissed with a verbal hand wave of some kind like:
"your logic - not mine".

Dismissing logic - and going by just argumentative evidence instead of crushingly conclusive evidence ...  and wanting to 'pick a side' (or also passivity and apathy) - often leads to the symptom. 
The 'doctrine approach'.  
////////////////////////////////////

s
#1613 is wrong on both counts. 
On the first point - by illogic of what 'I' would do.  
On the second point - by both incompleteness and inaccuracy.

////////////////////////////////
I wish I had more time right now to respond more to @MARattigan and his latest posts.   Including about the two knights each thing.
Usually his posts are the best posts here.

Regarding the projects with 'Szygy' and other such professional projects I'm confident they've considered most if not all of the points discussed here and far beyond that and much more accurately and thoroughly.
They would have much experience with the 'vicissitudes' of this thing.

But that doesn't mean we should go 'doctrine'.
Regarding the plus of a rook versus the plus of a piece we could discuss what is resignable. 
We could discuss as to where a tournament official would adjudicate a hopeless position.
The policy I used to hear - is that if one side wants a draw -
then the official would go by:  (what I remember)
'If in his opinion a master could not swindle a C player into blundering away the draw then the official would adjudicate the game as a draw'.  
Officials might also adjudicate if a player announces he has mate by force in a certain number of moves - and if the official agrees then the win is awarded.
But even a rook up?   Will the official award the win if there's still 'play in the position' ?   Yes I know it could be looked up.  For both over the board and online chess.   And for different Federations too.
Point:  finding better ways to go at the center of the task instead of trying to slice cheese off it.   

No. It was deliberate.

The only thing in the FIDE laws that mandates the pawns start on their second rank is art. 2.3.

The en passant arts. are 

3.7.4.1
A pawn occupying a square on the same rank as and on an adjacent file to an opponent’s pawn which has just advanced two squares in one move from its original square may capture this opponent’s pawn as though the latter had been moved only one square.
3.7.4.2
This capture is only legal on the move following this advance and is called an ‘en passant’ capture.

and don't specify the specific rank on which an e.p. capture may occur. So the position shown is illegal irrespective of arts. 2.1-2.3 as well as by art. 2.3. If the position were moved up a rank it would still be illegal irrespective of arts 2.1-2.3 (but then not also by art 2.3).

Well we're in agreement that its illegal.  Then.
And apparently doubly illegal.
But the illegality of the fifth rank there would make the presence and position of the knight irrelevant.
Because it would be illegal anyway.   

Because the double pawn move can only be made from the second rank - not from the third rank. 

White has enough material advantage to win here

Re the illegal position. My point was that such positions are the only positions that are illegal by the arts. excluding 2.1-2.3. You could say those positions are "intrinsically illegal" as you put it in an earlier post - the rest aren't.

A side can win because they have material disadvantage.

Say one side has only a knight.  The other side has much more material than that clustered around its King at h8.
Nf7#  Checkmate !   Wham !!  

So nobody wants to accept Lasker's point about a rook up?
That's okay.
I think most players would accept (whether openly accept or secretly accept) it wins unless the other side has 'compensation'.
Compensation doesn't have to take the form of recovering the material.
It can take the form of having a mating attack for example.
Or exposing the other side's King badly enough - that under tremendous pressure - the other side cannot find good enough moves fast enough and falls on time on the clock.  
'Compensation' could be - the rook-down side doesn't recover all of the material - but recovers enough material to draw.

@playerafar

#1620

But none of these concepts is well defined. They have use in practical chess , but don't translate to perfect chess.

Smothered mate (the position - not the sequence) looks Perfect.  To me.
Chess does allow for considerable perfection. 
That's part of its close relationship to mathematics.  

@MARattigan
We could have at least three categories of 'illegal' positions.
1)  Pre-illegal by logic.  The position need not even be generated in the first place.
For example - Kings adjacent.
2)  Intrinsically illegal but still needs to be generated to so determine.
Why?  Because not all pre-illegals could be logically pre-determined in a practical way.  So would need to be generated - and only then does the computer say ...  'No - already illegal on the board'.
3)  Illegal because it couldn't get there.  The worst category.  
I saw a problem once in the USCF chess magazine - where the solution was argued on the basis that castling couldn't be legal even though it looked legal while somehow it was argued that the King would have had to have moved earlier. 

Note:  illegal positions ...  versus illegal moves.
And - arriving at positions by computer generation rather than by moves.
'Moves' need only come in when the process has to go to the 'solving stage'.
And even there - might be 'solved' by relating them to checkmate/stalemate/hopeless draw positions and yes - to lesser material situations but only where that could be in the context.  
Relating positions - as opposed to generating positions by moves.

If anyone here solves chess it will surely be @playerafar.

@Elroch - such flattery ! 
But surely you see that @tygxc has far more optimism about the task -
he is likely to survive me - and be around to pursue it and therefore is  more likely to be 'the one'.  

Folks - in considering the nature of the forum topic and title -
the subject is actually broader than one might think ...
because the nature of chess itself is involved.
This could mean or does mean that the general subject of chess and most if not all of its aspects are potentially involved.
We're talking about solving chess - means we're talking about chess !  

And I found it ! -
a problem from Jason Rosenhouse in the 1990 Chess Life magazine.
Remembered it from 30 years ago.
The logic is somewhat intricate as to why black can't castle there. 
In this mate in two problem.
Chess Life, April 1990   Mate in Two
White to move.

Solution here and also remarkable explanation as to why castling has to be illegal in the diagrammed position !  
http://educ.jmu.edu/~rosenhjd/Problems/Direct_Mates.pdf
Can 'you' figure out without 'peeking' why that is ?
When I saw this over 30 years ago I was thinking -
"Whaaatt???  How could there be mate in two here???
Black would just castle long - and there's no mate at c7 because black's knight protects that square."
But black Can't Castle long!  It has to be Illegal ! 
So the mate in two stands up.

And - Stockfish seems to know that castling long would be illegal here -
can be seen by clicking the analysis button under the diagram board.