Chess will never be solved, here's why

@playerafar

I agree the 1/2-1/2 is a pain - even when you find where to fix it. If the setup program doesn't know the result then it shouldn't display a result. That's just bad design.

Regarding post #1723 - 
with lines involving a rook or a queen with different evaluations -
I've seen Stockfish change its mind about evaluations - if you give it enough time to run.  Sometime radically change its mind.
From loss to win or vice versa.  
Could there be discrepancies regarding numerical evaluations - because the runtime on one of the lines has been different?  

There is an oddity that if there is a result in a sideline it gets displayed as the result of the game. I often come across this posting analysis diagrams for vote chess, where you discount a sideline with a different result and then have to go to the tab to remove that from being displayed as the result for the whole diagram!

Open Source Chess Engine - Stockfish

@playerafar

Or just look at them, it's obvious what they do from looking. If you have a chess GUI (the thing that shows you the board) you can usually go into setup to look at the FEN at any stage and cancel out again. Some keep a running FEN on display as you play.

It's true that the bit about castling wouldn't take many examples to guess. The bit about e.p. and the two move clocks is best read rather than guessed, I feel.

[To change the subject, did @playerafar say he had remembered that problem position from 30 years earlier? Wish I had that good a memory!]

The first three points already established.
The truth of those points doesn't establish validity or practicality about reading all about Fen - I think you continue to miss the point about 'user friendly'.
If I want to talk about how gasoline automobiles are causing global warming - do I have to read up about how automobile transmissions work?  Or about how they make the alloys for the cars?

Much better:  present your points compactly -
so that all readers might get a logical picture - an overview.
An overview that pertains directly to the forum topic.
But that's my opinion and to each his own.
But neither your opinion nor mine is 'exclusive to others'. 

I didn't remember the position.  I remembered that it existed.
That's why I had to look it up.
Not surprised it was available.  Since its so Remarkable.

And we have yet to talk about why castling would be illegal in that position - as opposed to the very different issue of Stockfish regarding it as illegal.
The argument is that the bishop at b3 had to be a promoted piece - but that wouldn't be possible with castling still legal there.  

@playerafar

I think we got that far in the end. Very nice puzzle.

Now there's a post from the best poster here.
Interesting that you were able to find many instances of such puzzles where an illegality was 'deduced' and therefore a particular solution was available.
When I first saw that problem - the internet was hardly in full swing -
but now there's probably many computers even generating such positions.
Tons of positions that 'couldn't get there' even though they look legal.  Or where a move looks legal but isn't.  

@Optimissed

Clearly if you have neither a weak nor a strong solution you have nothing to distinguish between. 

That doesn't mean that the concepts are indistinguishable or the solutions if found are indistinguishable.

Well there's some more confirmation.  
But to get chess.com to fix it - that's a task in itself.
With so many members nowadays - and so much involved ...
perhaps the good old days when these things got directly in front of erik 
and fast
(Eric Allebest - the owner/founder/manager of chess.com)
are no more.  
The staff get millions of reports a week perchance ...
and there could be so many programming issues.
It took Years to transition chess.com's V2 interface to V3.
Literally years.  
Maybe - to make some changes - they'll have to wait for V4.  
I'm not a programmer - so I don't know how much code they'd have to rip up to make that 1/2 - 1/2 go away ...  or how much risk they'd have to take on creating other bugs by doing so.  

@playerafaer

I think that fix should be pretty trivial.

Sometimes though - there's a pleasant surprise.
It gets talked about - and then a few weeks later - they've at least partially fixed it !  

Me too.  But you would know better on that one.

#1710
"Hyatt (author of the Crafty engine) estimated at most 10% of the time was spent in generating new positions."
For playing chess on a desktop in a short time like 1 minute / move the engine cannot calculate all the way to hit the 7-men endgame table base and thus time has to be spent with the evaluation.
Analysing chess on a cloud engine of 10^9 nodes / second and with ample time like 60 hours / move allows to reach the 7-men endgame table base to retrieve the exact table base evaluation draw / win / loss and thus can do with a very basic evaluation function just to loosely guide the search.
In the TCEC superfinals Stockfish with a simpler evaluation function but deeper search defeated LC0 with a more elaborate evaluation function but more shallow search.

Even if the 10% rule were applicable, then we would end up at either 50 years, or 5 years with more cloud engines.

#1746
You cannot strongly solve chess by working from the 7-men endgame base towards the opening and generate a 32-men table base, as it would take 10^36 nanoseconds and 10^36 bits of storage.
The only feasible way is weakly solving chess by calculating from the opening towards the 7-men endgame table base, needing 10^17 nanoseconds.
Checkers and Losing Chess were also weakly solved from the opening towards the endgame table base.

@tygxc
You and I are in agreement that 'pure' solving of chess is just not feasible with today's computer hardware/software/programming.
Its just not strong enough.
And its good that that seems to be established through the forum for now.

"The only feasible way is weakly solving chess by calculating from the opening towards the 7-men endgame table base, needing 10^17 nanoseconds."
But 'weakly solving' is subject to a big range of definitions.
We could maybe argue that that's actually been done already.  
By both humans and computers.
If the required standard of 'weakly solving' is weak enough.

But I'm beginning to appreciate the difficulties of even just generating a proper 3-piece and 4-piece tablebases that links up properly with higher tablebases.

#1747
Definitions are clear:
"weakly solved means that for the initial position a strategy has been determined to achieve the game theoretic value against any opposition"
https://web.archive.org/web/20170912011410/https://pdfs.semanticscholar.org/8296/bc0ab855841088b31190c9f2923951853d7b.pdf 
The game theoretic value of chess is a draw.
Drawing for white is easier in chess than drawing for black due to the extra tempo.
A strategy for chess consists of a game tree.
Weakly solved for chess thus means that from the initial position for all (reasonable) white moves at least 1 black move has been found that draws.
That is also how Checkers and Losing Chess have been weakly solved.