Chess will never be solved, here's why

@6306

"disprove your own point" ++ No, proved my own point.

"Edwards, Jon (2525) vs. Miroslav Michálek (2480)" ++ This is clear human error.
Here is a more typical game: a draw in 35 moves, optimal play from both sides.
https://www.iccf.com/game?id=1164259 

How do you prove it is optimal play?

The well-known proof technique of hubristic proclamation?

@6308

"How do you prove it is optimal play?"
++ By statistics: > 99% sure to be optimal play with no errors by either side.

Nothing has ever been proved by statistics.

Rather statistics provides support for uncertain belief.

You acknowledge an example explicitly when you say:

"99% sure to be optimal play"

This fails to support your previous (still) unsubstantiated claim:

"a draw in 35 moves, optimal play from both sides".

I conclude you don't even know the difference between certainty and a belief state of 99% probability.

This seems a bit circular..... it's optimal if there are no errors. The definition of an error would be a move that is not optimal, right? It seems like you'd need to be more specific around the term "error" to make this concept useful.... maybe to the point of "in this position, any move other than xxxx is an error". 

Circularity is a key part of @tygxc's reasoning.

I'm coming to the belief that some people actually assert that these two are the same.... a belief state of 99% IS, in fact, proof / certainty. And that seems to be the essence of most of the disagreement. 

(I don't mean to make 99% a hard number......just an example of a high probability. Could be 99.99% or other.)

And in some cases, we might all agree. For example, "Yes, we all agree that it's certain. Smoking causes lung cancer".  Hasn't been "proven", but in this case, close enough. There's no point whatsoever in further debate. 

But some would say solving chess is not the same, and we are pursuing an absolute, mathematical proof, not one that we can all agree is "close enough" or "for all practical purposes".  

Like so often, this reveals that a large fraction of this debate is semantic. While several of the people here understand the accepted academic viewpoint about logical propositions and the reasoning that can prove them (eg the solution of checkers, the proof of the 4-colour theorem), those who "debate" against the consensus are using several of the same terms but incorrectly in crucial cases.

A "proof" to @tygxc is not a proof according to the standard meaning of the word. It's not that he is knowingly applying a different paradigm (including confusing different uses of the same terminology). Rather he gets confused and thinks he is advocating something of equal status to a rigorously proved result.

@6310

https://www.iccf.com/event?id=85042

136 games = 119 draws + 17 decisive games.
Assume chess a win.
Fit a Poisson distribution with 119 / 136 probability of an odd number of errors / game.
It is impossible, thus Chess is a draw.
Fit a Poisson distribution with 17 / 136 probability of an odd number of errors / game.
It is possible with average 0.143 error / game.
Games with 0 errors: 118
Games with 1 error: 17
Games with 2 errors: 1

Notice there is an anomaly in the games of SIM Bock, Steffen.
Strike the 16 games of SIM Bock Steffen.
120 games = 116 draws + 4 decisive games
Assume Chess a win, it is impossible to fit a Poisson distribution of errors / game with 116 / 120 probability of an odd number of errors / game. Thus Chess is a draw.
Fit a Poisson distribution with 4 / 120 probability of an odd number of errors / game.
It is possible with average 0.0345 error / game, i.e. 1 error / 29 games
Games with 0 errors: 116
Games with 1 error: 4
The above game is 96.6% certain to be an optimal game without any error by either side.

@6311

"more specific around the term error" 
++ An error (?) is a move that changes a drawn position a loss, or a win to a draw.
a blunder or double error (??) is a move that changes the game state from a win to a loss.

Intuitively, I think chess can be solved. My belief has nothing to do with human ability, the power of computers, present or future, or the time it might take.....or if it will ever be done. The question to me is simply, "Does a solution exist?".... not "Will we ever find the solution?". I think those are two very different questions...both interesting. 

It's hard for me to understand how there would not exist an algorithm that would guarantee at least a draw....just my thinking. I think it's unlikely that humans will ever devote the necessary resources to getting it done, but seems to my crude mind that the solution is out there waiting to be found......and again, that's just an opinion. 

Aside....I know (from experience) what an "error" is. But I don't think that's helpful if you can't pin down in every position which moves are errors and why. And the simple fact that "this move leads to defeat 98% of the time" isn't persuasive to me, although it might be to some.

Obfuscating, non-standard terminology.

Both are blunders. One is a whole point blunder, the other a half point blunder.

@6314

"there would be more than a 100 million move orders"
++ There are 10^44 legal positions, of which 10^17 are relevant to weakly solving chess.

"you have to know the best move against every move played"
++ Not the best move, but one of the good moves that are no errors

"even the best chess engine can’t play perfectly" ++ But when it runs for 17 s on a 10^9 nodes/s cloud engine, then a table base exact move is among the top 4 engine moves

"costs a fortune to use" ++ I estimate 3 million $ during 5 years

"would force mate in 279 moves" ++ No, Chess is a draw

"you could get to the same positions with dozens upon dozens of move orders"
++ That is why positions are key, not move orders

"yes, chess can be solved" ++ indeed

"nothing we have or can have, will be able to solve it"
++ 3 cloud engines of a billion nodes/s and 3 grandmasters can do it in 5 years.

This is not the kind of specificity I was thinking of.  I realize this is clear in your mind, but it's not clear in mine. I don't disagree with these "definitions" but they don't help with the fundamental questions I have. 

There is no doubt that chess with a choice of drawing rules is simple to represent mathematically. Technically, chess with a drawing rule that prevents infinite games falls into a different class of games to basic chess, but maths has no problem with either.

(The added definition that a theoretically drawn basic chess position is one where neither side can force a win brings it closer to finite versions of chess.  Note also that, unlike for many key notions in computer science, there is a terminating algorithm that determines the values of all basic chess positions with this added definition).

@6322

"if there are no mistakes, the game is a draw" ++ Yes, correct.

"between 0.3 and 0.6(for playing e4 at least), which is an indication of a draw" ++ Yes

"that advantage could grow exponentially as the game goes on" ++ As one side errs

"not even chess engines can calculate the best of the best moves"
++ There are only good moves, errors (?), and blunders (??)

"some trash opening" ++ There is knowledge. 1 a4 cannot be better than 1 e4 or 1 d4, 1 Nh3 cannot be better than 1 Nf3

"it would be some forced mate in 200 moves" ++ No, Chess is a draw

"we can’t prove that there’s a forced mate in 200 moves" ++ There is no such #200

"nothing can calculate move orders that far" ++ Yes, 10^9 nodes/s cloud engines can calculate from the initial position to a 7-men endgame table base draw or a prior 3-fod repetition draw.

To me, solving chess is purely mathematical with zero consideration of such things as confidence levels or statistical treatments. 
If someone "solves" chess to the point of being 99.99% confident, I'll bet on him every time, but I would not consider his solution a true solution.  In most cases, 99.99% is plenty good. In solving chess, it's "close but no cigar". 

But the reason I'm following is because I'm open to hearing arguments on both sides.....which I'm definitely hearing.