Will computers ever solve chess?

I don’t think it’s likely, but more importantly,  it’s impossible for anyone to know. 

 Perfect game would be game with no errors of any kind.  I’m not referring to errors as evualted by humans or current engines, no errors period.

That’s not what I kept telling you you’re wrong about. As you well know, the disagreement is your claim you can KNOW if a perfect game was ever played.

Show me a perfect game.

The term "errors" is ambiguous. A perfect game is a game where neither side makes a mistake that would change the theoretical outcome of the game.

Or to use the language of godofhell  A perfect game is where there is a never changing true evaluation. [ it will stay a draw if a draw is the result of a perfect game or it will stay a win if a win is the  result of a perfect game]

And the term "to know" is somewhat ambiguous. Many people if they are 99.99% sure of something they believe  they "know it" For many people --they are not 100% sure of anything--they could be a figment of someone's mind for example.

I am 99.99% sure that a perfect game will be a draw.

I am 99.99% sure that perfect games have been played according to the rules of chess.

By the way, early draws can be perfect games because in many early draws neither side made a mistake which would change the true evaluation of the game.

However even without an early draw, I think many perfect games have been played. I would guess several of the draw games between Stockfish and Alpha Zero were perfect games. 

Why do i thing this? Because i have a ton of evidence to support my statement.  Most of the evidence has not even been brought up here. 

Regarding my statement that there have been games that i have played perfectly--here is one example of a game I played perfectly:

Here is a game [quick play game] where my opponent played perfectly

I had Black...

Humans can certainly know If SOME specific moves are perfect, such as mate in 1. 

How did you determine if 1...e6 doesn’t lose by force? 

Also, I was referring to games you claim to KNOW we’re played perfectly by both sides and ended in a draw. A perfect GAME, not one side making ONE perfect move.

How about a game you claim both side played with no errors and ended in a draw?

So it has happened thousands of times where one side or the other played perfectly [with no errors on his part which would change the theoretical result of the game]

It happens a lot less for both players to play a "perfect game"

which would require both players to not make a mistake [which would change the theoretical result of the game]

However with the billions of games played and with the increasing abilities of chess engines and humans over the decades--I am sure there have been perfect games played--both rather long type games and also very short games. And i am sure thousands of such games have been played. [just in the very short games there have been thousands of such games played]

You know very well i have already answered this question in the other forum. However I will give one such game here:

I think I might see the problem. USArmy keeps asking what a perfect game is because knowing what a perfect game is and perfect play is helps to figure out if a computer will ever solve chess. Ponz keeps replying that a game of a few moves is an example of a perfect game. 

I think Ponz has a very different opinion of what the definition of perfect is. To Ponz a perfect game is a draw, so, if a game is a draw it was probably perfect. Or his example above, where he deemed certain moves perfect, therefore, the game was perfect. To someone like me, and probably USArmy, that's impossible to know because we define perfect has having no better option. Not just no errors, but no improvements either. Which means people like me think no perfect game has ever been played because we have no idea if a better option was possible. Ponz on the other hand already knows (somehow) no other better option was available, so thousands, probably many many millions of games have been perfect. So to me there is a slight hiccup, computers still haven't solved chess. If "perfect" games have been played, and since perfection is probably essential to computers solving chess, why hasn't chess been solved?

You will find almost no grandmaster who will claim either side made a mistake in the game per post 4707.

I also claim that neither side made a mistake in that game.

It’s good to see you walk back your previous assertions. 

Humans can know a perfect move in SIMPLE positions to solve (a forced mate, avoiding mate, etc). But for every single move you have to be able to see it through to the end to know it’s perfect, to call the game perfect.

There are 10 (To the 120 power) possible game variations. Billions of games is like a drop in the ocean. 

How did you determine a future computer that solved chess won’t find that 3...Nf6 loses by force?

The problem with this is we have been defining "perfect game"

Also humans do not usually play perfect games. But how well or poorly humans play games has nothing to do with "solving chess"

In other words  perfect games can be played by accident.

Also the best chess computers do not play perfectly. [they are getting close however]

"No errors period" isn't clearly defined though.

For example if there's a mate in 4 and instead a person plays a mate in 6 is that a mistake? Why?

What if the mate in 4 was, practically speaking, more risky and the mate in 6 was risk free and easy.

If we're in a drawn position and I play a move that makes things more difficult for you, is that a better move than if I made it easy? Why? And what if that move is only difficult for some players while others would struggle to draw vs a different move? And even so, how can we accurately judge which move is most difficult to face? Sometimes players lose because they become complacient when the draw is easy. Sometimes players lose because you give them a winning position, but the complications cause them to err.

So it seems to me the only objective way to judge a perfect game is the way a tablebase would see it. Moves either maintain the evaluation (win, draw), or they make it worse (draw, loss)

We’re speaking in the context of chess being solved. If a computer ever does that there will only be three evaluations for ALL positions.

1. 0.00

2. White Mate in X moves

3. Black Mate in X moves 

To simplify this discussion let’s set aside assessments for “better” moves with the same outcome, for example two moves draw, but one creates more (or less) variations that transpose into a win.

An ERROR will be a move that changes the outcome with BEST play on both sides.

Billions of games compared to 10 to 120 power is small in that regard but it is big to the number of games   of perfect games played.

To use an anology  the game of checkers has been solved and it is known that quite a number of perfect checkers games have been played and it is also known that the number of possible games which can be played drawfs the number of actual games which have been played.

Ok good. I agree. In this context it's going to only be one of those 3 types of evaluations.

I think it's pretty safe to say, then, that perfect games have been played.

Perfect Najdof or KID games where the players enter sharp lines? No, probably not.
Perfect exchange slav or exchange french games? Yeah, probably quite a few.