Solving chess

 For those of you who just do not have the capacity to understand online humor, this was SARCASM!

 No, it would be "White mate in --"  I just don't know that number. White has some advantage, whether it be 0.10, 0.15, or even 0.01.

No, when all is said and done, and the game is solved, White's first move advantage can only be one of three values:

-999.99999999 (Solution is a win for black)

    0.00000000 (Solution is a draw)

+999.99999999 (Solution is a win for white)

Chess is a draw. Anyone who doesn't recognise this is just demonstrating that they understand absolutely nothing about chess. Ask any master player (or even expert).

All +0.15 or whatever means is that black has got slightly more rope to hang himself with than white has. You've got to get to at least 0.75 or 0.8 in computer terms to even consider the possibility of winning by force.

A forced win is the same as forced mate, you just need a computer that can find that number. So, the advantage would be:

White mate in -?-

0.00

Black mate in -?-

Nah, it's obviously zugwang. Forced win for black.

Seriously, my bet would, like most people, be very strongly on a forced draw. How reliable that guess is is dependent on how reliable limited horizon evaluations of positions are at identifying the real truth about them. We know that the very best computers can get endings wrong because of the horizon problem, and if you think of the avaluation of the starting position, there is a faint possibility they might get that wrong as well. But it is very faint indeed, based on intuition.

But as for claiming that you need 0.75 (or whatever) to be winning, consider the situation where a strong computer plays another, and the evaluation goes from near zero to -999 during the course of the game. It would only be true to say that if a strong program gives an evaluation over a certain level, there is a strong probability that the position is winning (even if the evaluation is high, sometimes it could even be losing due to the horizon effect). A reason to think this probably does not happen for the starting position is that it is quite a quiet position with many playable variations, where such things are unlikely to occur.

You misunderstand both what a computer's evaluation means and how it's derived.  Consider that a computer's evaluation will fluctuate during a game even if the moves it has calculated as best to render it's evaluation are played.

You also misunderstand the drawing margin of many endgames.  As a simple example consider that a lone knight can't checkmate.  This means what was a +3.00 advantage in the opening does not remain constant over the course of a game.  Other examples can be made, and the more you know about endgames you realize this list becomes quite long.

The bottom line being the advantage (some say disadvantage) of the first move is not large enough to matter in terms of forcing a win.

A little harsh :)  But I agree with the sentiment.  Knowledgeable players will assume perfect play is a draw.

Actually, 0.15+ isn't enough. As a famous grandmaster said, that one tempo is worth  about a third a center pawn and up to half a flank pawn. White has an advantage somewhat, or these would not be the master stats:

  38.4%  1-0             31.9%  1/2-1/2         29.7%   0-1

White has an 8.7% win rate compared to black. Whatever it is that is causing so many wins for white would cause a win for white with two supercomputers playing.

Those statistics are from a few thousand games, but the total amount of all possible grandmaster level chess games is so huge that a few thousand games simply can't be a reliable sample. That 8.7% could perfectly be the error margin due to the relatively small size of the sample we're using, compared to the total possible grandmaster games. If we could progressively increase the size of the sample, we might discover that that 8.7% progressively decreases and tends to 0%, therefore, we could assume that perfect chess is a draw.

I was wondering where you got 0.15 from :)  Yes, a third of a pawn is more like it, but of course if both sides complete development then you can't count tempo for an advantage this way.  Especially if the position remains closed.

As for those percentages,  Grobe counters that in his post #14 with a nice example.  In short, you can't use what is comparatively very poor play with many blunders to predict a games behavior at it's extremes.  You  might as well say 1000 rated players score evenly with white and black therefore chess is a draw.

Or, if you think GMs are so mighty in their knowledge, you might consider that GMs since the first official world chess champion have considered best play to be a draw.

There is an interesting point though, as computers approach the upper limits of best play, they should score closer and closer to evenly with both white and black.  I wonder if this could be one method used as a predictor for such limits (as chess will never be completely solved).

For example programs cannot continue to get stronger forever.  So is the upper limit for computer ratings 3500? 4000? 5000?

You fail to realize that that statistics are not always correct. Probability is a mathematical neccessity to fill in what humans cannot know determinately.

Also, the win rate of White can be easily explained by the fact that these are HUMANS playing...modern day chess relies on the strategy of winning with white and drawing with black, due to there being a larger margin of error as white and less as black.

When these new perfect supercomputers come along they will refute the human error of winning with white. It's because we use stupid openings like e4 or d4. If we knew the secret opening that is perfect then the perfect infallible computer would evaluate it as -0.00000001, which means black has a forced mate. Black must have some advantage to cause humans to win as black 30% of the time even though supposedly it is inferior.

In summary:  The entire premise of this thread is based on a flawed understanding of what it means to solve chess and of what computer evaluations represent and how they are calculated.

We might, but it could also go up. You can't guess at these things.

Solving chess means a super computer with every possible position in it, and it would have to know what to do at these positions. Computers say white has an advantage, if you have any advantage you will win with perfect play.

Only such a supercomputer as you described can accurately evaluate the starting position, and such a supercomputer as you described cannot exist.

Unfortunately, this argument of yours has already been disproved by an earlier post stating a material advantage of a knight (3 pawns) would be useless if there is nothing else on the board. I'd bet everything on a draw in this scenario, despite a supercomputer playing with the knight and any chess amateur with a grasp of the rules playing with a lone king.

Blunders don't account for such stable results like this:

 move   games       1-0        1/2       0-1

1.e4     597,035     38.4%     31.9%     29.7%

1.d4     455,024     38.7%     34.6%     26.7%

1.Nf3    126,616     36.8%     38.3%     24.9%

1.c4       96,598     38.5%     35.5%      26.0%

Then what have you been doing this whole thread?

As I have pointed out before, the win rate of White can be easily explained by the fact that these are HUMANS playing...modern day chess relies on the strategy of winning with white and drawing with black, due to there being a larger margin of error as white and less as black.