Solving chess

Computers say white has an advantage today because chess is not yet solved.  They too are currently working with imperfect information and are making assumptions.  Assumptions that may well be wrong.  It's not like someone solved chess to arrive at the +0.15 analysis and then we all forgot.

Once chess is solved, the starting position's correct evaluation will be only one of three possible values:

-∞

0.0

+∞

The current "best guess" of +0.15 is a lot closer to 0.0 than it is +∞, and most people who understand these things agree that chess is most likely a draw with perfect play.

Also, given that K+N vs. K (or K+B vs. K) is a 3.0 point differential, but still dead drawn, I think it would be wise to let go of the misconception that any advantage, no matter how small, can be converted to a win.

Which computer claims an advantage for white anyway?  I'd like to know which program matthew is using.

Whichever one it is, notice that it's not a "chess oracle" which only speaks the truth as whichever chess engine you happen to be quoting will lose games even against itself.  That fact that it will lose as white means it's evaluations are not as if they were from an oracle.

Better yet, load nearly any drawn endgame into an engine and see what its evaluation is.  If a computer can't correctly evaluate a position near the end of the game, why would you trust it to evaluate the opening position when there are exponentially more lines to calculate.

Or perhaps you could look into how programs play chess in the first place, and how they're not turned to evaluate opening positions in the first place.

I mean there are so many examples but this is already on page three.

As for the statistics again why are you trying to use something like that to predict the outcome of a perfect game?  Those stats aren't even related!  TheGrobe puts it nicely when he points out that 0.15 is closer to zero than it is infinity. 

Although if you're so fond of statistics, notice also there are more draws than wins in top level play.  (Although again this is a bad example for a number of reasons).

Just to clarify, I pulled the number 0.75 out of mid-air just to illustrate the point. Of course, it's nowhere near clear but like that in real life. It is possible to construct a position which Fritz scores about +15 which is clearly a draw:

EDIT: "clear but" should read "clear cut"

Do you know why computers say that white has an advantage on move one? (Rhetorical question: if you did, you wouldn't have started this thread.) It's because they're programmed to do so. Try this with your engine: evaluate the position after 1. e4 e5 and compare its value to the position after 1. e3 e5 2. e4. Is the (absolute) value of the positions same at all depths? If not, you'll see that whatever advantage the computer gives at the beginning is not very reliable. It's just an arbitrary (and volatile) value representing the advantage of having a tempo over your opponent.

The advantage for white is the tempo.

ny computer would say thats 0.00

K+N vs. K is dead drawn. White's material advantage is machted by the fact he can't mate black with the lone knight and king.

The question at the root of solving chess, though, is whether or not it really is an advantage (or even possibly a disadvantage) and the fact is that no-one really knows, including those who programmed your chess engine to say it was +0.15.

Yes, which was precicely my point. 

Try it. You might be surprised.

And infinity is an over statement, +1.00 would be the same. It's certain that you would win with a +1.00 score. (with perfect play) I don't see how black has more chances to mess up either.

The point is that it's not certain, and in fact there are positions that initially appear very imbalanced that are dead drawn and which engines grossly mis-evaluate.

Nope

Here are some positions where the extra material out weighs the value of the first move, sometimes by a lot of points.  The computer agrees, but these are all draws.

In order of the diagrams, Houdini 1.5 (one of the strongest if not the strongest engine around) gives the following evaluations of these drawn positions:

-0.53  (black's bishop is the incorrect color)
-3.74  (A well known fortress with Rf4)
-1.36  (Nh3 and g4 will draw this one)
1.78  (A book draw that can get complicated in analysis)
3.32  (Black can shuffle with Bf6-g7-f6 or Kh8-g8-h8)
1.18  (Black can shuffle Kb4-c5-b4 or Bc1-f4-c1)

You don't get it, yes there are positions where an extra pawn won't win but that's not what I mean by "+1.00". What I mean is that if you have a position where you have an advantage that truly is +1.00 you will win with perfect play, even if you are down material.

Ok.

So the point become that the first move isn't that much of an advantage.  Computer play, GM play, anyone's play you can see that you can't force an advantage that large :)

No, you don't get it.  Your argument is entirely circular.  There are only three types of positions:

1. White can force the win
2. Black can force the win
3. Dead Drawn

We know which is the case for many positions (although computers still evaluate some of these incorrectly), but unless we know it for the starting position we've not solved chess, and a computers best guess of +.15, or +1.0 tells us nothing about the actual outcome with perfect play.

More to the point, the notion of a +1.00 vs a +999.9999999 vs a +infinity evaluation is actually meaningless when you know what perfect play is -- they are all the same, white wins.  We might as well just say +.  The reason that we use an analogue scale is to approximate our best guess at the relative advantages with imperfect play -- basically at the level the engine doing the evaluation is capable of.  What this means is that these evaluations are completely meaningless in the context of solving chess, and trying to invoke them results in a circular argument, as you've demonstrated.

I do believe he's a better player than I was at 12 or 13, but also this may be difficult to grasp when you're 12-13.

Did I mention I just found out he's a kid

- I sense much fear in you, young Jedi.

Lucky for you (and me and all chess lovers) there are so many moves that even if the analysis for 1 position took 1 bit of information there are some problems as to where you could possibly store this solution for all positions.

As in enormous problems... as in you could have a digital storage device with as much mass as the observable universe and it wouldn't all fit in there

For this reason I don't think memorizing the solved lines will be doable (just like EGTB I'm sure there will be multiple moves along the way that draw or win).  And as other threads have explored, there's no way currently to even theoretically try to solve chess.

I dont believe chess will ever be "solved" . If I am wrong though and it is I dont believe humans will be able to remember the solution so it won't matter.... people will still play and there will still be competitions between humans.