Alex_Svitashev wrote on 04/24/13 at 02:58:06:


There are a lot of assumptions/opinions in the OP.  
Interpreting the last question as "how would you use a large database to study the rate of best moves played by an average, GM or WC player ?" :

First I would define a best move in some game-theoretic setting. Two main possibilities :
a) best moves in a position are those maintaining the initial status of the position (win, draw, lose). In this respect, all moves are equally best if they maintain a winning position, and all moves are equally best in a losing position, since none can improve the evaluation of the position. Note that this status in not always known, as chess isn"t solved yet.
b) you can define additional metrics such as "in winning position: shortest distance to mate" or "in losing position: maximum severity of winning sequences", with such severity related to some probability that the opponent loses its winning status, or any other criteria you see fit.

Having chosen a metric you're happy with, you can take a subset of measurable chess positions (e.g. ending tablebases when the metric is either distance to mate or position status), and check your big game database against it. For each position both in the tablebase and in the gamebase, you can score the frequency of (or distance to) "best moves" played in each category. 

That's what I would do anyway if life wasn't so damn short. 

I think it is more likely that in many positions, there are in fact quite a few perfect moves. If current position isn't a forced win for either side, then any move that doesn't lose by force is as good as any other, theoretically. You can observe this by playing a drawn endgame against a computer using tablebases, in many cases any move draws and it just appears to play randomly.

Perfect games have been played many times, for instance there are a lot of short known theoretical draws that are almost certainly perfect (if the game is in fact a draw), and quite a bit of current mainline opening theory may be perfect in the sense that it doesn't lead to a position that loses by force.