Ah, Bill, thank you for your understanding reaction to my pungent expression.
The way Shlomi and Bill are both talking, my sense is that they are concentrating on the details of the architecture of solver design, and further that their analysis would only apply to an "imperfect" solver. To put it in pragmatic terms, aren't you both saying that a solver's solving algorithm could trap it into shutting itself off from exploring a winning path by the route which it took to reach a specific given position? I will submit that any solver that could do that to itself is improperly designed, or, to be more tolerant about it, is not designed with the goal of eliminating any chance of a false impossible. For Shlomi's original objective of finding solutions to as many deals as possible in the shortest possible time, ignoring intractables (and considering them in the same category as "impossible"), such a solver design would be valid. But not for a solver whose objective is to provide a fully reliable verdict as to the solvability of any position.
As to "game theory" and Shlomi's comment that a "game" must involve more than one player, I'm reminded of an astute comment Mike Keller once made about Freecell. That Freecell is not really a "game", it's more in the nature of a "logic puzzle".
Best regards, ---------------Adrian
Received on Wed Jun 26 2002 - 05:25:30 IDT