On 2003-06-30, fabio guillermo rojas uttered to [EMAIL PROTECTED]:
>Chess games seem to be a combination of rule behavior ("try top control
>the center board," "castle early as you can") with a little bit of search
>through a gigantic strategy space.
True, but how are those rules generated? They are generated as local
heuristics conducive to maximum utility at the end of the game. Why isn't
the conventional, full solution being used? Because of computational
constraints -- we can't search through the whole game tree. So this would
suggest the problem is indeed a one of utility maximization, only this
time with resource bounded rationality.
Still it's an interesting example, because there is a tradeoff between
payoffs from computation and its cost. You wouldn't want to ruin the
utility you get from the game by planning each move for 10 years. It
becomes still more interesting when you go to speed chess. The first
interesting thing is that, generally, the computational cost in
combinatorial problems rises exponentially with lookahead, so there might
actually be a way to sort-of quantify the utility in this case. (Aye, that
becomes perilously close to cardinal utility.) Second, I think we would
also expect less experienced players to prefer speed chess -- ceteris
paribus, knowing the right heuristics is less important in speed chess
because the other player's actions can always stall a long strategy.
That means everybody's constrained to simpler play, and so the
considerable intuition and long-term strategic talent of a grandmaster
would be wasted. (The argument isn't watertight, of course.)
--
Sampo Syreeni, aka decoy - mailto:[EMAIL PROTECTED], tel:+358-50-5756111
student/math+cs/helsinki university, http://www.iki.fi/~decoy/front
openpgp: 050985C2/025E D175 ABE5 027C 9494 EEB0 E090 8BA9 0509 85C2
