Dec 9, 2004

The PIE rule

The more long-term the goal is, the smaller the relevance of the PIE rule. There is a strong temptation to think this way, but I am in some doubt. For example in Go, it turns out that once the board size is past a small minimum, the komi is remarkably constant. It seems to be about 7 for all board sizes greater than 5x5. This suggests that the long-termness of the goal (at least of some games) is irrelevant - the PIEness is always about the same; though of course it diminishes in PROPORTIONAL importance to the other moves.

Another reason is that the initial advantage can be built up with good play to its final conclusion. The PIE rule is a tool for the placing player to reduce it to a value very near zero. Ideally, a perfect use of PI implies that only a perfect player can use that setup to achieve victory

p.s. For some reason I'm reminded of a pair of comments about playing against GOD (Game Optimization Device) and the DEVIL (DEVice of ILegitimacy)

* GOD always makes the optimal game-theoretic move; but
* DEVIL always makes the best move given what your overwhelmingly likely response is to be.

If GOD plays a perfect game; DEVIL may play even better (!) because it exploits your weaknesses.

No comments: