Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Posting-Version: version B 2.10.2 9/5/84; site sunybcs.UUCP
Path: utzoo!watmath!sunybcs!ugfailau
From: ugfailau@sunybcs.UUCP (Fai Lau)
Newsgroups: net.puzzle,net.games.chess
Subject: Follow up to TWO PLAYER GAME AND WINNING STRATEGY
Message-ID: <2942@sunybcs.UUCP>
Date: Sun, 16-Mar-86 01:09:36 EST
Article-I.D.: sunybcs.2942
Posted: Sun Mar 16 01:09:36 1986
Date-Received: Mon, 17-Mar-86 04:35:13 EST
Distribution: net
Organization: SUNY/Buffalo Computer Science
Lines: 24
Xref: watmath net.puzzle:1518 net.games.chess:316


	In reference to some responses to the original article I posted
in net.puzzle and net.games.chess, I found one fundamental flaw of the
evaluation of perfect play to be the scoring system. In various
situations the same score can by assigned to two moves of very
different qualities. In other word, a second best move in one
state of the game may mean a delay in checkmate while a second
best move in another state may lead to instead defeat.
Since the points are assigned to a given move according to
the quantity and quality of the alternatives availables to 
that move, it is not an accurate indication of 'deviation'
from the perfect play. The system is similar to comparing the 
performances of two students from classes of different curves
sololy on the basis of their grades. Perhaps if there is a way
to assign scores based on an ABSOLUTE scale..., but then
the original question can't be answered until such system is
derived.


+-----------------------------------------------------------------------------+
|  Fai  Lau                                                                   |
|  ECE / CS   SUNYAB                                                          |
|  BI: ugfailau@sunybcs                                                       |
+-----------------------------------------------------------------------------+