 |
 |
Dr. Osman's message is puzzling. My paper does not mention the AHP. Here is the abstract: "Preference measurement plays a fundamental role, and is necessary in order to introduce the real numbers and operations on them, in game theory and economics. It is not possible to escape the need to construct preference functions by assuming that payoVs are in money units and that each player has a utility function which is linear in terms of money. The mathematical operations of game theory are performed on preferences for objects rather than on empirical objects, preference scales are not unique, and preference spaces are not vectors spaces. UndeWned sums and ill-deWned concepts which are based on preference scales where whose preference is being measured is not speciWed are the source of many game theory errors. For example, the characteristic function of a game, the Shapley value, the concept of imputations, and von Neumann and Morgenstern's solution of a game are ill-deWned. Von Neumann and Morgenstern's utility theory and its variants cannot serve as the foundations of any scientiWc discipline because their construction rules are contradictory and for additional reasons." The paper does deal with logical errors at the foundations of decision theory. Here is a sample: "Howard says that a normative theory establishes norms for how things should be (In Praise of the Old Time Religion [13, p. 29]) and appears to suggest that decision theory says how you should act in compliance with von Neumann and Morgenstern's assumptions [13, p. 31]. His comments on "second-rate thinking" and education [13, p. 30] seem to indicate that he believes that those who do not share his praise for the old time utility religion need to be re-educated. In the context of logic and science this position is untenable - mathematical theories do not dictate assumptions to decision makers. Furthermore, educating decision makers to follow Xawed theories is not a remedy for "second-rate thinking." Flawed theories should be corrected rather than be taught as the norm. Unfortunately, according to Edwards [7, pp. 254-255]), Howard is not alone. Edwards reports as editor of the proceedings of a conference on utility theories that the attendees of the conference unanimously agreed that the experimental and observational evidence has established as a fact the assertion that people do not maximize "subjective expected utility" and the attendees also unanimously stated that they consider "subjective expected utility" to be the appropriate normative rule for decision making under risk or uncertainty. These utility theorists are saying that although decision makers reject the assumptions of the mathematical theory of utility, they should accept the conclusions which these assumptions imply. This position is logically untenable." It seems that Dr. Osman responded to my paper before reading it and attached a paper that cannot be a response to mine since Mrs. Saaty's paper was published before mine was written (to the best of my knowledge Rozann Whitaker does not hold a doctoral degree and being Tom Saaty's wife constitutes her only qualifications in decision theory). The logical relevance of Dr. Osman's message to my paper is unclear. Jonathan Barzilai, D.Sc. Professor, Dalhousie University