[Next] [Previous] [Top] [Contents]
The Possible Incomensurability of Utilities and the Learning of Goals - Bruce Edmonds
3. Practical Examples
That this is not such a far-fetched example for economics, consider the example where firms effectively engage in complex marketing games against each other, successively placing products and advertisements, where the aim is merely to survive. They may well use a variety of ad hoc indicators as to the extent they will avoid extinction (like profit, predictability of cash-flow, customer loyalty, turn-over, diversity of product range etc.) but it is likely that any firm which simplifies their strategy down to a single indicator will not do very well, however sophisticated it is.
Now quite apart from the theoretical possibility of mapping the agents' utilities onto one utility, it is even less sure that this is remotely practical. The trade-offs between the different utilities might vary so critically from context to context that it would be foolish for the agent (or us) to model it by mapping them onto a single utility function (even a single ordinal utility ordering). Thus even if one supposed that one must be able, in theory, to map these utilities onto a single utility, this might well be beyond the capabilities of the agent or us in many contexts.
The Possible Incomensurability of Utilities and the Learning of Goals - Bruce Edmonds - 05 SEP 97
[Next] [Previous] [Top] [Contents]
Generated with CERN WebMaker