As the Theorem treats them, voters are already utility-maximizing agents who have a clear preference set which they act on in rational ways. The question: how to aggregate these?
It turns out that if you want certain superficially reasonable things out of a voting process from such agents—nothing gets chosen at random, it doesn’t matter how you cut up choices or whatever, &c. - you’re in for disappointment. There isn’t actually a way to have a group that is itself rationally agentic in the precise way the Theorem postulates.
One bullet you could bite is having a dictator. Then none of the inconsistencies arise from having all these extra preference sets lying around because there’s only one and it’s perfectly coherent. This is very easily comparable to reducing all of your own preferences into a single coherent utility function.
As the Theorem treats them, voters are already utility-maximizing agents who have a clear preference set which they act on in rational ways. The question: how to aggregate these?
It turns out that if you want certain superficially reasonable things out of a voting process from such agents—nothing gets chosen at random, it doesn’t matter how you cut up choices or whatever, &c. - you’re in for disappointment. There isn’t actually a way to have a group that is itself rationally agentic in the precise way the Theorem postulates.
One bullet you could bite is having a dictator. Then none of the inconsistencies arise from having all these extra preference sets lying around because there’s only one and it’s perfectly coherent. This is very easily comparable to reducing all of your own preferences into a single coherent utility function.