[EM] Voting methods & utility

[LAYTON Craig]

David Catchpole wrote:

>I think it's wholly appropriate to discuss utility when discussing voters'
>_behaviour_ but not their vote. A vote may well reflect the _cardinal_
>utilities of a voter but they will in no way (in fact it is apparent that
>it will not) reflect any kind of continuous utility.

I don't know what you mean by utility in relation to voters' behaviour.  If
you're equating utility with preference satisfaction, I've already stated
this is a crude approximation.  Do you mean that the more voters who are
satisfied by the outcome, the higher the utility?  In most cases this is
true, but it is hardly a matter of significant differences in utility.  As I
have said, there is strong reason to believe that disutility should be
avoided over maximising utility, and as a result it is more important that
few people are strongly opposed to the outcome of an election.  This can be
encouraged by a system in which more people have a representative in
government.  I have written a diatribe on this, so I will not repeat it.

This is the reason I have not written much on single winner systems.  I
think all single winner systems that 1) choose a winner from the Smith set
and 2) do (1) in such a way so as the voters are happy with the outcome (ie
don't pick a winner from the Smith set out of a hat, this violates 2) are
all roughly utility equivalent.  I see no reason to refine a system further
for single winner political elections, although there is much merit in
refining voting systems for other purposes, most importantly deciding
between particular proposals, rather than representatives.  But at present
systems for deciding between proposals or courses of action are badly
hampered by assumptions that preferences are linear.  I've not seen much
progress on this front.