[EM] Social Utility

Blake Cretney bcretney at postmark.net
Wed Apr 18 22:58:26 PDT 2001

One perspective on elections is that we should attempt to maximize the
utility of the outcome.  If, for example, each voter accurately
represented the value of each outcome to themselves, then we could
pick the result with the greatest value.  This is sometimes referred
to as social utility.  Of course, it is only true social utility in
the utilitarian sense if each person is accurate in their assessment.

Let me suggest two fundamental kinds of elections, which I will call
subjective and objective.  In a subjective election, the question can
only be asked with reference to the opinions of the individual voter,
but with this knowledge can be answered with certainty.  In an
objective question, the answer depends on some unknown fact about the
external world, independent of the voter.

For example, if the election is over what television program should be
watched on a common TV, then there is no right answer independent of
the wishes of the voters (assuming they are the ones who will watch
the TV).  On the other hand, if you knew the desires of the voters, a
program could be found to maximize utility.  This is what I call a
subjective election.

In a subjective election, it makes sense to try to determine the
perceived utility of each option to each voter, since this provides
the answer we want.  So, we might hope that an election method should
approximate this.

But let's say you have a group of people voting on whether the
universe will always expand, or will eventually contract.  This is an
objective question.  The correct answer is independent of the will of
the voters, and to answer it, the voters have to make educated guesses
about the world around them.  Any answer will only be a guess, but we
would want the method to pick the guess most likely to be right, based
on the votes.  It isn't as clear that utility measures would be useful
in this kind of election.

It seems to me that people who make great use of social utility in
their arguments would tend to assume that elections are primarily
subjective questions.  Is that true, or am I just attacking a straw
man?  The other question is, if elections are primarily objective
questions, what does that mean for election methods?  I think this
tends to weigh against trying to approximate Ratings.  Any comments?

Blake Cretney

More information about the Election-Methods mailing list