[EM] Some brief campaign argument (Approval)

[LAYTON Craig]

Richard wrote:

>Given your SU rankings (A > C > D > B), it strikes me that you are using a
>different definition of SU than I am. When I speak of SU, I am referring to
>an L1-normalized SU. Does this contradict an accepted definition of SU?
>If we calculate SUs according to this method, then we get (D > B > A > C).
>So Approval picked the second-highest SU.
<snip>
>I suspect you are using absolute SU calculations. If we accept this method
then
>we would indeed conclude that B has a poor SU in spite of appealing to two
>factions of voters who otherwise have no common candidates (so B succeeds
as
>a compromise).

This is perhaps what I should have concentrated on from the beginning.  I
originally intended to discuss it in a discussion of Cardinal Ratings, but I
didn't want to write pages and pages that no one would be interested in.

The reason I use absolute SU calculations is because absolute utility is
what is actually important.  It is important to note that absolute can
diverge significantly from weighted utilities.  I am suspicious of the
assumption that it will all even out in the end.  I strongly suspect that
some factions of voters are much more likely to have a particular pattern of
utilities than others.  What I mean for example is that Candidate C's
supporters (voting C,B,A,D), on the whole, might have very low expected
utilities (average 5,3,1,0) while Candidate D's supporters (voting D,A,B,C)
have very high expected utilities (99,95,94,92).  It isn't unreasonable to
suggest that factions who vote in different ways have very different actual
utility patterns and values.

This is, of course, not a very useful concept, because it makes discussions
like which voting method is best for social utility not very productive.
>From the sense of reported absolute utilities, it is very difficult to be
able to measure your own utilities against the utilities of others, if not
epistemologically impossible.

>Except in mechanical systems (such as computer simulations) or where the
outcome
>has a directly measurable payoff to the voters (such as a monetary reward),
I don't
>believe in absolute utilities. They just don't make sense where human
passions are
>involved. Absolute utilities are the flip side of the "one man one vote"
argument,
>and just as specious.

As normalised utilities are simply distorted absolute utilities, that are
increasingly separated from the goal of utility maximisation, it doesn't
make sense to say that one should pay attention to normalised utilities, but
not the real utilities they were derived from.  If the goal is utility
maximisation, you have to accept that the ultimate goal is absolute utility
maximisation.  If you can't accept this, then you should not support a
method on the basis of SU maximisation (I note that Richard didn't claim to
support Approval solely on SU maximisation).

Qualification: I don't totally disregard SU arguments, I'm just pointing out
the problems.