My two cents worth on utility:
1. Utility can be a useful concept for an individual to use in making a
decision, even though it may be impossible to calculate. For example, if
candidates A, B, and C have equal priors of winning, and my preference order is
A>B>C, then I might decide to approve a "ho hum" candidate B on the basis that
it seems to me that B is closer to A than to C in my own personal value system.
Neither Approval nor Dyadic Approval requires more than this "half way mark"
kind of utility judgment.
2. In the PR context, one could have as a theoretical common denominator of
utility for each candidate the probability that the candidate would represent
your point of view on a random issue that comes before the representative body.
Of course, not all issues are of equal importance, so you want to take into
account that someone might disagree with you on all of the wedge issues, but
agree with you on the one or two life or death issues. Soon you see that even
this utility is more of a gut feeling than anything else. Furthermore, there
are interactions among the candidates that could make your utility for one
candidate depend on the election of another candidate that has a positive
influence on him.
3. I am skeptical of using utility additively (as in range voting), especially
in capitalist countries (but also in those communist and socialist countries
where the wealth is controlled undemocratically) because under all systems with
non-democratic economies (including capitalism) benefits don't get spread
around equitably.
To see what I mean, consider the two person game direct payoff matrix
(100, 0) | (20, 20)
------------|-------------
(20, 20) | (0, 500)
By direct payoff, I mean that the values in the matrix are to be paid directly
to the players by some third party (say fate).
In one setting (analogous to capitalism) the row player should choose the first
row and the column player the second column. Each player gets twenty of their
units of utility. It doesn't matter if the twenty utility units are comparable
or not, because they are not going to be shared. [This (20, 20) result is
analogous to an approval election outcome.]
In an equitable society, it would make a difference if the utilities were
comparable and additive. If the row player sees 250 of the column player units
as more valuable than 20 of his own units, then he should choose the second
row, with the understanding that the 500=250+250 units are going to be shared
equally. [This 250+250 result is analogous to an outcome of range voting in an
equitable society.]
Thus two brothers with neighboring farms could benefit more by cooperative
planning (based on weather probabilities) than they could by going it alone,
especially if they were willing to share the revenues and costs.
Nafta raised the gross national product of both the USA and Mexico, but only
because the rich became super rich, while the poor got poorer.
The last four U.S. administrations have been very effective in caulking up all
of the leaks that were supposed to supply the poor with trickle down.
Sincere range voting minimizes expected (total or average) Bayesian regret, but
that is desireable only in an equitable society where the mathematical
averaging is reflected in reality by actual sharing of the costs and benefits
throughout the population.
[Our version of capitalism is "Socialize the cost (by taxing everybody), but
privatize the profits," and its corollary, "subsidies for the rich corporations
and their stock holders, but market discipline for the poor." Many
fundamentalist Christians while giving lip service to "No man can serve both
God and Money," paradoxically support this brand of capitalism with religious
fervor.]
That's why (in the real world) I prefer the sincere approval winner to the
sincere range winner: sincere Approval tends to maximize the number of voters
that get a result that they can live with, without having to beg the stingy
super utility winners to spread around their loot.
Fortunately, under range voting most voters will soon learn to use approval
strategy, so range voting will give results as good as approval voting in
non-equitable societies. It's a shame, however, that range ballots have to be
used to get approval results, when simple approval ballots would do.
The least we can do is use ballots that have simple options for the voters that
don't want to specify range numbers for each candidate.
Unfortunately, both approval and range voting are vulnerable to disinformation
campaigns that tend to distort the (subjective) prior probabilities.
Methods that make judicious use of pairwise comparisons can ameliorate this
problem. DMC uses both approval info and pairwise info, but in the end gives
more weight to the pairwise info, which is appropriate here in the U.S.A. where
the fat cats can manipulate prior probabilities like silly putty.
Forest
<<winmail.dat>>
---- Election-methods mailing list - see http://electorama.com/em for list info
