[EM] "Higher Resolution Methods"

[MIKE OSSIPOFF]

>My criticisms of Approval go double for Cardinal Ratings.  At least with
>Approval, by assisting voters to use optimal strategy, the chance of a
>non-majoritarian outcome is significantly decreased, as is the difference 
>in
>power between individual votes (as voting power is somewhat standardised).
>This important issue is disenfranchisement of voters (typically poorer or
>minority voters) who do not have the same access to information or
>understanding of strategy as other voters.

True, with CR, it would be important to make strategy information
available. If it's successfully made available to everyone, then
the problem that you speak of would be avoided.

Though Approval is better than CR, for the reason that you describe,
CR _might_ be easier to enact, because people are so familiar with
points ratings. Of course, during the enactment campaign, CR's
equivalence with Approval should be mentioned, because otherwise
people would be saying "Why didn't you tell us that before?".

>Of course, these problems are
>further solved by using ordinal rankings, but then again, ordinal rankings
>can have other problems, as most of you have pointed out.

Those problems depend very much on what rank-count is used. That
makes all the difference.

As I said in a reply that seemed to not post, I like the best rank-counts, 
the versions of Condorcet's method, the interpretations of the count rules 
that he proposed. I like those because of their
strategy criterion compliances. Approval does quite well in that
regard too, and I consider it a strong 2nd choice. Sometimes, if
people won't accept rank balloting, or if it's impossible to get
an adequate rank-count enacted, then Approval can become the best
proposal. Maybe sometimes it would be more prudent to start with
the simple, minimal-change Approval proposal.

When scored by social utility simulations, Approval consistently does
better than IRV. Social utility matters; it's a measure of the overall
summed rating of the winning candidate. It matters because if there's
going to be an election some years in the future, and we don't know
what method will be used, or what the candidate linup will be, what
the voter configuration will be, or where we'll be in it, etc., then
your expectation, estimated now, for the outcome of that future
election, is better if the election is going to be by a method that
does well in social utility.

In terms of the pairwise preferences of a strategizing voter, Plurality
messes up in several ways: When you vote for Middle, just to keep
Worst from winning, though you prefer Favorite to Middle, Plurality
forces you to vote Middle over Favorite. And it doesn't let you vote
Favorite over anyone. Approval gets rid of both of those faults.
Only one lack remains: You don't get to vote your preference for
Favorite over Middle, when voting Middle over Worst. But, overall,
Approval goes a long way toward getting rid of Plurality's problems.
For such a simple method, Approval brings big improvement. Anyway,
when the count results show Favorite with more votes than Worst,
voters will realize that they don't need to give Middle a vote next
time. This is never shown with Plurality. Approval quickly homes in
on the voter-median candidate, and stays there.

Weber & Myerson have a definition of voting equilibrium, which
can be summarized by saying that an equilibrium is an outcome that
is consistent with the predictive beliefs that caused people to vote
as they did. They point out that, in Approval, with 1-dimensional
policy-space, the median candidate is the only one who can be
elected at equilibrium. With Plurality, nearly _any_ 2 parties that
the voters expect to be the 2 frontrunners can always remain the
frontrunners, purely because of that belief.

In Approval, if you vote for Middle, it's because you believe Worst
will get more votes than Favorite. If the voter median is among the
voters who prefer Favorite, then Favorite has a majority. If they've
voted for Favorite as well as Middle, as they always would, then
the count results will contradict their belief that Worst would outpoll
Favorite.

Though Approval doesn't let you vote every pairwise preference, it
at least reliably counts each one that you vote, which is more than
can be said for IRV. I'd rather be the one to decide which of my
pairwise preferences will be counted, rather than have IRV decide that
for me by its capricious procedure.


Mike Ossipoff



_________________________________________________________________
Get your FREE download of MSN Explorer at http://explorer.msn.com