[EM] recent postings

[Forest Simmons]

It does my heart good to see so many spirited postings on the list near
the fourth of July with some new writers in the mix!

My perspective on single winner methods has moved more and more towards
the point of view that ranked ballots are costly in terms of voter
patience (as opposed to the cost of voting machines, ballot counting,
etc.), and that the best condorcet methods just barely justify that cost
in public elections, if at all, and that the other methods based on ranked
ballots fall far short of justifying that cost, though various ranked
methods including Borda may have other applications in other venues.


Chris Benham recently pointed out again that IRV voters tend to rely on
the guidance of candidates or parties, rather than figuring out their own
rankings.

In other words, IRV has all of the cost of a ranked ballot system, but it
functions as a Candidate Proxy method.  Why pay for IRV when you can get
the same result from Candidate Proxy at bargain basement prices?


My only concern about Approval is that it can be manipulated somewhat by
fake polls, so that strategic approval voters are either too optimistic
(setting their approval cutoff too high) or too pessimistic (setting their
cutoffs too low).

MCA addresses this problem at the expense of a slightly more complicated
ballot, but not nearly as complicated as a ranked ballot.


Two bits of information are necessary to specify the three levels used in
MCA, but this is somewhat wasteful since two bits are sufficient for four
levels.

An interesting question is how to make optimal use of the four levels that
a two bit ballot affords us.  In other words, if we are going to go beyond
Approval and Candidate Proxy, to a slightly more complicated ballot (two
bits, rather than one), how much can we get for the extra cost?

Kevin has proposed "four level Condorcet."

I think that has a lot of merit.

Others have proposed generalizing MCA slightly.

Here's another idea:

Let A and B be the two candidates with the most votes in the top two
places (i.e. least in the bottom two places).

If the set {A,B} has members both below and above midrange OR both members
adjacent to the midrange on the same side, then set the approval cutoff at
its midrange nominal value 1.5 (halfway between zero and three).

Otherwise move the approval cutoff up or down a notch to 2.5 or 0.5,
depending on which side contains the set {A,B}.

The approval winner based on the cutoff determined according to the above
rules is the method winner.

Preliminary testing based on standard problematic distributions of voter
utilities and fake polls seem to show that the method is fairly immune to
manipulation.

Alternately, one could use Adam's cutoff, but that would tend to
non-monotonicity, since winning the "first round" can give too much of a
disadvantage in the second round. In other words, adding enough votes to
change your favorite from second place in the first round to first place
in the first round could make him go from winner in the second round to
loser in the second round because of the handicap given the first round
winner by Adam's cutoff.

I think the method with my cutoff, which doesn't care which of {A,B} is
the frontrunner, is more apt to behave monotonically.

Forest