[EM] Re: IRV's Top Preferences Considered First Feature

[Forest Simmons]

On Thu, 12 Jun 2003, Forest Simmons wrote:

> Indeed artificial ballot constraints tend to elicit insincere ballots.

Note how Approval shines in this regard:

An Approval ballot may be considered as an unconstrained plurality ballot.

At a more subtle level Approval may be considered a CR ballot in which all
votes are constrained to the extremes of the ratings range. But it turns
out that this constraint is not artificial since it precisely conforms to
the dictates of optimal CR strategy regardless of the amount of
information available.

Note, however, that there are other uses for CR ballots than the standard
CR method of giving the win to the candidate with the highest average
rating.  In some of these other uses, including approval runoff,
constraining to the extremes is not in general consistent with optimal
strategy.  In this context this constraint would be almost as artificial
as the IRV and Coombs constraints.

Forest