[EM] Introductory Message

[Bart Ingles]

James Green-Armytage wrote:
> 
>         I may as well point out that there is some controversy on the list about
> the merits of approval voting. I happen to be a person who is a bit more
> dubious about it than some. One serious drawback that it seems to have is
> that it doesn't pass the "majority" or "mutual majority" criterion as
> defined below:
> 
> majority criterion: If a majority of the voters prefers all of the members
> of a given set of candidates over all candidates outside that set, and
> they vote sincerely, then the winning candidate should come from that set.

The majority criterion seems to be more important to some people than to
others.  Although it is defined above using sincere preferences, it can
also be defined in terms of actual ballots.

I could agree that the criterion is fairly important when defined in
terms of actual voted ballots.  When you have an real set of ballots
showing that a certain candidate is preferred by more than half of the
voters, then explaining why a different candidate actually won could be
problematic. 

Systems which fail the majority criterion on actual ballots include:
Borda Count, the U.S. electoral college system, and CR (but at least
with CR the reason for overriding the majority is relatively easy to
explain).  Systems which pass the majority criterion as defined in terms
of actual ballots include: Plurality, top two runoff, instant runoff,
the Condorcet family of methods, and approval voting (whenever a single
candidate receives a first-choice majority under any of these systems,
that candidate wins).

When defined in terms of sincere preferences, I view the majority
criterion as somewhat desirable but not particularly important.  For one
thing, it's not usually possible to know what peoples' sincere
first-choice preferences are in the real world.  Due to strategy
incentives, most methods are prone to "false positives", in that the
ballots will tend to show a first-choice majority even when there is
none in terms of true preferences.  This is more pronounced with
Plurality and the runoff-based methods, but could also be the case with
the Condorcet methods if a cycle seems likely.

With any decent non-Duvergerian system (i.e. a system which doesn't
artifically promote a two-party system), the more candidates present,
the less likely you are to ever see a first choice majority (either
sincere or on the ballots).  Thus the majority criterion is not likely
to be put to the test in a healthy political system with several
candidates.

Bart Ingles