[EM] supermajority

James Green-Armytage jarmyta at antioch-college.edu
Sun Dec 26 18:39:12 PST 2004



Dear election methods fans,

What to do when we want to use a method that offers the benefits of
Condorcet but where a supermajority requirement is appropriate, e.g. where
70% of the electorate should consent to a new course of action before the
status quo is changed?

Here's an idea...

Let's assume that the 'base' method is ranked pairs, although the same
principle can be used with many majoritarian methods. Let's assume that
the supermajority required is 70%, although the same principle can be used
for any supermajority up to 100%.

Ballot: Assign each option a ranking, and also indicate whether or not you
consent to that option being implemented instead of the status quo. So the
second part is a yes/no ballot, but instead of an 'approval' ballot, I'll
call it a 'consent' ballot, to avoid confusion...

Tally: Do a ranked pairs tally. We'll call the winner option A. Is A
consented to by at least 70% of the voters? If so, option A is selected.
If not, then the status quo remains in place.

Commentary: The advantage of this method is that it minimizes the
incentive to vote tactically on the consent ballot, because it cannot be
used to give an advantage to some option that you like better than the
sincere base method winner... at least, not unless there is another round
of voting in the event of the status quo remaining in place.
__________________________

Above is my primary proposal for supermajority voting situations (e.g.
constitutional amendments, etc.) I am also toying with an alternate idea,
which is as follows.

Ballot: Ranked ballot and consent ballot.

Tally: Eliminate all candidates who are not in the minimal dominant set,
and who are not consented to by at least 70% of the voters. If any
candidates remain, hold a ranked pairs tally between them to determine the
winner.

Commentary: This is a bit less 'strict' than the method above, and may be
slightly more likely to produce a non-status-quo winner, but I think that
it does not minimize tactical incentive on the consent ballot to the same
degree. Hence, I prefer the first proposal, but I'm very much open to
further discussion on the issue. Also, please let me know if this idea or
similar ideas have already been proposed.

my best,
James Green-Armytage




More information about the Election-Methods mailing list