I realize that it's for Shugart, & not me, to define the standards
that he proposes, but I'd just like to sa a few words:
I've heard methods called indecisive if they don't return 1 winner--
if they often return a tie instead. Copeland's method, and other
methods related to it are indecisive in public elections, and often
require a tie-breaker. Copeland can't do other than return a tie
when there is a circular tie with 3 candidates, for instance--maybe
In public elections, Condorcet, Approval, Condorcet with the various
disapproval & NOTB features, Double-Complement, & Approval-with-1st-choice
are all decisive. Except that the disapproval or NOTB features could
be said to make the method indecisive if they prevent it from picking
any winner in an election.
As I've said, that could get the single-winner reform in trouble with
the public, if that happens, but not if they unnderstand that disapproval
or NOTB is completely separate from the new voting system, & that they
needn't go together, and if the public has been very insistant about having
those "reject everyone" features.
As for simplicity, as I've said, there's no way that we can say what's
too complicated for the voters. In Cambridge, they voters are quite
satisfied with an electoral system that's many times more complicated than
any of the single-winner methods that we propose, and is much more
complicated than the Condorcet's method would be even if fancier
features were added to it.
Cambridge, then, is an example of how we go wrong when we say that
something is too complicated for the public to accept. As I said,
the public themselves have to be the ones to decide that; it isn't for
us to decide it for them.
And as I've said, though Double Complement can be simply stated in a
definition for the public, explaining why it should be better than
ordinary Runoff & Plurality (which it's halfway inbetween) could be
a hard job. Justifying any claim that it is as good as Condorcet's
method would be impossible. Convincing people that they should do
without rank-balloting? I have no idea how that would be attempted,
when people know about the better rank-balloting methods.
Anyone intending to tell the pulbic how they can get significant
improvement over Runoff & PLurality with a method that's halfway
between them has quite an explaining job. In fact they'd have quite
a job jusifying such a claim to the Single-Winner Committee.
Anyway, to summarize: Decisiveness isn't a decisive standard for
choosing from the methods proposed to the Single-Winner Committee,
since all of those methods are decisive in public elections,
except that the disapproval & NOTB features could cause indecisiveness
if they disqualify all of the candidates--something extremely unlikely
to happen, considering the much better selection that will be available
to voters in a Condorcet election.
A standard based on simplicity depends on being able to say what's
too comlicated for the public, and I don't believe that's for us to
say. Who'd have expected the public anywhere to accept the complicated
I'm not saying that decisiveness & simplicity shouldn't be listed
as standards, and am only commenting on how they seem to me to
relate to our evaluation of methods. But I recommend not including
a standard that isn't violated by any of the methods that we propose,
since that just comlicates things for ER.