[EM] Unranked IRV versus Approval

Anthony Simmons asimmons at krl.org
Mon Apr 2 22:03:16 PDT 2001

>> From: Tom Ruen
>> Subject: Re: [EM] Unranked IRV versus Approval

>> Anthony,

>> One-vote definitely makes a difference for
>> multiseat elections if you want PR. Unranked IRV
>> is cumulative voting w/o the elimination part.

Hmm, what I was asking is:  What is the significance
of "one vote per voter"?  How does it matter how many
marks a voter makes on the paper?  Basically, why
should anyone care?

It's a natural human tendency to accept what we have
heard unchallenged all of our lives, but it seems to
me that this business about "one vote per voter" is a
flat earth theory.

I mean, the actual vote isn't an individual mark.  If
it were, then you could turn Plurality in a "many
votes per voter" method by requiring voters to vote
against all but one candidate.  It would be logically
equivalent to the usual way Plurality is done, but
one of the equivalents satisfies "one vote per voter"
and the other doesn't.  Therefore "one vote per
voter" is not, as they say, even well-defined.  In
other words, it's a phantom.

"One vote per voter" never had anything to do with
how many marks are made on the ballot.

>> My biggest problem with Approval is that I'm
>> uncomfortable awarding victory if there are more
>> than one majority winner. If there are 2 majority
>> candidates, that means some voters "overvoted"
>> their compromises. We can offer a runoff, but
>> that's what we're trying to avoid!

Ah, I see the problem here.  People are trying to fit
reality into a conceptual framework that fits their
own thoughts.  If there is more than one candidate
that is approved of by a majority, that's a fact, and
no choice of election method is going to change that.

So, why is it a problem?  Is there a real reason?

>> I'd prefer Approval in decisions/elections that
>> allow runoffs and no elimination. Just let voters
>> start with their full set of favorites and keep
>> voting, slowly reducing their choices as they are
>> comfortable and declare a winner when exactly one
>> candidate exceeds 50% or 60% or whatever. But
>> that's not going to work in big elections. Maybe
>> it won't even work in small elections, but
>> actually consensus is the real test for agreement.
>> In theory you could demand 100% support for one
>> choice for victory. Approval runoff is less
>> competitive than plurality runoff and therefore it
>> seems more able to find a consensus choice with
>> less hard feelings.

>> Overall for political elections I see that it is
>> plurality that is failing us and runoffs are the
>> solution people are used to. I really haven't
>> heard anyone, outside of election methods,
>> complain that runoffs fail them.  [...]

Well, there have been a few, like, IIRC, the
Australian Electoral Commission.  When you hear about
IRV outside of this group, who do you hear from?

>> [...]  Without ranked
>> ballots, you can't even measure properties like
>> Monotonicity.

>> I would be interested in experimenting with
>> approval in practice, but I'm not willing to
>> impose it on anyone. For me it is an experiment
>> only.

>> I do plan now that when pollers call me in the
>> future and ask who I support, I'm going to make an
>> effort to pick two choices I like. Then I expect
>> they'll say they'll put me down as undecided and
>> I'll say NO, I'm decided FOR these two and AGAINST
>> all others. Perhaps it'll get them thinking of
>> making new categories and approval will be born!
>> Well, I can dream.

>> That's my thoughts for now, whatever they are
>> worth.

>> Tom

More information about the Election-Methods mailing list