At 08:07 AM 4/26/2010, Kristofer Munsterhjelm wrote:
Abd ul-Rahman Lomax wrote:
At 11:57 PM 4/24/2010, Kevin Venzke wrote:
Hi Abd,
--- En date de : Sam 24.4.10, Abd ul-Rahman
Lomax <[email protected]> a écrit :
> This is what is common with the
> use of voting systems criteria to study methods. Scenarios
> are created, sometimes cleverly, to cause a failure of a
> criterion. Does it matter if those conditions never exist?
> It should.
For the simple question of whether the criterion is satisfied or failed,
no it doesn't. Of course people then do go on to disagree about whether
certain criteria are important, and why. There is nobody who thinks every
single criterion is important.
That's right. But until utility analysis
started to be done, the arguments had
practically no foundation, they were just ideas
about what democracy should look like,
sometimes intuitions, and sometimes quite
deceptive. Some criteria may be positively
harmful, and Later No Harm is one of those. No
method that maximizes utility can satisfy Later
No Harm, no method that finds the best compromise winner can satisfy it.
And no method that maximizes social utility,
overall satisfaction, can satisfy the majority
or condorcet criteria, as fundamental as they
seem, when only a single ballot is used. They
can by using a second ballot to ratify (or
reverse) an original election that finds the utility maximizer.
To me it seems that wouldn't be the case either.
Consider the case of a top-two runoff (with
Range as the first round) where the two winners
are both supported by a minority. Then no matter
who wins the runoff, the method as a whole has
failed the Majority (and Condorcet) criterion.
The criteria are not designed to apply to
multi-round elections. You can't even tell from
the information given. What if more voters vote
in the runoff? It happens, you know. A runoff is
a separate election, that's very important to
understand. If write-ins are allowed, or if the
plurality preference is clearly in the runoff, as
two examples, the Condorcet and Majority criteria
are satisfied, assuming that the runoff method
satisfies them. Bucklin doesn't satisfy
Condorcet, technically, but in practice it does,
I'm pretty sure. It does satisfy the Majority criterion, clearly.
Unlikely? Perhaps, but one failure is enough. Of
course, you could then argue on basis of social
utility (as you have), but you can't say the method passes the criteria.
You had two elections. The first one failed. The
votes in it may determine ballot position, but
they do not have any effect on the winner. They
are then like any nomination rules, for example
petition signatures or party endorsements. We
don't claim that a method fails Condorcet because
the nomination processes don't allow a condorcet
winner to appear on the ballot!
One could also formulate criteria based on
score, for instance: "if a candidate X is given
more than half the points given by voters in
that election in total, he should win" - a score/Range version of Majority.
I'm actually proposing using a Range ballot for a
Bucklin method as primary, then examining the
ballot, if there is majority failure, for various
kinds of winners, and then including the
important ones on the runoff ballot. I do
advocate allowing write-ins on the runoff ballot,
so there are actually *no* eliminations, just a
kind of suggestion. It's rare, but write-ins do
win elections, sometimes. A situation where a
Condorcet winner, by some quirk, got eliminated
is a case where it quite well might happen; in
this case, there would be good data from the primary.
Because I'd use Bucklin for the runoff as well,
it's conceivable that a Cndorcet winner could
fail there, but this can only happen when there
are excessive approvals by the supporters of the
Condorcet winner. Essentially, it's either bad
strategy -- and it's bad strategy when there is
no excuse, the Condorcet winners supporters
should know that this is, indeed, a Condorcet
winner in the primary -- or it is a case of small
preference strength in the votes for the
Condorcet winner, compared to large preference
strength in the votes for the actual winner of
the election. In other words, this is a case
where the Condoret winner was not ideal.
That's rare, but if we somehow had the magic
perfect voting system, it would find this
situation and would thus fail the Condorcet
criterion. The question I ask: is, then,
Condorcet failure of this kind a Bad Thing, to be avoided?
But ordinarily the Condorcet Criterion is quite
important. That's why I'd like to make sure tha
this winner, if there is majority failure, ends
up on the runoff ballot. It allows those
supporters to make a real, informed choice. For
the Condorcet winner to fail, they must accept
it, they must, at least, stand aside and not
insist. If they persist, and if the Condorcet
winner remains such through the expression of
exclusive preferences, which requires that their
preference strength be strong, they will prevail.
----
Election-Methods mailing list - see http://electorama.com/em for list info
