Concluding Tobin Reply

[Mike Ositoff]

Let me just clarify a few things I may have left vague, or add
a few things that I may have omitted, when I replied to Tobin
about margins & standards.

First, Hugh, we could save ourselves a lot of trouble. I told you
why Young fails our standards & criteria. Though those violations
have truncation & order-reversal as their mechanisms (and truncation
seems to happen in every rank-balloting election), truncation &
order-reversal are important only as mechanisms of failure rather
than as primary standards. So it's missing the point to direct the
discussion to the mechanism while ignoring the primary standards.
You can disagree about how Young or Condorcet does with regard to
those standards, or you can explicitly say why your standard is
more important to you (and exactly what it is), and, why it should
be more important to us. But if you're going to argue for 
Young vs Condorcet, don't dodge the standards issue.

As Demorep has pointed out, Arrow, and Gibbard & Satterthwaite
have pointed out, in their own way, that there's no perfect method,
if we're talking about methods that only use 1 balloting and whose
only input is a ranking from each voter. Methods of that type aren't
the only kind. Steve & I have proposed several enhancements that
make much better methods possible. Still, methods allowed by that
limitation, which I'll call "simple methods", are the basis for the
enhanced methods that Steve & I proposed here.

Since no simple method can be perfect, let's not attack Condorcet(EM)
because it isn't perfect. Order reversal can be a problem under
certain conditions that probably aren't relevant to public political
elections? Sure, no method is perfect. Conceivably someone could
benefit from insincerely ranking his top choices equal? Sure, no
method's perfect.

But, though this is beside the point, as I said at the beginning
of this letter, since standards are the real issue, I'd like to
reply to a few especially unreasonable criticisms of Condorcet
based on order-reversal and incentive to rank equally. No, I'm not
going to repeat the order-reversal discussion that I recently posted.

Hugh said that, one shouldn't be concerned about Young having a
truncation problem when Condorcet(EM) doesn't, becaues both
methods have the same order-reversal problem. Well, he implied that
their order-reversal problem is the same. Nothing could be farther
from the truth. It will sometimes be necessary, with Young, to
protect against order-reversal by engaging in _defensive_ order-reversal.
By ranking a less-liked alternative over a more-liked one. This is
in stark contrast to Condorcet(EM), where it isn't even necessary
to rank a less-liked alternative equal to a more-liked one. 

Yes, yes, I realize that Markus showed an example where voters could
benefit by insincerely ranking their top choices equally. But that
example was never represented as other than a chaotic natural
circular tie. I claim that what happens under those conditions isn't
nearly as important when there's a Condorcet winner, a lesser-evil
to protect, an LO2E problem for the method to avoid.

As Markus agreed, Schlze & Tideman also can give incentive to rank
one's top choices equal insincerely.

What about Young? Young can force you to insincerely rank your top
choices equal in order to protect truncation (innocent or strategic)
from defeating a Condorcet winner and electing the truncators'
candidate. So it's Young's method, not Condorcet(EM), that has that
problem when it can do real harm, when there's an alternative that
we'd all agree should win (nearly all of us).

So the fact is that Young is the method that has serious problems
in the 3 areas that Hugh named: truncation, order-reversal, and
incentive to insincerely rank one's top choices equal.

But as I said, this is all beside the point. We like Codorcet(EM)
(with or without the Smith set) because it meets standards that
are important to us and which concern voters. If you want to argue
for a different method, show that it's better by our standards, or
tell us why we should consider other standards to be more important.
As I've said, though, as long as LO2E is so important to voters,
it will remain important to electoral reform advocates, I suspect.
Young thoroughly fails our standards.

Mike