[EM] Guarding the Instant Runoff movement (Our Mike is mentioned)

[MIKE OSSIPOFF]

> > I said that as Condorcet is usually defined either all methods fail
> > it, or Plurality meets it.
>
>Plurality can be defined (and usually is defined in the academic
>literature) on preferential ballots.

No doubt anything can be defined in any number of silly ways.
I've seen a few definitions of Condorcet's Criterion, including
academic definitions, and they didn't say anything about
assuming that Plurality uses rankings. Maybe you're saying that they
were thinking that when they wrote the definitions, but then I suggest
that it would be better if they said what they meant. No, my statement
is accurate, that, as Condorcet's Criterion is usually defined,
either all methods fail it or Plurality passes it. But admittedly
I can only speak for the wordings that I've seen. Maybe the ones
you've seen specified that funny assumption, and maybe they're more
typical. Maybe, by some coincidence, all the academic definitions
that I've seen of Condorcet's Criterion were atypical, in not
saying that it's assumed that Plurality collects rank ballots.

>You claim that as plurality
>depends on LESS than the complete preferences of the voters
>plurality cannot be defined on preferential ballots.

No, I didn't say that it _can't_ be defined that way. I say that
I've never heard of it being defined that way. And I say that that's
a really silly way to define a method that doesn't use rankings.
And I also say that, though it can be defined in that way,
a criterion definition that requires a funny definition for one
voting system, or a few voting systems, is  a funky criterion
definition. A legitimate criterion definition would apply uniformly
to all methods, rather than having to specify a contrafactual
definition for one method.

Or maybe you're saying that the academics assume, for the purpose
of applying criteria in general, to all methods, that the input
of the voting systems is rank ballots. That sounds more like what
you were saying before. Aside from how nonsensical an assumption that
is, on the face of it, I'll repeat that you still haven't said
what uniform rule these academics of yours have for marking the
Purality ballots, or the Approval ballots, etc., based on the ranking
that the voters are assumed to vote.

So which is it? Do you say that the criteria require that Plurality
collects rankings, and counts all of the preferences for purposes
of determing whom the criteria require to win, but gives the voter's
Plurality vote to his top-ranked candidate? Or is it that, for
all criteria & voting systems, you determine compliance in a way
that assumes that the voter votes a ranking, no matter what the
voting system is, and then, from that ranking, you somehow decide
how to mark that voting system's own ballot, for the purpose of
determining who actually does win? And if that's it, then how
do you decide? What uniform rule to you use for all criteria &
voting systems?

So it isn't that I'm saying that it's impossible to define things
the way you vaguely refer to. I'm merely saying that you haven't
done so. Unless you answer the questions in the above paragraph,
your compliance-determining system that you refer to remains
undefined, and so does Beatpath GMC.

>But when
>you re-think your argument then you will observe that only when
>plurality depended on MORE than the complete preferences of the
>voters plurality couldn't be defined on preferential ballots.

No, you're missing the point. Of course a rank ballot tells whom
you're voting favorite. But how do you derive an Approval ballot
from that ranking? Or a single-winner Cumulative ballot? What's your
general rule? For Plurality you say the voter's top-ranked candidate
gets his Plurality vote. What do you do in Approval & single-winner
Cumulative? What's the general rule?

>I don't think that the fact that you cannot see that plurality can
>also be defined on preferential ballots demonstrates a problem of
>the Condorcet criterion or beat path GMC.

The problem is that you can't tell me what is the general rule
for getting the information that the voting system needs from
the voted ratings or rankings (the ratings or rankings that you
assume that people vote).

Or maybe there's no general rule. Maybe you just stipluate that
Plurality is dealt with as you specify. And do you have some other
special rule for Approval? For single-winner Cumulative? If so,
that's funky, shabby, & sloppy. Add those to your list of adjectives,
if you have to come up with a different contrafactual assumption
for each voting system in order to apply your system to all of them.
And, by the way, if you do use special assumptions for each of those
methods, then what are they?

Look, I'm not trying to make this discussion into a hostile argument.
I'm just asking you if you want to define the system that you've
been vaguely referring to.

So don't get so defensive. Either define the system that you refer
to, or take the time to go and look it up, so that when you talk about
it you'll be able to say what it is.

Mike Ossipoff

_____________________________________________________________________________________
Get more from the Web.  FREE MSN Explorer download : http://explorer.msn.com