[EM] condorcet definition dispute

[Warren Smith]

> WDS: ...
> If [the Condorcet winner criterion] means (2a) "if we redid the same
> election but using a DIFFERENT election system, namely majority vote,
> and demanding votes `logically consistent' with the originally-cast
> votes, then A would win"

>Robla: This is what it means.  I'm quite confident that the literature will
back me up on this.

--REPLY BY WDS:
dear Rob.  Pelase let us know if you ever find a single instance, in any literature
whatsoever, of the condorcet criterion being used to consider a class of voting
methods in which the votes are more general than ranked-preference-order ballots
(or partial-orders) i.e. in which it is not possible to derive those votes from their
ranked-abbreviations.

I have never encountered any such example in the literature.

Therefore, I am creating my own literature.

Because it is necessary to now do so.
I repeat, the condorcet criterion, and many other criteria EM often discuss, are
naturally uniquely defined only for ranked-ballot type voting
methods.  I do not think there should be (nor is) any dispute on that.
When one attempts to apply those criteria to more general kinds of voting methods, often
you need to think more and define more and reconsider the definitions, etc etc.


---

Now let me elaborate.  Consider the word "DIFFERENT" in the start-portion of this
email.  That word was not necessary to use, IF you only are considering
ranked-ballot methods, because every ranked-ballot method (except perhaps
for something crazy nobody would bother to consider) reduces to majority vote
in the 2-candidate case.  Therefore, generally nobody ever bothers to write it down.
If you do not write it down, then range voting is a conodrcet mrthod.
If you do write it down, then range voting is not a condorcet method.

So now you see the wellspring of the definitional difficulty, and this was a
difficulty/ambiguity which simply never occurred to Condorcet, but which
does occur to us when we try to think about range voting.

ok?  I hope this is clear now.
wds