[EM] Center for Range Voting Formed

[Adam Tarr]

Let me stress that the important debate is the stuff I mentioned in my
previous method.  That's the meat.  This stuff is just definition
crap, which always manages to raise people's hackles, but in the end
is not all that relevant.

Warren Smith wrote, in response to Rob L:

> >Robla: However, the Condorcet winner criterion is quite easily and
> unambiguously applied to Range Voting ballots, since a ranked ballot can
> be easily derived from a Range Voting ballot.
> 
> --that isn't fair because a range voting ballot cannot be derived from
> a ranked ballot.  This is a one-way noninvertible transformation.

Two responses here:

1)  Who says Condorcet has to be done on ranked ballots?  My personal
favorite ballot is the graded ballot: the voter ranks every candidate
from A (or A+ if we want more slots) to F.  I like this ballot because
it is very intuitive and obvious whether A or F is the "highest"
ranking.

This is, or course, a rated ballot, not a ranked ballot.  But it can
be used for any ranked ballot voting method that can support equal
rankings.

2)  (voice of Peter Falk in the Princess Bride) "Who said life is
fair?  Where is that written?"

Seriously, who the hell cares that it's a one-way transformation?  OK,
so information is lost.  Big deal.  Is squaring a number unfair, too?

The point is, we can do it, and we can draw valid conclusions about
the method by applying criteria to that transformed ballot.  The onus
is on you (or any other advocate of a method that only applies to
rated ballots) to prove that the information that was lost is useful
in some sense.

In this case, it's clearly useful in the case where voters can be
trusted to be honest and fully expressive, but not conclusively useful
in other cases.

At any rate, none of this prevents us from making the transformation
and checking the method's ability to satisfy certain criteria.

> --as I described last post, the Condorcet winner criterion is quite easily seen
> NOT to be unambiguously applicable to Range Voting ballots, and you have to make
> a choice of definition.

No, it is unambiguous.

What you are doing, in reality, is creating a NEW criteria, which only
applies to range ballots.  Now, that is your prerogative.  You can
call it the "range Condorcet criteria" or somesuch, if you want.

But the Condorcet criteria is already well-defined, and you don't get
to change the accepted understanding of a DEFINITION because you don't
like it.

As a math major, you surely understand that without agreed-upon
definitions, we may as well be arguing in different languages.  Define
your criteria.  Call it whatever you like.  But it is not the
Condorcet criteria, which is already well-defined and cannot be
changed on your whim.

> You can either tell your electorate "I, Rob L. demand that in the re-vote,
> you STAY with the same votes, dammit, no strategy changing allowed"
> in which case, range voting is a condorcet method.  Or you can tell them
> "I, Rob L. demand that in the re-vote, you MUST be strategic but MUST
> have been honest in the first vote" in which case range is not a condorcet method.

The reality is neither.  The reality is that the existing ballots can
already be interpreted as ranked ballots, and all pairwise contests
can be derived thereof.  No additional voting is required.

> Time to admit I'm right :)

Your right about your own definition, but it's not the Condorcet
Criteria.  Make your own criteria, write your own book.  But you don't
get to redefine that which is already well-defined and generally
accepted.

-Adam