[EM] equal rankings IRV

James Green-Armytage jarmyta at antioch-college.edu
Sun Jun 13 14:55:02 PDT 2004


I wrote:
>
>> 	Once again, in the whole votes version, if I equally 
>> rank A B and C (in first place, for example), my vote will 
>> count as a whole vote for each of those candidates in the 
>> first round.

James Gilmour replied:
>
>So if you do this but I do not, you get three votes while I get only one?

I wrote:
>> In the fractional votes version, my vote will 
>> count as 1/3 of a vote for each of them at first, then when 
>> one is eliminated, 1/2 for the remaining two, and later as a 
>> single whole vote for whoever lasts the longest.

James Gilmour replied:
>
>Now everyone has only one vote.

I reply:
	Yes, I already implied that this might be an issue in the e-mail that
James Gilmour is replying to. (I wrote "This may come into play if people
are worried about some one person one vote mumbo jumbo and hence reject
the superior (?) whole votes version.")
	My feeling is that this is a rather superficial issue, but one which
might cause legal problems, 'cause, you know, some bureaucrats are either
superficial-minded or they're happy to use any reasonable-sounding excuse
to knock down something that would cause a change in the routine, or both.
	The one person one vote issue here is equivalent to the one we have in
approval, and for that matter, in Condorcet's method. 
	That is, yes, you can give more than one candidate a whole vote, or you
can just give one candidate a whole vote. The reason that this is NOT
unfair is that, in each round, you can only add one vote to the score
differential between any two candidates. 
	Do you know what I mean? If I vote for Kerry and Nader in first place,
and Bush in second place, I am adding one vote on Kerry's side of the
Kerry-Bush vote margin. I'm also adding one vote on Nader's side of the
Nader-Bush vote margin. I'm adding zero votes to the Kerry-Nader vote
margin. Now, let's say that Nader is eliminated first. In the second
round, my vote counts as one (not two) whole votes in favor of Kerry. That
is, once again, I'm adding one vote to the Kerry-Bush vote margin. Which
is totally fair.
	But what if I had strictly ordered my rankings Nader>Kerry>Bush? In the
first round, I'm adding one vote to the Nader-Bush margin. I'm also adding
one vote to the Nader-Kerry margin. I'm adding zero votes to the
Kerry-Bush margin. So in each case, the maximum amount of difference that
I can make between the vote margin of any two candidates is *one*. As in,
one person one vote? Maybe... at least that's how I would argue it to the
Podunk State Supreme Court if it came to that. 
	So, unless you can come up with an example where ER-IRV(whole) seems to
be really unfair, please acknowledge it as being fair.
	By the way, as far as I know, ER-IRV (either version) meets most of the
groovy criteria that IRV meets, such as mutual majority, dominant mutual
third, independence of clones, and whatnot. But people are very very
welcome to challenge me on that, point out criteria where plain IRV passes
but ER-IRV fails.
	And anyway, yes, ER-IRV(fractional) does meet one person one vote in the
more generally accepted way. And thus, maybe it is the more politically
viable. A compromise from a compromise... maybe a little too much
compromise for me. But I still suppose that the fractional version is an
improvement from plain IRV. Any more comments on fractional versus whole?

best,
James






More information about the Election-Methods mailing list