[EM] Wikipedia / ballot taxonomy

James Green-Armytage jarmyta at antioch-college.edu
Sat Jun 5 01:58:02 PDT 2004


Eric Gorr <eric at ericgorr.net> writes:
>Thinking about this some more and based on a private comment from 
>James, I think the only natural division that might actually garner 
>general agreement is to have a single division...multi-winner and 
>single-winner methods.
>Allow things like whether a method is one-vote or multi-vote, what 
>the ballot style is, etc. to be left to the pages for the individual 
>methods themselves.

For the purpose of the Wikipedia page, I agree. Why drive ourselves nuts
arguing about it? We should just change the page back to the way it was, I
think.

Still, my interest has been piqued in the question from an academic
standpoint. To begin with, I see three very basic questions that you can
ask about a ballot type.

1. Does the ballot allow as many "scores" as there are candidates, without
allowing more scores?
	If so, I'd call it a rank ballot. Condorcet and equal-rankings IRV are
examples.
	
2. Does the ballot allow more scores than there are candidates?
	If so, I'd call it a ratings ballot. Cardinal ratings from 0 to 100, with
5 candidates, is an example.
	You could further divide this category into 'natural', 'whole', 'integer'
and 'rational' number ratings, the latter allowing you to give a candidate
a rating of 69.443, or 0.315, for example.

3. Does the ballot allow fewer scores then there are candidates?
	If so, I'd call it a limited-ratings ballot. (Other possible names are a
limited-rank ballot, a limited-levels ballot, or simply a limited
ballot... etc.)

	So far, things are pretty straightforward. But it seems that there is
another question that cuts across these other questions:

4. For every score allowed, does the ballot allow more than one candidate
to receive that score?
	If not, and it's a rank ballot, then I'd call it an exclusive rank
ballot. Standard IRV would be an example.	
	If not, and it's a ratings ballot, then I'd call it an exclusive ratings
ballot. I'm not aware of any such method, but it would be possible.
	If not, and it's a limited-ratings ballot, then I'd call it an exclusive
limited-ratings ballot. Plurality would be an example. You can give
candidates a rating of either one or zero. However, while you can give
multiple candidates a rating of zero, you can only give one candidate a
rating of one.

	This exclusivity factor seems to be similar to what Tom was getting at
when he was talking about "one person, one vote" systems. However, I think
that his way of putting it is much more likely to be misleading.
	Oh, and I do think that combined systems make up a class of their own,
although we don't talk about them all that much. It would be possible to
have a ballot with both a ranked vote and an exclusive/limited plurality
vote, where you didn't necessarily have to vote for your first ranked
choice as your plurality choice...
	Anyway, not a matter of ground shattering importance perhaps, but that's
my two cents on it.

best,
James




More information about the Election-Methods mailing list