[EM] Comparing ranked versus unranked methods

[Forest Simmons]

On Mon, 4 Feb 2002, Adam Tarr wrote (among other words): 

> 
> I think that it is worth comparing methods that do not require ranked 
> ballots and methods that do require ranked ballots on separate 
> planes.  Since methods that do not require ranked ballots tend to be very 
> easy to implement, understand, and tally, they have a significant advantage 
> when pushing for reform on a small scale.
> 
> In single winner, this means basically approval voting or plurality 
> voting.  Are there any strategic advantages to plurality in this comparison?
> 
> In multi-winner, there are a wide variety of methods that don't need ranked 
> ballots.  I prefer open party list; some prefer closed party list or single 
> non-transferrable vote.  Plurality at-large is a pretty poor method by 
> almost all accounts.
> 

I wonder if you have thought about Proportional Approval Voting (PAV) that
we have explored on this list from time to time as an alternative to STV
PR.  It seems to have some of the same kinds of advantages over STV that
Approval has over IRV including, of course, the simpler ballot, but not
including polynomial summability. So like STV it has to be limited to
situations where summability is not an important consideration.

One of the best things about it is the way it rewards candidates that
appeal to groups having overlapping interests, or from the other point of
view, awards all groups with at least their fair quota of representation
and groups with overlapping interests with extra serendipitous
satisfaction. This tends to elect representatives with broader support
than those elected under STV, even though they may not have as many first
place preferences. 

As you say it is hard to make direct comparisons between methods based on
Approval Ballots and Methods Based on ranked preference ballots.  The
standard approach is to start with sincere Cardinal Rating ballots, and
then try to deduce the Approval and Preference Ballots for comparison
purposes.  Any other approach tends to give meaningless results in most
cases.

If you are interested in PAV I could point to some messages in the
archives or give you a brief outline. 

My initial interest in joining the list was in PR methods since I
considered the single winner case adequately resolved by Approval and/or
Condorcet. But I was surprised at the complexity of some of the single
winner issues and became fascinated by IRV psychology, etc. because of the
IRV initiative here in Oregon.
 

Forest