[EM] Condorcet/Range voting?

[Mike]

I got an interesting email from Warren D. Smith a couple of days ago 
concerning range voting, and I was wondering if anyone had considered it 
as a Condorcet completion method, kind of like how Borda is used in 
Black's method. I did a Google search for "condorcet range," ("range 
voting" "condorcet completion") and a couple of variations but didn't 
see anything). It seems like it should work pretty well -- even using 
maximum strategy, range voting is equivalent to Approval, but voters can 
provide more information on preference rankings than either Approval or 
Borda would allow.

To make it work right, you'd probably have to do something like the 
following:
1. Using ordinal rankings, see if there is a Condorcet winner. If there 
is, this is the overall winner.
2. If there is no Condorcet winner, take the Schwartz set and normalize 
the range scores in this set for each voter from 0 to 1.
3. Use these scores to fnd the range voting winner.

Basically, you'd be looking at ordinal rankings, and if that was 
insufficient to find a winner you would use normalized cardinal rankings 
(that's so a person with a 0-1-4 ranking of the Schwartz set candidates 
does not have less of a say than the person who used 0-25-100).

Since someone had probably already discovered all of the properties of 
such a method (it's really hard to come up with something new), if there 
is a link or a good description of this method the web, just let me 
know. I'm especially interested in what voting paradoxes come up with 
it, and comparisons with other Condorcet completion methods. Thanks!

Mike Rouse
mrouse1 at mrouse.com