[EM] Newbie de-lurks to propose a clone-proof Borda method

[Ken Kuhlman]

I wrote: 

> "Clone-ness" thus can't be a binary factor that is either "on" or "off," 
> but a continuous function.  


I fear this isn't blunt enough, so let me restate: 

The fundamental problem with a plain Borda count isn't clones per se, it's 
that it doesn't directly take into consideration how candidates do relative 
to each other on individual ballots.  (Likewise, the fundamental problem with
plain Condorcet is that it doesn't directly take into consideration which 
position candidates are ranked in).   

A correlation component seems to me to be the simplest way of adjusting 
the Borda method for how pairs of candidates tend to fare against each other. 

The more time I work with it, it also seems the best. But, I've gotten too 
close to the proposed CIBR solution to be able to be certain that my 
objectivity hasn't been compromised.   Thus, I'm asking the help of the group 
to vet the proposal before I sink more time into it. 


The biggest problem I've found so far is that the method further complicates
the problem of determining what to do with partial ballots. (What's the clone
component of a ballot that doesn't specify a full ranking?  Do we need to 
spread the votes across the remaining sub-permutations, or is there a 
shortcut?)


If anyone's interested, I have an implementation of the method that I'd be 
glad to share.   It's too large to post to the list, so mail me directly if 
you'd like to help & want a copy.    Please continue to direct general 
comments/criticism to the list, however.

Thanks,
-Ken
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.electorama.com/pipermail/election-methods-electorama.com/attachments/20050525/32eab9fd/attachment-0003.htm>