[EM] Re: paradigms...

Rob Brown rob at karmatics.com
Wed Sep 8 18:28:24 PDT 2004


Paul Kislanko <kislanko <at> airmail.net> writes:

> To which I reply "you are entitled to your opinion, but if you cannot prove
> that all orderings of n-1 candidates by a single voter will be consistent
> with the orderings of n candidates by THE SAME voter for ALL voters, then
> your opinion doesn't count."

Count according to who?  

First of all, we are talking about human preferences, and I am not denying 
that a human could have contradictory preferences.  Anyone who has ever had a 
girlfriend knows that.

> I might think my hypothetical voter is illogical, too, but that's not the
> question. The question is can you reconstruct the original ballots from a
> pair-wise matrix? If not, then you can't claim a result based upon the
> pair-wise matrix is the "will of the people."

I never claimed it was.  Doesn't Arrow supposedly prove otherwise as well, and 
we accept that?

I think in order to have a perfectly accurate "will of the people" system, you 
would have voters assign a score or weight to each candidate, so that they can 
say just how much they prefer A to B.  However, because this would be prone to 
strategic manipulation, you must use a lie detector to make sure that they are 
being sincere.

But having a lie detector requirement is too costly and expensive and privacy 
infringing.

Likewise, having a ballot where you can express all pairwise preferences is 
not worth the benefit afforded by being able to capture this additional 
information from voters.  It would be exceptionally cumbersome to use.  I've 
done the best I can at making an easy to use interface for ranking candidates 
( http://www.karmatics.com/voting/ as well as 
http://article.gmane.org/gmane.politics.election-methods/5460 ), do you have 
any suggestions for how to make a ballot that expresses full pairwise 
preferences -- and that is something that people will find usable enough to 
accept?

Because if your method works great in theory, but it never gets adopted for 
any real election because it is cumbersome and confusing and is therefore 
rejected by the public....well then, in my opinion, it doesn't count either.




More information about the Election-Methods mailing list