[EM] Fast Condorcet-Kemeny calculations -- in polynomial time

Jameson Quinn jameson.quinn at gmail.com
Thu Dec 22 00:27:18 PST 2011


2011/12/22 Richard Fobes <ElectionMethods at votefair.org>

>  But as far as I know, the algorithm always finds the highest score, even
> in the most complex cases.
>

If you aren't 100% sure you have the right answer, you probably don't have
the right answer 100% of the time.



> Yes, a _proof_ that the highest sequence score has been found may be
> NP-hard.  Yet, as the calculation descriptions point out, that becomes an
> issue only in situations that are like finding the highest sand dune in a
> desert.  In such cases different experts will argue about which candidate
> really should have won.
>

I understand that you're claiming that the only cases where your algorithm
might not give the right answer are unrealistic cases where the "wrong"
answer is not actually very wrong. Still, if you can't prove you have the
right answer, you possibly don't.

So your own claims contradict themselves. It would seem that you do not
have a polytime algorithem for finding the Kemeny-Young winner, but just
for probably finding that winner. I would not be surprised if your
algorithm was right >99.9% of the time. But if the alternative when your
algorithm is wrong is a 10% chance of a civil war which kills millions,
then that still could be an unacceptable risk. And your claim to have
solved the problem, when you haven't, actually reduce my confidence that
you've even solved it 99% of the time.

Jameson


> Richard Fobes
>
>
>
>  2011/12/21 Richard Fobes <ElectionMethods at votefair.org
>> <mailto:ElectionMethods@**votefair.org <ElectionMethods at votefair.org>>>
>>
>>
>>    As previously promised, I am revealing how Condorcet-Kemeny
>>    calculations can be done fast.
>> ...
>>
>
>
> ----
> Election-Methods mailing list - see http://electorama.com/em for list info
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.electorama.com/pipermail/election-methods-electorama.com/attachments/20111222/65088615/attachment-0004.htm>


More information about the Election-Methods mailing list