Boost logo

Boost :

From: chaos_at_[hidden]
Date: 2005-04-12 09:02:41


> Joel wrote:
>
> I'm not sure about votes like:
>
> vote:
> 37
> 37
> 37
> 37
> 37
>
> and votes like:
>
> vote:
> 50
> 50
> 83
> 83
> 75
>
> Does the IRV voting allows listing the same candidate several times? Will
> this skew the resuls?

Yes, and no, respectively. Only the top vote is counted, and if that
candidate is eliminated, all of the votes for the eliminated candidate can
no longer be counted. Essentially, that person is forefitting some of
their choices.

>
> Also:
>
> vote:
> 0
> 75
> 75
> 38
> 1
>
> Does not look correct. Don't we number candidates from 1?

Yes, but since zero is a one time fluke it cannot possibly win. It will
be eliminated toward the very beginning of the counting and will really
just delay the person's real votes from counting. Can't really hurt the
final result at all.

> OTOH, the attached Python script I've just wrote by looking at
> http://www.electionmethods.org/IRVexample.htm produces the same results:
> '67' is the winner and '75' is the runner-up.

Thanks for taking the time to double check my program's results. Glad to
see they came out the same :)

-Jason


Boost list run by bdawes at acm.org, gregod at cs.rpi.edu, cpdaniel at pacbell.net, john at johnmaddock.co.uk