Splitting the votes

[David Catchpole]

On Wed, 7 Oct 1998, David Catchpole wrote:

> On Wed, 7 Oct 1998, New Democracy wrote:
> 
> > Dear David Catchpole,
> > 
> > You wrote: "2) It should be hard for any candidates to split the vote."
> > 
> >      No, I disagree. It is acceptable for candidates to split the vote of
> > their party - as long as there is a way for those votes to come back
> > together.
> 
> By "split the vote" I mean a retraction of a/group of candidate/s would
> result in an advantage/disadvantage to other candidates. Obviously
> "split" comes from the nature of the phenomenon as regards FPP. I
> apologise to people if they inferred that I was referring only to that
> phenomenon (It was late, and I was... um, not quite up to it). "Allowing
> votes to come back together" means "splitting" as I define it does not
> occur in the first place.
> 
> 
Woops! There should also be a point that all rank-ordering methods have
the occasional potential for a "split," still. The only safe result is a
Condorcet result (if it exists) and exclusion methods are particularly
hairy. Check out Fishburn, et al in Lijphart's "Choices of Electoral
System" (or a title like that...) which has a nice'n'screwy example of an  
election which ordinary STV seriously defiles. Though I don't agree with
Fishburn's assumptions from it, the example is a good one for frightening
oneself.