[EM] corrctions to older psts re IRV public election data

[Scott Ritchie]

On Fri, 2005-11-11 at 21:35 -0500, Warren Smith wrote:
> Arguably STV multiwinner electiosn are still of interest for single-winner
> purposes since the FIRST winner is a single-winner IRV winner.
Not really.  Consider the following:

300 voters, 2 winners (Droop quota of 101)
101 A
99  B, C
100 C

Here the first STV winner is quite clearly A, but A is not the IRV
winner.

> Incidentally the comment by somebody that non-montonicity spotting in IRV
> elections is NP-hard, is misleading.  That result is only true in an unrealistic
> limit where both the number of candidates & voters tnd to infinity.
> 
> If the #candidates is held fixed nd the #voters is made large (more realistic) then
> the task is fully in polynomial time.  MOst of these NP-hardness results in
> voting theory are almost completely uninteresting for this exact reason.

I'm sorry, but this explicitly conflicts with the conclusions of the
paper I linked to here, where the set C of candidates and V of voters
are most certainly finite numbers:
http://www.isye.gatech.edu/people/faculty/John_Bartholdi/papers/stv.pdf

Could you clarify?

Thanks,
Scott Ritchie