[EM] SD only as a replacement for IRV

[Bart Ingles]

MIKE OSSIPOFF wrote:
> 
> Incidentally, though PC uses rank balloting, as does IRV--PC, like
> all Condorcet versions, is incomparably easier to implement than IRV.
> IRV requires far much more computation time and computer memory
> use, compared to PC (or any Condorcet version) or Approval. That's
> because, unlike those methods, IRV has to keep all the rankings in
> memory and process them all in one big long operation. The current

Or re-read the ballots every time a candidate is eliminated...

I also wonder how state or federal IRV results would be tabulated if
counties are to continue to do the actual counting, as they do now. 
Each county would have to determine its first-choice vote totals, then
forward them to the state, which would then decide which candidate was
eliminated, so that the counties could calculate the next-round totals,
etc.

-Bart


> problem in Florida underscores how IRV's much greater computational
> task will be a security & verification nightmare.
> 
> Mike Ossipoff