[EM] CVD wants Alt.V to be fairer but it isn't: misleading website

[MIKE OSSIPOFF]

Markus said:

Does that mean that you want the CVD to promote FPP?

I reply:

You didn't ask me, of course, but I'd like to answer anyway. It would
be an improvement if CVD would stick with FPP. That's partly because
FPP meets Participation and IRV doesn't, and partly because IRV
can be expected to reliably do worse than FPP when candidates' support
tapers gradually away from the voter median, as in the following
example:

60: ABCDE
70: BCADE
100: C
83: DCEBA
75: EDCBA

The incoming votes from the extremes build up the vote totals of
nonmedian candidates. Merrill calls that the squeeze effect, and it's
the reason why IRV does so much worse than Approval by SU index.

Obviously there's lots of room to change those numbers, and the same
thing will happen. All it takes is candidates' support tapering
gradually away from the voter median candidate.

So it seems to me that FPP is better than IRV.

IRV, in examples like that, fails to elect the SCW, even when he's
the favorite of more people than any other candidate is. At least
FPP won't do that.

Summary: FPP is better than IRV because of the above kind of example,
and IRV is worse still because it poses as a reform. Better to keep
FPP till we can get Approval or Condorcet.

In local elections we here use Runoff. Runoff is another method that
is much better than IRV. The SCW is sure to win if he comes in 1st or
2nd in the 1st balloting. Runoff does significantly better than IRV
in SU.

******

Marcus continues:

I consider the independence from clones criterion to be more
important than the participation criterion. Therefore I prefer
IRV to FPP.

I reply:

Can I call that IFCC? Doesn't IFCC say that if we delete from the
ballots 1 or more candidates from a clone set, and recount the ballots,
that shouldn't change the matter of whether or not the winner comes
from that clone set?

That doesn't seem very useful. If we turn it around and add clones,
and hold a new balloting after adding them, the criterion only applies
if no one changes their vote.

How likely is it that no one would vote differently if new candidates
are added? A criterion that assumes that seems real useless.

The IFCC would be more useful if it spoke of adding candidates and
holding a new balloting in which voters don't necessarily vote the same
way, but whose votes or vote-changes are somehow constrained.

For instance, maybe they always vote to maximize their utility
expectation (we then have to specify utilities in our examples),
or maybe they just always vote in an undominated way. Maybe, then, they
don't change their vote unless to not do so would result in a dominated
strategy. These are just possibilities that I mention. But saying that
they don't change their vote is entirely unrealistic.

So say a method meets IFCC. Would it meet a more realistic version?
Maybe not, and so I suggest that meeting IFCC doesn't mean a whole lot.

Mike Ossipoff



Markus Schulze
schulze at sol.physik.tu-berlin.de
schulze at math.tu-berlin.de
markusschulze at planet-interkom.de



_________________________________________________________________________
Get Your Private, Free E-mail from MSN Hotmail at http://www.hotmail.com.

Share information about yourself, create your own public profile at 
http://profiles.msn.com.