Only 5 candidates needed for SD nonmonotonicity

[MIKE OSSIPOFF]

EM list--

Isn't it true that Narkus's SD nonmonotonicity example would
work with just 5 candidates? I've been mistaken before, but
it seems to me that the EFG cycle isn't needed. Keep the same
defeats, but leave out candidates F & G, and add the defeat
EA14. A still wins in Act I, and, in Act II, when DE13 is dropped,
CD15 is still the weakest defeat that's in a cycle.

Though nonmonotonicity doesn't seem to cause trouble in a
practical sense, it is ridiculous just the same, when you make
one change to your ranking, and the method reacts oppositely.

The academics really hate nonmonotonicity, and some consider it
IRV's worst problem, something that I never agreed with. But I've
never heard of any academic publicly talking about IRV's
nonmonotonicity to the public where CVD has been proposing IRV.

Though Tideman & SSD seem, merit-wise, the best rank methods
for public proposals for political elections, I like to offer
several alternatives to IRV, when opposing IRV. In case someone
could be more likely to like something with a super-brief
definition, SD & PC are still worth including. SD, so far as
I'm aware, meets every criterion that IRV does, while still
meeting BC, and all that goes with that. SD's nonmonotonicity
doesn't change the fact that it meets every criterion that IRV
meets. So, just in case the briefest-worded alternative methods
are the ones best able to beat IRV, SD is very much worth including
in the list of alternatives sent to people to whom IRV is proposed.
Along with Approval, Tideman, SSD, & PC. PC wouldn't fare as well,
though, because it fails criteria that IRV meets--Condorcet Loser,
Majority Loser, & Mutual Majority.

But when it's just a matter of replacing Plurality, and IRV
isn't proposed, then PC seems the best super-simple Condorcet to
offer. It fails some criteria that the academics like, but
Plurality fails those criteria too, when they're worded in a way
that fairly tests Plurality. So PC meets everything that Plurality
meets, except for some criteria failed by every rank method better
than Borda. Participation failure, for instance, can be excused
as unavoidable for an adequate rank-method, because it isn't
completely unreasonable that one's various pairwise preference
votes can act against eachother.

When writing to people to whom IRV is being proposed, I
present Tideman as most likely the best, with SSD & Tideman as
the best two, between which the merit comparison is really
a matter of opinion.

Mike Ossipoff



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