[EM] Election methods in student government...

[Kevin Venzke]

Tim,

--- Tim Hull <thully at umich.edu> a écrit :
> OK, wasn't sure about Condorcet - I knew that to meet the Condorcet
> criterion with a voting method you had to violate later-no-harm, but not
> that the finding of a Condorcet winner in and of itself violated LNH.
> Thanks for clearing that up...

Well, perhaps you can merely *find* the Condorcet winner, and somehow
use that information, without failing LNHarm. But if you regularly elect
this candidate, then that's the same as satisfying the Condorcet 
criterion.

> Regarding the single winner methods, it seems that IRV or MMPO may be the
> way to go there if one wants to maintain later-no-harm.  Range seems to
> be a
> good choice with respect to non-LNH compliant methods, and reweighted
> range
> for multi-winner.

I like approval (which fails LNHarm, but it is at least quite clear how
and why additional choices may hurt your other choices) and I would
have to tweak MMPO to recommend it.

> I'm curious - is there any other multi-winner PR methods (besides STV)
> that
> satisfy later-no-harm, exclusing variants of party list and asset?  Also,
> what tie-breaking methods for IRV/STV/MMPO satisfy LNH?

For MMPO (where ties are a big issue) the most obvious tiebreaker would
be first preferences. You could also do random ballot.

MMPO elects the candidate against whom the greatest pairwise opposition
is the least. Here's an extreme situation of this:

1000 A
1 A=C
1 B=C
1000 B

C wins. I suggest less sensitivity to pairwise opposition scores of
under a majority. I recently suggested (as the "MTR" method) a version
that reduces all scores to "majority" or "less than a majority" and
resorts to the first preferences tiebreaker more often.

> I figure that the decision mostly comes down to whether LNH or
> monotonicity
> are more important in our elections.  I know many people will bullet vote
> -
> probably more than in a hypothetical national election - if the voting
> system fails LNH.  However, monotonicity is also a concern...

Among single-winner methods, FPP, DSC, and MMPO are all monotonic.

For multi-winner methods I think a party list system would be easiest...
If you're concerned that a given voter is represented proportionally
in the result then I suppose there are various positions on policy
around which party lists could be formed. If not, if it's just an enormous
popularity contest, then the efforts to be proportional seem to be a
bit unnecessary...

Kevin Venzke


	

	
		
___________________________________________________________________________ 
Découvrez une nouvelle façon d'obtenir des réponses à toutes vos questions ! 
Profitez des connaissances, des opinions et des expériences des internautes sur Yahoo! Questions/Réponses 
http://fr.answers.yahoo.com