[EM] full ranks in MDDA (not)

[Kevin Venzke]

Warren,

--- Warren Smith <wds at math.temple.edu> a écrit :
> But I still feel unhappy about the whole idea (which I think Rob L is also
> unhappy about) that in these systems voters (i) have incentive to truncate their
> ballots (ii) have incentive not to, and they have to tradeoff these two strategic
> incentives.  It would be better if voter always has incentive to give more
> info.

Then I think you should use a method which satisfies LNHarm and is immune to burying
strategy. As far as I can tell, this limits you to plurality, IRV, and probably DSC.
(Doing a couple of test scenarios suggests to me that DSC is sufficiently insensitive
to lower rankings to be rather immune to burying strategy.)

>  I mean, for all we currently seem to know, it might even be that the
> best voter strategy in these sysems is actually just to plurality-vote,
> in which case these systems might act just like the plurality system with
> strategic voters - just like now! - which is not so wonderful.  I am not saying this
> is true, but I am saying it isn't currently clearly extremely not true.

I haven't done expectation math with MDDA, but I have done some with ICA, and it
seems that in common simple scenarios, ICA strategy agrees with Schulze(wv) rather
than Approval. (I don't think it is realistic to fear that these methods might be
strategically equivalent to FPP.)

Kevin Venzke



	

	
		
___________________________________________________________________________ 
Appel audio GRATUIT partout dans le monde avec le nouveau Yahoo! Messenger 
Téléchargez cette version sur http://fr.messenger.yahoo.com