[EM] approval defection; another easy method

[Kevin Venzke]

Hi,

 --- Russ Paielli <6049awj02 at sneakemail.com> a écrit : 
> But now let's say that Approval has been in use for several election 
> cycles, and the Green is starting to catch up to the Democrat. Well, the 
> Green Party will certainly urge their supporters to drop their vote for 
> that pesky Democrat. But if they take the advice, they risk handing the 
> election over to the Republicans, of course.

Here is a method that is still as easy to count as Approval, and which
seems to address this problem. I got this idea from Chris Benham, but
as far as I know I'm the only person who likes it. I call it "the vote for
and against" or "VFA."

1. The voter votes FOR and AGAINST one candidate.
2. The candidate with the most FOR votes wins, except that a candidate
receiving more than half of the AGAINST votes is not allowed to win.

So you might get votes like this:

40 Repub >> Green
25 Democ >> Repub
35 Green >> Repub

And the Green would win. This might correspond to ranked ballots:

40 Repub
25 Democ
35 Green>Democ

In which case we could complain that Green mustn't win due to the
Plurality criterion. But I note that electing Repub would seem to
be a majority rule failure (I'll be vague about that for now),
and any method that elects Democ here is still going to have a
defection problem, since on these ballots:

40 Repub
25 Democ>Green
35 Green>Democ

There is no way to elect Democ. (That is, Democ can't win, but you
permit those voters to make Democ win by deserting the Green.)

The VFA method fails Clone-Loser (since you could divide up the AGAINST
votes), but I don't see a good way to use it. If there are two large
factions, one faction can benefit from running two candidates if 
1. they believe they won't have a majority, and
2. the other faction also runs two candidates.
But in this case, the other faction has no incentive to run clones.

The VFA method clearly fails FBC, but it uses this in a strange way
that is perhaps even positive. In FPP and IRV, central voters have
incentive to desert their favorite candidate because the criterion for
elimination is having few first preferences. But in this method,
flank voters have incentive to desert their favorite in favor of a
*central* candidate, since the criterion for elimination is having
too many *last* preferences.

In other words, if the Repub voters really strongly object to Green
winning, they will have to move their vote to Democ.

In a different way, VFA approaches FBC nicely. Suppose there are
two strong flank candidates and one central candidate, and you
assume that one of the two flank candidates will receive a majority
of the AGAINST votes. (This should be likely, since a vote against
the central candidate would seem to be a waste.) In this case you
are free to vote for the central candidate if you like him best.
No matter which of the flank candidates is your compromise choice,
you can't spoil the result.

Any thoughts on this?

Kevin Venzke



	

	
		
Découvrez le nouveau Yahoo! Mail : 250 Mo d'espace de stockage pour vos mails ! 
Créez votre Yahoo! Mail sur http://fr.mail.yahoo.com/