[EM] old school MCA issue

[Forest Simmons]

On Wed, 22 Oct 2003, [iso-8859-1] Kevin Venzke wrote:

> Here are some messages that might be of interest:
>
> Forest proposes the method (MCA):
> http://groups.yahoo.com/group/election-methods-list/message/9571
> Alex proposes that any number of favorites be permissible:
> http://groups.yahoo.com/group/election-methods-list/message/9582
> Forest agrees but expresses doubts:
> http://groups.yahoo.com/group/election-methods-list/message/9628
> Joe Weinstein coins "Majority-Choice Approval," supports the unlimited version,
> offers some other commentary:
> http://groups.yahoo.com/group/election-methods-list/message/9692
> Forest replies to this:
> http://groups.yahoo.com/group/election-methods-list/message/9707

Thanks for taking the time and effort to summarize the evolution of MCA.


>
> It seems that ever since, we've favored permitting any number of favorites
> to be listed, because that permits MCA to meet weak FBC and clone criteria,
> like Approval.
>
> Could I interest anyone in reconsidering this point?  It has come to be my
> feeling (and it seems to have been an original suspicion) that in realistic
> scenarios, there is little or no use for the middle rank, if any number of
> candidates may be "preferred."  (I'll spare you the reasoning at this point.)
>
> This would be quite different if only one candidate could be preferred.  We
> would have a method where only one candidate could be a majority favorite,
> which would be easier to sell, I think.  Forest suggested (back then; maybe
> he's changed his mind) that a majority favorite would be rare, anyway.  (I suspect
> he is or was right, though: Even in Plurality, a weak third candidate can
> prevent anyone from having a majority.)
>
> Who would gain from ranking Compromise as "preferred"?  One who thinks Compromise
> could be a majority favorite, and that Compromise is better than the expected
> winner by greatest approval.  It's true that such voters could give the election
> away to Compromise, but: 1. They'd be likely to do the same thing in an Approval
> election (that is, approve Compromise, believing Favorite is hopeless), and 2.
> Favorite's true approval, at least, would be revealed in the results.
>
> This does not look bad to me.  What do you think?  Does it seem to anyone that
> this method would preserve two-party rule?  Is FBC too crucial to fail so clearly?
>

I still have mixed feelings on this.  Another idea that would encourage
use of the middle level is to raise the top slot quota, i.e. the quota for
a win on the basis of favorite votes alone.

To see this suppose that a candidate had to be marked favorite by 99
percent of the voters in order to win on that basis.  The chance of
winning on that basis would be so small that the top slot would be
considered only "expressive" as opposed to "instrumental" so there would
be virtually zero incentive to move a compromise up to that level.

I'm not suggesting raising the quota to 99 percent, but somewhere between
50% and 99% (perhaps 55%) there has to be a happy median that would tend
to spread out the approval votes more or less evenly among the two upper
levels.

There probably isn't a "one size fits all" happy median quota, though,
since it could depend on the number of candidates, their distribution in
issue space, etc.

The other problem is that MCA loses its simple majoritarian appeal when
we change the quota to something other than the magical 50%.


Forest