[EM] FBC wording

[Forest Simmons]

On Tue, 10 Feb 2004, James Green-Armytage wrote:

> Kevin Venzke <stepjak at yahoo.fr> writes:
> >No one's alternatives to FBC have resembled mine, and I think my "weak"
> >version is wrong, but I don't see what's wrong with my "strong" version.
> >Any comments from anyone?  Did I leave a loophole of some kind?
> >
> >Keeping in mind that this criterion is supposed to be unattainable:
>
> 	I think FBC isn't totally unattainable, but it's incompatible with
> certain other properties that are basically necessary for most election
> purposes, such as universal domain. I think that this method passes FBC,
> for example:
>
> 	The options are laid along a spectrum before the voting takes place, by
> some sort of authority figure. Voters then are unable to order their
> preferences in a way that is incompatible with the order of the spectrum.
> That is, if the order of the spectrum is A-B-C-D-E-F-G-H-I-J, then you can
> cast a vote A>B>C>D>E>F>G>H>I>J, or J>I>H>G>F>E>D>C>B>A, or
> C>B>A>D>E>F>G>H>I>J, or E>F>D>G>C>B>A>J, but not A>C>B>D>E>F>G>H>I>J, or
> A>J>I>H>D>B>C>G>C>E. I suppose that truncation would also be disallowed.
> The winner of course, would be the candidate(s) unbeaten in pairwise
> comparison.
> 	As far as I know, there is no possibility of a Condorcet paradox when
> options are necessarily laid down on a spectrum like this, and therefore I
> think that there is no favorite betrayal incentive.
> 	Correct me if I'm wrong.
> 	Probably I am stating the obvious here, but I just find this method to be
> somewhat interesting. Kevin's statement isn't really an error, but just a
> short way of saying something that we usually don't need to bother
> mentioning, that is, those "obviously necessary" criteria such as
> universal domain. I just wanted to talk about this method, because I think
> that it might have some application for public voting. For example, let's
> say that you were taking a public vote to determine what the prison
> sentence should be for possession of a pound of cocaine. If the possible
> options are 1 month, 2 month, 3 months, 4 months..., then it would be
> irrational for people to vote as 1 month > 4 months > 3 months > 2 months.
> Hence it would be unnecessary for the method to allow this. Probably the
> applications for this are fairly limited, but still it's nice to know that
> we can produce such a trouble-free voting system for certain purposes.

Very interesting example.

In general I wonder if one could have the voters submit two rankings of
the candidates: one being their perceived order of the candidates along
the (to them) most important issue axis, and the other a preference order
that makes sense in conjunction with this "spectrum."

The spectral orders could be combined some way to get a consensus
spectrum, and then all of the preference orders could be minimally
adjusted to "make sense."

The CW of these adjusted preference orders would be the method winner.

Forest

>
> ----
> Election-methods mailing list - see http://electorama.com/em for list info
>