[EM] Squeezing the most from the ballot [was Party's list or voters ranking, Let the voter choose]

[Forest Simmons]

By the way, there are several reasons for not requiring the "predefined"
ballots that you mention as current practice:

(1) The strategies reflected in those predefined ballots are based on
polls that cannot be as accurate as the actual proxy counts.

(2) If the proxy actions are predefined then there is no room for the
horse trading that can ameliorate the "tyranny of the majority."

(3) A true majority winner will automatically know that (s)he has a
majority of loyal supporters before acting as proxy, and so will know to
leave the truncated ballots truncated.  In other words these hybridized
methods are almost sure to satisfy the majority criterion, even if the
corresponding non-hybridized methods don't.


Having said that, there are still possibilities for some useful
constraints on the proxies' actions:

1. Their approval ballots should not contradict a predefined CR ballot.
In other words, if you approve someone that you gave a CR of x, then you
have to approve all candidates that you rated above x.

2. You have to rate yourself at the top of the scale.

etc.


Forest



On Thu, 10 Jul 2003, Forest Simmons wrote:

>
>
> On Wed, 9 Jul 2003, [iso-8859-1] Anthony Duff wrote:
>
> >
> > How about offering the voter a choice?  Let the voter
> > choose to either (A) mark 1 box to vote for a party’s
> > predefined ranked ballot, or (B) complete the ballot
> > with their own ranking.
> >
> > This is a method that is in practice and works quite
> > well.  It is particularly useful when there are a
> > large number of candidates.
> > Most voters will take option (A).  Few voters take
> > option (B).
> >
> > Option (B) is more complicated to tally, count and
> > track transfers, and so is helpful that few take it.
> > However, it is important, in principle, that voters
> > have the (B) option so that they are free to vote any
> > way they choose.
> >
> >
>
> What you describe is essentially a hybrid Candidate Proxy method.
>
> Any method could be hybridized more simply as follows:
>
> Rule 1.  Any ballot with a single mark ("bullet vote") is automatically
> filled in according to the recommendation of the candidate whose name was
> marked.
>
> Rule 2.  A virtual candidate named "No Proxy" is listed among the real
> candidates, so that someone who wants their bullet vote to remain a bullet
> vote can make two marks on the ballot, thereby circumventing rule 1 above.
>
> In other words, by default a bullet vote is considered a proxy
> designation, but this default can be over-ridden by marking "No Proxy."
>
>
> Since, as you say, most voters are inclined towards the proxy option, the
> default should be in this direction.
>
> The Approval/Proxy hybrid is especially attractive since it only requires
> a one bit (per candidate) ballot, the same ballot used in FPTP Plurality.
>
> Suppose that we had this hybrid going in the last U.S. Presidential
> election.
>
> Since Nader called Gore and Bush, respectively, "Tweedle Dumb" and
> "Tweedle Dumber," I assume that he would have approved Gore in the proxy
> vote if he himself lacked the support to win (and assuming Bush didn't
> offer some believable major green concessions for Nader support).
>
> Nader supporters who couldn't stomach approval for Gore even if hell
> should freeze over, could approve Nader and any other candidate besides
> Gore, if only "No Proxy" for example, and that would guarantee that their
> ballots would not count for Gore.
>
> The main difference between what I am proposing and what you say is
> already in practice is that the proxies get to decide where to put their
> approval cutoffs after the rest of the ballots are counted.  So they know
> how much loyal voter support they have before they have to decide how much
> if any approval they should extend to other candidates while acting as
> proxies.
>
> At present I cannot imagine any other method that could squeeze more out
> of a one bit ballot.
>
> And I seriously doubt that any single winner method currently in use for
> public elections could be expected to give superior results, no matter how
> complicated the ballot.
>
> Forest
>
> ----
> Election-methods mailing list - see http://electorama.com/em for list info
>