[EM] "Higher Resolution Methods"

[Forest Simmons]

I agree 100% with what you said.  However, you did misinterpret one of my
(unintentionally ambiguous) statements, the one that says  "But optimal
strategy voting based on Cardinal Ratings results almost surely in an AV
ballot."

Here's what I intended that sentence to say:  A marked Cardinal Ratings
ballot will probably look just like an AV ballot if the voter uses an
optimal strategy. In other words, given the chance to use the most
expressive type of ballot, the cunning voter will "waste" the opportunity
and vote as if it were an AV ballot.

By the way, the problem you pointed out is one reason I said "strongly
suggests" instead of "implies."  

Forest

On Wed, 17 Jan 2001, Martin Harper wrote:

> The optimal strategy for cardinal ratings will depend on how you count 
> the ballots... if you use the cardinal ratings to determine ordinal 
> rankings for beatpath or IRV, the strategy will be different to if you 
> simply add them all up. Strategy is a result of the method used, not the 
> ballots used...
> 
> 
> Forest Simmons wrote:
> 
> > I believe that "higher resolution methods" have a place, but mostly in
> > non-political contexts like pattern recognition, where the grading or
> > judging is done by mechanical sensors or where the graders have no stake
> > in the outcome.  Ice skating competition judges seem to have set grading
> > standards, since they routinely give scores within 10% of each other.
> > 
> > Let me summarize again why I favor AV.  Cardinal Ratings (high resolution) 
> > can be considered a refinement of both AV information and Ordinal Ranking
> > information, so it is more "expressive" than either.  But optimal strategy
> > voting based on Cardinal Ratings results almost surely in an AV ballot.
> > This strongly suggests to me that although the AV ballot may not be as
> > expressive as the others, it carries the most relevant information for the
> > actual election.
> > 
> > In a mechanical or otherwise non-strategical situation, the Cardinal
> > Rating would be preferable if available.
> > 
> > Forest
> 
>