[EM] I stand corrected - thank God for little favors

[Blake Cretney]

Donald E Davison wrote:

>      I stand corrected. And, I wish to thank you for pointing out this

> needed correction. You have been most helpful.

>      I am more than willing to embrace this corrected defination of

> Approval Voting(AV).

>      One question: Does the definaion also apply to Borda Count??



The requirement that a candidate with a majority of first-place votes
should win is the Majority Criterion.  It is listed in the EMR
(Election Method Resource).  I don't list Approval as passing or
failing this criterion since it doesn't even have first-place choices.
 Borda fails this criterion.

> 

>      Anyway, my current rankings of the single winner methods are as
follows:

> 

>      On a scale of one to ten - one being the best.

> 

>         1. Choice Run-Off(aka Alternative Vote)

>         2. Run-Off without Elimination(RWE)

>         3. Salva Count

>         4. Salva Voting

>         5. Condorcet

>         6.

>         7.

>         8.

>         9. Borda Count

>        10. Approval Voting(AV)

> 

>      I put Borda Count down lower, in the event it also has the same

> defination of first count choices as AV.

It isn't a matter of how first-choices are defined.  It is how they
are used.


> 

>     You say that your Resource site is objective. If so, then you should

> include the methods Salva Voting, Salva Count, and Run-Off without

> Elimination. These three methods are valid methods and any one of them is

> far superior to Approval Voting, which you do list.



Inclusion is not based on merit of the method, because that would be
too subjective.  I personally think that Borda is disastrous, but I
still list it.  Here are the major considerations I am currently
using:



Is the method obvious?

I try to add methods that people are likely to think of again.  For
example, Paul Dumais recently re-suggested Nanson's method without
ever having heard of it.  It's obviously helpful to provide
information on such methods.  Although I don't consider the methods
you mention to be particularly complicated, I don't think that they
would be frequently re-invented.



Has the method been properly studied?

Looking at my web site you will see a page with a long list of
criteria.  For each criterion, I have listed the methods that pass and
fail.  Adding new methods to this list takes a significant amount of
my time.  I am therefore obviously encouraged to add a method if
someone else has already done some of the work for me.

I am particularly concerned that some of these methods may have
properties that you are unaware of.  For example:

Does the method suffer from vote-splitting?

Does the method suffer from the opposite problem, that running more
candidates helps a party (what I call teaming)?
If these methods did suffer from one of these, would it matter to
you?  I don't want to add a method just to have it withdrawn by the
author after it is better understood.


Is the method popular (or has it been popular)?
Borda and Approval have both been well-known and fairly popular for a
long time.  The methods you propose are not even YOUR first choice.

I see no limit to the number of methods that you could invent and
rank between IRV and Condorcet.  These three methods are only a few
months old.  I think you can understand why I worry that adding these
methods would create a precedent that could result in more growth than
I can handle.

>      The people who like Approval or Borda or Condorcet should also like

> the higher three because those three also allow the compromise candidate a

> chance to win.


I think you are oversimplifying the positions of these three
different groups.  I'm not sure exactly what you mean by a "compromise
candidate".  If you mean a candidate with low plurality scores, then
clearly being a compromise candidate is not in itself enough to
justify election, or similarity to Approval, Borda, or Condorcet.

If these methods are being proposed as a compromise between IRV and
one or more of Approval, Borda, and Condorcet, then I think some more
work should be done on proving that they fill this role.  Try to show
under which circumstances they give the same results as IRV, when they
give the same results as one of the others, and when they give results
all their own.

---
Blake Cretney
See the EM Resource:  http://www.fortunecity.com/meltingpot/harrow/124