[Election-Methods] RE : Re: RE : Re: RE : Re: rcv ala tournament

[rob brown]

On Dec 30, 2007 6:51 AM, Kevin Venzke <stepjak at yahoo.fr> wrote:

> Rob,
>
> --- rob brown <rob at karmatics.com> a écrit:
> > Anyway, as I'm sure you know, that system (rank/rate your candidates
> > honestly, let the system generate the most strategic approval ballot) is
> > simply DSV, and it works out to be the same as Condorcet.
>
> How does your method resolve:


Not sure what you are referring to as "my method".  I use DSV as an
explanatory device (i.e. "a software agent which takes your preferences as
input and produces the most strategic approval ballot"), but I advocate any
condorcet method, and don't see a lot of point debating the differences
between the various condorcet resolution methods.

I have messed around with producing my own method, but its main benefit is
that it produces nice stable *scores* for each candidate that can appear in
a bar chart. ( http://karmatics.com/voting/bars-demo.html )

With regard to:

49 A
> 24 B>C
> 27 C>B


C wins, condorcet winner.

My scoring system gives C a score of 35.13, vs 25.75 for B and 15.12 for A.
I think those are pretty reasonable scores.

Matrix/scores:

c 35.13
b 24.75
a 15.12

vs.
> 49 A
> 24 B
> 27 C>B


a wins (by my method), but it is a condorcet cycle.   b comes in second.

a  37.02
b  33.37
c  29.61


> May I assume a voter is allowed to bullet-vote?


Sure.  If they prefer one candidate to all the others, and consider the
others to be equal, a bullet vote is a sincere ballot.  (they should also be
able to do things like B=C>A, in my opinion)


> If not, how does your method resolve:
> 40 A>B>C
> 35 B>C>A
> 25 C>A>B
>

Condorcet cycle again.  My scoring system gives b a razor thin margin.
(score 35.8 to 35.64)

b 35.8
a 35.64
c 28.56

Does your information allow rated information? It seems this should be no
> problem since you would not have any incentive to exaggerate your
> preferences when the method does its best to get you what you want.


My scoring system would just reduce that ordinal rankings, partly for the
convenience of being able to express that as a pairwise matrix.  But I have
no problem with a system that the users enter ratings, as long as they
aren't tabulated a la Range Voting.

In the past, I tossed out a proposal (intended more as a thought experiment
than a practical suggestion) for a DSV method that voters entered cardinal
ratings:
http://listas.apesol.org/pipermail/election-methods-electorama.com/2005-December/017781.html

I think if that method was actually used, there would be almost zero benefit
to attempting to vote strategically, and there would be literally zero
benefit in the majority of elections.  But even in those where that was any
benefit at all, to get that benefit would be exceptionally hard because of
the need to know others preferences with extreme precision.

Actually, I think in real world elections, that is true for any Condorcet
method.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.electorama.com/pipermail/election-methods-electorama.com/attachments/20071230/65cd3890/attachment-0003.htm>