I actually already have a mode in my program to run multiple elections at each point and average the color of the winners over those rounds.

It doesn't hurt anything to have more than three candidates as long as each one gets a color reasonably distinctive from the others. As long as each choice has some region where they are dominant, there will be a pretty visible blending edge with neighboring choices.

If you want to add new election methods to test, my source is available via subversion respository at
http://voteutil.googlecode.com/svn/sim_one_seat

On Nov 12, 2008, at 5:56 PM, [EMAIL PROTECTED] wrote:

It recently occurred to me that in the case of three-candidate elections, Yee-Bolson Diagrams can be
generalized to lotteries:

In the three candidate case the traditional Yee-Bolson Diagram of a method is a coloring of voter/candidate space in which each point of the space is assigned the color (red, green, or blue, say) assigned to the candidate that would be elected if the mean of the (standard normal) distribution of voters were at the point
in question.

To adapt this to lotteries we generalize the color possibilities to all possible hues in the RGB system of colors. In this system RGB(p,q,r) is the (full intensity color) that is made up of red, green, and blue in the proportion p:q:r . The respective pure red, green, and blue colors that go with deterministic methods are
given by RGB(1,0,0), RGB(0,1,0), and RGB(0,0,1).

Ideal deterministic methods yield Yee-Bolson diagrams wherein the colored regions are Dirichlet/Voronoi
regions relative to the candidate positions.

These ideal diagrams are benchmarks for comparison with diagrams based on other methods.

It seems to me that the (generalized) Yee-Bolson diagram of the Random Ballot method could serve as
another benchmark.

This idea could be the basis of a good master's degree project.

By the way, for me the most convincing case against IRV is the original paper by Ka-Ping Yee.

See

http://zesty.ca/voting/sim/

Forest
----
Election-Methods mailing list - see http://electorama.com/em for list info

Reply via email to