[EM] Visualizations of Plurality, Hare, Approval, and Condorcet

[Brian Olson]

On Apr 20, 2006, at 8:06 AM, Ka-Ping Yee wrote:

> I'm new here and hope to learn from others on this list.  Thank you
> for indulging me with this request for feedback.
>
> I did some simulations of Plurality, Approval, Hare, and Condorcet
> methods (under certain simplifying assumptions) and produced
> visualizations of the results, which are posted at:
>
>     http://zesty.ca/voting/
>
> I'm curious what you all think of this.  I was already skeptical of
> the Hare method but the results surprised even me -- are they
> surprising to you?

Nifty!

Interesting methodology and good results. I really like how the  
middle-exclusion and non-linearity are portrayed.

And for a lot of people, a two-dimensional political debate is  
substantially adequate. I hear a lot of reference to a system of  
liberal-conservative being measured on the two axes of economics and  
social policy. So, given that kind of debate, I think a lot of people  
could identify with these graphs. You could even point to one and  
say, "This is kinda how the 2000 US Presidential election, here's  
Bush, here's Gore, and here's Nader." { (-.2, -.2), (.1,.1), (.5, .5) }

I did a plain utility simulation. It's here:

http://bolson.org/voting/sim.html

Instead of having an opinion space for voters and candidates, I just  
gave each voter a set of uniform distribution random opinions on each  
candidate. The goal was to determine the ability of election methods  
to maximize social utility. I think my most interesting results are  
that single vote gets worse results when there are many choices and  
IRV doesn't keep up wit the rest, and that those methods also degrade  
more rapidly under error.


Anyway, your graphs are really neat, and I'm kinda tempted to repeat  
such an experiment with a few tweaks of my own.