[EM] Yee/B.Olson Diagrams (YBD's): the next step

[Brian Olson]

I did some small multiples runs. They kinda look good
http://bolson.org/voting/sim_one_seat/20081202/

Brian Olson
http://bolson.org/
(Sent from my iPhone)

On Dec 2, 2008, at 9:28 AM, "Raph Frank" <raphfrk at gmail.com> wrote:

> Ultimately, there are more degrees of freedom than a 2d picture can
> show.  If you hold one of the candidates fixed and let the other two
> be placed anywhere, then you have four degrees of freedom.  However,
> one of those degrees of freedom can be dropped due to rotational
> symmetry.
>
> This gives the same 3 degrees of freedom as Forest's 2d + scale  
> method.
>
> On my implementation at
> http://ivnryan.com/ping_yee/results.html
>
> I have some user control.  The user can place the centre of the
> population at any point and then the winners are highlighted.  This
> was mainly for multi-seat competitions.
>
> In the single seat versions, it will highlight the single winner and
> shows 2 circles (enclosing 95% of the population and 50% of the
> population)
>
> Something similar could be done for Forest's idea.  There could be 2
> pictures side by side.  The one on the left would show all the
> configurations and the one on the right would show the 'current'
> standard diagram.  When you move the mouse to a position on the left
> diagram, it would update the right hand diagram so that it shows the
> candidate positions for that configuration and the strandard Yee
> diagram.
>
> There would also need to be some way to adjust the scale.
>
> In principle, one way to show the 3d effect would be to have 3d 'fog'
> where there colour matches the pathology.  If the user could rotate
> the fog, it would give an impression of the confguration.  The other
> option is to just have say 10 scale settings that can be used.
>
> One other potential problem is that it would require considerably more
> computer time.  Each pixel would require as much time as an entire
> standard diagram.  Also, there would need to be some way to
> automatically determine if a diagram has a given pathology.
> ----
> Election-Methods mailing list - see http://electorama.com/em for  
> list info