[EM] OpenSTV 2.1.0 released and new OpaVote features

[Juho Laatu]

On 17.9.2012, at 21.08, Richard Fobes wrote:

> On 9/15/2012 3:02 AM, Kristofer Munsterhjelm wrote:
>> On 09/15/2012 09:55 AM, Juho Laatu wrote:
>>> On 15.9.2012, at 6.05, Jeffrey O'Neill wrote:
>>>> You can also now save Condorcet results in HTML format but still
>>>> working on the best graphics to visualize Condorcet results.
>>> One solution is to support minmax(margins). With that method you can
>>> simply draw a histogram that indicates how many new (first preference)
>>> votes each candidate would need to win (or tie) with the winner /
>>> current leader.
> 
> Here is the URL to a results page at VoteFair.org that shows another way to graphically display pairwise-comparison results:
> 
> http://www.votefair.org/cgi-bin/votefairrank.cgi/votingid=41541-56251-09157

Some observations.

The results contain two histogram style visualizations (= numeric value for each candidate), "VoteFair ranking score" and "Traditional vote count" / "plurality". I discussed earlier about histograms that could show "distance to victory" or "quality of the candidate" (from the used method point of view). The presented histograms do not fall in these categories but offer additional information, I guess mainly to allow people to compare the reasults to some other ways to measure the candidates.

When reading the results I noted that the top two candidates were tied but all the other results were transitive. That means that the top two candidates had 10 pairwise victories each (+ one tie). The number of victories of the ramaining candiates were 9, 8, 7, 6, 5, 4, 3, 2, 1 and 0. These values would give us one additional histogram that would show the number of pairwise victories. Those values would tell quickly, at least to an experienced observer, that other than the two top canididates are ordered transitively. That piece of information would be valid information for most Condorcet methods, but only partial, since it only indicates the order of the candidates unambigiously if there are no pairwise cycles or ties. (Well, there is no requirement of single-winner methods or Condorcet methods to give a ranking, but the intended order is often obvious anyway.)

That's all. Just some observations on the presented extar histogram style information. It is not very easy to generate "distance to victory" or "quality of the candidate" measures for all Condorcet methods.

Juho

> 
> Note that the pairwise comparisons are sorted according to popularity.
> 
> The length of each bar indicates how many voters support that choice compared to the other choice in the pair.
> 
> The "summary" section uses bar lengths that basically sum up the pairwise counts.  This means that the length of a bar for a higher-ranked choice is not necessarily always longer than the length of a bar for the next-lower-ranked choice (but such cases become rare as the number of votes increases).
> 
> Richard Fobes
> 
> ----
> Election-Methods mailing list - see http://electorama.com/em for list info