Difference between revisions of ":2003 Debian Project Leader"
(query re IRV detailed results)
m (1 revision: Importing from the Electowidget test site from a few years ago)
Revision as of 10:55, 4 October 2009
Voters in the Debian Project Leader election were asked to rank the candidates, not rate them. However, Electowidget currently only accepts rating ballots.
So, I had to fake it. Here's the formula I used:
- All unranked candidates were also unrated.
- A scale of -5 to 5 was used
- NOTA (the horribly misnamed "None Of The Above"...should be "None Of The Below") always received a score of 0, if ranked
- All candidates above NOTA were given a score of (6 - ranking). So the first ranked candidate was given a 5, the second a 4, and so on.
- All candidates below NOTA were given a score of (NOTArank - ranking). A candidate one slot below NOTA was given -1, two slots below were given -2, etc.
- The program counts all rated candidates above non-rated candidates
Note that neither Schulze(wv-mod) and Instant Runoff use the absolute ratings, but rather, rankings are derived from the ratings. Therefore, the only thing that's really important about chosing the ratings is to choose ratings that transform consistently back into the rankings that were cast. The nice part about anchoring NOTA at zero is that an approval tally is also possible.
The IRV detailed results section of this article is obviously nonsense. First, it says the IRV winner is Robinson. Then it says that in the first round Robinson has 168 votes (how to get this from 160 unshared and 20 shared first-place votes is unclear). Then it says that in the second round Robinson has only 10 votes. Finally, in the last round it says Garbee is left after everyone else has been eliminated. - DPJ, 2006-07-27 10:19 UTC