move to "range voting"
Hi there. I hope that I won't offend anyone by this move. I just don't think that it makes any sense to call this method "cardinal ratings". My main reason is that "cardinal ratings" describes a ballot type rather than a tally method. A secondary reason is that the term "cardinal ratings" is somewhat redundant in itself.
If people really don't like the name "range voting", I'd suggest "ratings summation", or "ratings sum".
I suppose that "average rating" works as well, but it raises a question: what is the impact of a ballot that doesn't rank a particular candidate? If the ballot counts as the lowest possible score (e.g. 0 on a 0-100 scale), then "ratings summation" is the most appropriate name. If the ballot doesn't affect the candidate's score at all, then "average rating" is the most appropriate name. For political election purposes, I think that a candidate not ranked on a given ballot should be assigned the lowest score, so that candidates who are unknown to most voters are unlikely to win. Hence, I favor the term "ratings summation" rather than "average rating".
Am I making any sense?
My best, James Green-Armytage 04:39, 19 May 2005 (PDT)
I don't object. This is under "range voting" on wikipedia as well. I suppose "CR" was popularized by Mike Ossipoff. I'm curious to know who coined "range voting," though. I first remember seeing it in Warren Smith's paper (which was very opinionated and contained a number of errors). Kevin Venzke 11:38, 20 May 2005 (PDT)
Good question; I don't know. Maybe a good topic to bring up on the list. I was under the impression that it had a fairly well-established usage, but I could be wrong. Annoyingly, most of the google hits I get are clones of the wikipedia article (with this one at the top), and none of them seem to cite a seminal paper. Even if "range voting" doesn't have a good usage history, it's still probably better than the redundant "cardinal ratings", but perhaps "ratings summation" would be more descriptive? James Green-Armytage 10:12, 21 May 2005 (PDT)
Comment on the Condorcet Criterion violation example. If 51 voters vote A100>B90 vs. 49 voters with B100>A0, then isn't it possible that B-preference voters have an incentive under Range Voting to bullet vote for B instead of rating A truthfully? In that case, Condorcet correctly avoids giving that incentive. --Araucaria 17:40, 5 August 2008 (UTC)
Alternate title suggestion: Score Voting
The rangevoting.org site has recently been proposing Score Voting as an alternative term. --Araucaria 23:21, 13 August 2009 (UTC)
Example is flawed
IMO, the example of converting ranking to scores is flawed, and I assume this is inherited from the Wikipedia source. The example with scores from 1 to 4 is essentially the same as Borda, and doesn't represent what voters might do in an actual Score Voting election.
One might reasonably expect that Memphis voters would vote 100% Memphis, 90% Nashville, 30% Chattanooga, 0% Knoxville, which would have a much different score result, though probably Nashville would still win. --Araucaria 23:21, 13 August 2009 (UTC)
Range Voting could be interpreted as a method to approximate the centroid of a data set by finding the choice with the smallest sum of squared distances to each voter. However, since an individual voter has no way of measuring absolute distances, the best they can do is to compare their relative distance from every candidate, giving their closest candidate the highest score and the most distant candidate the lowest score. The question then is, what function of relative distance best approximates that least sum of squares property? --Araucaria 23:21, 13 August 2009 (UTC)