Ralph Suter wrote (Wed.Dec.15):
authors) or meeting the "Favourite Betrayal Criterion" (i.e. never having a strategic incentive to Compromise).
To cheerfully assert, as W.D. Smith does, that "minimizing Bayesian regret" trumps majority rule is tantamount to saying
that more emotional voters should have more power than less emotional voters, which in my view is nonsensical and unfair.
One of the aims of a voting method should be to minimize the advantage well informed strategists have over sincere voters.
Under Range Voting, well informed strategists will have much more clout than naive sincere voters.
To me, it is axiomatic that a single-winner voting method should, with sincere voting, reduce to FPP when there are only two
candidates. RV doesn't.
No voting method is invulnerable to informed strategy, but meeting No Zero-Information Strategy is very easy to meet, so
why not at least achieve that? RV doesn't.
The paper Ralph refers to, by Smith, Quintal and Greene is paper 82 at
"So we recommend allowing blank entries, and averaging each candidate's non-blank entries to compute their final score."
They specify that making all the blanks 0s (the lowest possible score) is "inferior".
That means that if all the voters except one are sincere, and the sincere voters all ignore some candidate that they know
nothing about, and also give out no maximum scores; and the single other voter gives the unknown candidate a maximum score;
then the unknown candidate will be elected, with one vote!
I dislike methods that pose any sort of problem for voters who simply want to vote their full sincere ranking. Approval is
horrible in this respect: voters must agonize on where to place their approval cutoff. Range Voting with many more
available slots than candidates also can pose a problem: voters must agonize on exactly how many points to give each
candidate, in the knowledge that any point could be decisive (which is not he case in "Approval Margins", the method I
like that uses that type of ballot.)
Mike Ossipoff (Thu.Dec.30) speaks up for 3-slot CR as being more saleable and appealing to voters than Approval,
and more proposable than high-intensity CR or unrestricted ranking methods because of the limitations of current
voting equipment.
That is fine, but is 3-slot CR the best 3-slot method? I say definitely not! In my humble opinion the best 3-slot method
is 3-slot Schwartz // Disapproval.
(I also consider my "Majority Approval Runoff" 3-slot idea to be far superior to 3-slot CR.)
Chris Benham
CB: Range Voting is fine if all one cares about is "minimizing Bayesian regret" (Warren D. Smith, one of the paper'sWill someone on the list who has studied range voting and compared it to Condorcet, approval, and other methods please comment on Doug Greene's paper? He appears to be saying that range voting is superior to all other single winner methods. Are there good arguments against this conclusion? Does range voting have serious flaws? If so, could someone briefly summarize them?
authors) or meeting the "Favourite Betrayal Criterion" (i.e. never having a strategic incentive to Compromise).
To cheerfully assert, as W.D. Smith does, that "minimizing Bayesian regret" trumps majority rule is tantamount to saying
that more emotional voters should have more power than less emotional voters, which in my view is nonsensical and unfair.
One of the aims of a voting method should be to minimize the advantage well informed strategists have over sincere voters.
Under Range Voting, well informed strategists will have much more clout than naive sincere voters.
To me, it is axiomatic that a single-winner voting method should, with sincere voting, reduce to FPP when there are only two
candidates. RV doesn't.
No voting method is invulnerable to informed strategy, but meeting No Zero-Information Strategy is very easy to meet, so
why not at least achieve that? RV doesn't.
The paper Ralph refers to, by Smith, Quintal and Greene is paper 82 at
http://math.temple.edu/~wds/homepage/works.htmlOn page 11, they propose what I believe is a nutty idea on how to handle blank entries.
"So we recommend allowing blank entries, and averaging each candidate's non-blank entries to compute their final score."
They specify that making all the blanks 0s (the lowest possible score) is "inferior".
That means that if all the voters except one are sincere, and the sincere voters all ignore some candidate that they know
nothing about, and also give out no maximum scores; and the single other voter gives the unknown candidate a maximum score;
then the unknown candidate will be elected, with one vote!
I dislike methods that pose any sort of problem for voters who simply want to vote their full sincere ranking. Approval is
horrible in this respect: voters must agonize on where to place their approval cutoff. Range Voting with many more
available slots than candidates also can pose a problem: voters must agonize on exactly how many points to give each
candidate, in the knowledge that any point could be decisive (which is not he case in "Approval Margins", the method I
like that uses that type of ballot.)
Mike Ossipoff (Thu.Dec.30) speaks up for 3-slot CR as being more saleable and appealing to voters than Approval,
and more proposable than high-intensity CR or unrestricted ranking methods because of the limitations of current
voting equipment.
That is fine, but is 3-slot CR the best 3-slot method? I say definitely not! In my humble opinion the best 3-slot method
is 3-slot Schwartz // Disapproval.
(I also consider my "Majority Approval Runoff" 3-slot idea to be far superior to 3-slot CR.)
Chris Benham
---- Election-methods mailing list - see http://electorama.com/em for list info