On Thu, 2005-08-11 at 09:29 -0400, Warren Smith wrote: > [Rob Lanphier wrote]: > > Here's a counterexample: > > > > 41 ballots: > > A:10 > > B:3 > > C:0 > > (Ranked equiv: A>B>C) > > > > 10 ballots: > > A:5 > > B:10 > > C:0 > > (Ranked equiv: B>A>C) > > > > 10 ballots: > > A:0 > > B:10 > > C:5 > > (Ranked equiv: B>C>A) > > > > 39 ballots: > > A:0 > > B:3 > > C:10 > > (Ranked equiv: C>B>A) > > > > Range Voting result: > > A:460 > > B:440 > > C:440 > > > > Condorcet winner: B > > B beats A: 59-41 > > B beats C: 61-39 > > A beats C: 61-39 > > > > Running the "erase the candidate" filter over this election doesn't > > change the fact that A beats B in a Range election, even though B is the > > Condorcet winner by a clear margin. ... > So your "B is the condorcet winner" does not imply "B is the beats all winner" > according to WEIGHTED votes. > > My point was Condorcet only considered <,>,= relations WITHOUT numerical > magnitudes, > just yes of no, as votes, hence the distinction between my 1st + 2nd > defintions of the Condorcet concept, dd not occur in his mind.
That would be /the/ definition of the Condorcet winner. I sincerely hope that you remove the extremely misleading proposition that Range Voting is somehow Condorcet-winner compliant. Your tactic a very similar tactic to one used by many Condorcet advocates which I also object to. Condorcet fails the "Independence from Irrelevant Alternatives" criterion (IIAC), made famous by Kenneth Arrow in his Nobel prize winning theorem. Many Condorcet advocates have tried to dance around this issue by redefining IIAC to be "Local IIAC", and pointing out that some Condorcet methods pass "Local IIAC", /before/ confessing that they fail IIAC as defined by Arrow. I personally believe that Local IIAC is a valid criteria, but I also believe that Condorcet advocates need to be honest and direct when the question comes up. Simply stating, "Condorcet methods fail IIAC. However, ..." can be just as convincing if the argument is a good one. For the record, I believe that Local IIAC is a very valid substitute for IIAC, for reasons we don't need to rehash just now. Similarly, I would ask that you are just as up-front about your new criteria. Name it something new, e.g. "Weighted Condorcet Winner". However, please do not spread disinformation about the classic definition of "Condorcet Winner". Regarding the example: 59 people out of 100 prefer B to A. Under the principle of "one person, one vote", it's extremely difficult to argue that A should win. Rob ---- Election-methods mailing list - see http://electorama.com/em for list info
