[EM] reply to Gilmour attack on range voting & social utility; CCd to RangeVoting

James Gilmour jgilmour at globalnet.co.uk
Thu Dec 8 04:08:42 PST 2005


> Abd ul-Rahman Lomax Sent: Thursday, December 08, 2005 2:40 AM
> > At 03:07 PM 12/2/2005, James Gilmour wrote:
> >On this point we shall have to disagree.  Just because you express 
> >your liking for A and your dislike for B more strongly than I do, 
> >does not mean your vote should count for any more than mine, or than 
> >anyone else's, when we are asked to choose between candidates A and 
> >B to be our representative.  You may be more upset than I am when B 
> >wins, but that does not give you any right for your vote to have a 
> >greater effect than mine in determining the outcome.
> 
> Let me put it this way. If we have a true democracy, the form of 
> government and its institutions are a matter of the consent of the 
> people. Suppose a majority of people decide that, say, Range Voting 
> is the best method, and it is a method which *allows* but does not 
> require people to cast a weak vote (but stronger than what everyone 
> agrees is allowed, an abstention).
> 
> In that case we could not say that allowing stronger expression to 
> count more strongly was undemocratic, or could we?

OK, if that is the will of the people, fine.  But I see and know of no evidence that is the will of the people - rather,
the reverse: all should have an equal say.  Certainly I have never seen any evidence that anyone would accept that my
contribution to an election outcome should depend on how passionately I feel about the candidates.  So I try to live in
the real world, where there are real tasks of practical voting reform to be tackled, not in some hypothetical "true
democracy".


> > [Re Mr. Smith:]
> >Please do not interpret this as my arguing for IRV over any other 
> >single-winner voting system, but it is obvious from
> >your comments that you do not understand the basic philosophy behind 
> >such preferential voting systems.  The preferences
> >after the first are contingency choices, to be brought into 
> >operation only if needed.
> 
> The same is true of Range Votes, actually. It's easiest to see with 
> Range2, i.e., Approval. If you vote for two candidates in approval, 
> at most one of these votes actually counts. The other is moot; that 
> is, the election outcome would not change if the vote were removed. 
> Essentially, with Approval, both votes are "contingency votes."

This is not correct.  In Approval both votes are simultaneous and the voter has no control of the "order" in which they
might be considered because that is irrelevant in this counting system.  In IRV the preferences are contingent choices,
under the control of the voter.

> The big flaw in IRV -- that does not exist with Condorcet methods -- 
> is, of course, that IRV, as generally proposed, fails to use all the 
> pairwise information from the ballots simultaneously.

In repeating this statement it is clear that you have failed completely to understand what I wrote.  This is a "flaw" if
you view the information on the ballots from an "inclusive" perspective (= social choice, maximising consensus).  But
IRV preferential voting  had a completely different origin, based in what may be described as the "exclusive" view of
representation (maximising diversity).  In IRV, successive preferences are contingency choices, to be brought into play
ONLY if they are needed, and they are directly under the control of the individual voter.

It would, therefore, be quite wrong to look simultaneously at all the information on an IRV ballot paper and say that
the IRV counting system had failed to use all of that information.  It doesn't do that because it was never intended to
do that, because those who invented this type of preferential voting never wanted it to do that.  However, if your
fundamental objectives are different, of course you will want to devise a different system for recording voters'
responses to the candidates and of 'counting' the votes.

James Gilmour







More information about the Election-Methods mailing list