[EM] Why I think IRV isn't a serious alternative 2

[Abd ul-Rahman Lomax]

At 04:47 AM 12/22/2008, James Gilmour wrote:
>In a post last night I wrote:
> > Sent: Sunday, December 21, 2008 11:14 PM
> > I am not going to comment of the rest of your interesting
> > post in detail, but I am surprised that anyone should take
> > Bucklin seriously.  I, and some of our intuitive electors,
> > would regard it as fundamentally flawed because a candidate
> > with an absolute majority of first preferences can be
> > defeated by another candidate.  Such a result may measure
> > some "compromise view" computed from the voters' preferences,
> > but it is not considered acceptable  -  at least, not here
> > for public elections.
>
>Chris Benham has kindly (and gently) pointed out my error, 
>off-list.  My comments above relate to BORDA, not Bucklin.
>My apologies to Abd and all for confusing the two systems and for 
>any confusion my comments may have caused.

No problem. It was simply confusing.

>Bucklin would, of course, correctly identify the majority winner in 
>the case described above.  But some of us take the view that
>Bucklin falls "one person, one vote" unless all voters are 
>(undesirably) compelled to mark preferences for all candidates -  but
>that is a completely different issue, and I am aware there is more 
>than one view on the meaning of "one person, one vote".

The general legal opinion seems to be that it doesn't fail that 
principle. It *looks* like the person has more than one vote, but, 
when the smoke clears, you will see that only one of these votes was 
actually effective. The voter has contributed no more than one vote 
to the total that allowed the candidate to win. Consider the election 
as a series of pairwise elections.

If the voter votes a single vote, the voter is casting a vote for the 
favorite in all pairwise elections involving that favorite. The voter 
is, however, abstaining from all other pairwise elections. If the 
voter approves another (whether unconditionally as in Open Voting or 
Approval, or conditionally as in Bucklin), the voter has voted in 
other pairwise elections, but has abstained from one, the pairwise 
election involving the two. There *is* additional voting power, but 
not violating one-person, one-vote.

In a ranked method, generally, such as STV, the voter may possibly 
vote in all pairwise elections, except that with STV some of these 
votes aren't counted. With a Condorcet method, the votes all count. 
Think of it as IRV with a different method of deciding whom to 
eliminate. Elect the winner of a rank unless a majority isn't found, 
if not found, proceed to the next rank. When the ranks are all 
exhausted and counted and added together, eliminate the candidates 
with less than a majority. Or eliminate the lowest-vote getters, thus 
finding the candidate or candidates with the most ballots showing support.

This "most ballots showing support" approach was cited with approval 
in Brown v. Smallwood, but then the court proceeded, in order to find 
Bucklin violating one person one vote, to ignore it, noting that 
there were more marks than voters.... thus confusing, in direct 
contradiction with what they had just quoted, marks with the ballots. 
The argument that they gave applies to any alternative vote system 
where the voter votes for more than one candidate, and that in STV, 
only one vote at a time is counted, is simply a procedural 
difference. The reasoning in Brown v. Smallwood was not repeated 
elsewhere, and Bucklin systems weren't elsewhere removed for 
constitutional reasons related to the canvassing method.