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

[Abd ul-Rahman Lomax]

At 09:25 AM 12/25/2008, James Gilmour wrote:
>Abd ul-Rahman Lomax > Sent: Wednesday, December 24, 2008 5:39 PM
> > 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.
>
>An appeal to effective" votes is sophistry.  Bucklin is not a series 
>of pair-wise elections and more than one of your votes is being
>counted when there is no first preference winner but only one of mine.

The vote is counted, yes, but, in the end, if you did not vote for 
the winner, and your ballot, in a recount, were to vanish, you would 
find that it would not change the result at all. NONE of your votes 
mattered. And if you did vote for the winner, ONE of them counted. 
Thus all the others were alternate votes that don't change the result.

Apply elimination to Bucklin, the final vote, as if it were IRV. No 
more transfers, that's all. The same thing happens. All the useless 
alternative votes are eliminated and we are left with two candidates, 
and the one with the majority of non-eliminated candidates wins.

What is sophistry is the idea that IRV, in doing this, is satisfying 
one-person, one-vote, and Bucklin isn't. There is actually very 
substantial legal opinion in the U.S. that Bucklin does satisfy OPOV. 
Minnesota, Brown v. Smallwood, is the one cited by FairVote, but, in 
fact, BvS decided on the basis of *any* alternative vote being used, 
it is quite clear that it applies to IRV as well; but it was also 
idiocyncratic, confirmed nowhere, and the American Preferential 
System was used in as many as 52 cities in the U.S., nowhere else was 
it found unconstitutional.

Yes, more vote than one is counted, but that's true with IRV as well, 
the only difference is the sequencing. In the end, with 
single-winner, what matters is how many votes the winner gets, 
compared to the runner up. But what if the voter has voted for both? 
In that case, yes, both votes are counted, but that's moot. The vote 
has no effect on the result. The ballot could be discarded, same 
result. (Except that there *could* be majority failure for the 
winner, unlikely but possible; in that case, we are looking, again, 
at only one vote being counted in the end.)

James, you are out on a limb. Voters unfamiliar with voting systems 
and how they work do often come up with what you've said as a 
knee-jerk response to Approval. However, and the matter has been 
considered for many years, it was argued and debated eighty years ago 
in the U.S., and it's settled, in fact, that Approval doesn't violate 
one person, one vote.


> > In a ranked method, generally, such as STV, the voter may possibly
> > vote in all pair-wise elections, except that with STV some of these
> > votes aren't counted.
>
>STV is not a series of pair-wise elections.  In STV the voter 
>indicates contingency choices.  These contingency choices (successive
>later preferences) are considered only in the contingency that the 
>voter's ONE vote has to be transferred.

That doesn't change the fact that the voter casts votes in *possible* 
pairwise elections. STV is a truly complex voting system, compared to 
just about everything else.

> > With a Condorcet method, the votes all count.
>
>Yes, all the marked preferences will allow the voter's one vote to 
>be used in as many pair-wise comparisons as the voter wishes to
>participate in.

Yes. Personally, I find it offensive that I can cast a vote and it is 
not even counted.

> > Think of it as IRV with a different method of deciding whom to
> > eliminate.
>
>I have heard this suggested for IRV (and STV-PR), but such a method 
>of deciding the next elimination would not comply with
>Later-No-Harm.

That's true. So? Which is more important, finding the best winner, 
the candidate who will most satisfy the voters, behind whom they can 
most effectively unite, or satisfying, in the extreme, LNH?

LNH isn't a criterion that actually improves results. It's one that 
supposedly motivates sincere votes, that's about the limit of it. It 
actually fails in this, to a large extent, people still bullet vote 
or don't use up their ranks, or don't vote for a frontrunner in the 
ranks they have.

Based on what I've seen so far, Bucklin sufficiently separates the 
first preference from additionally approved candidates that voters 
aren't impeded. They add additional preferences if they have weak 
preference against them, and not if they don't know any more to rank 
or they have strong preference. That's all. Same with IRV, in fact.