[Election-Methods] RE : Re: IRV/Approval/Range comparisons on Wikipedia

[Abd ul-Rahman Lomax]

At 08:59 AM 9/24/2007, Kevin Venzke wrote:

> > The method used in the Duluth
> > elections, the one found unconstitutional in Brown v. Smallwood, had
> > three ranks on the ballot....

>Is this sort of information regarding U.S. Bucklin elections available
>online? How did you get it?

Slogging through the quotation of the FairVote document by Wikipedia 
and then zillions of copies from that to all the copycat 
encylclopedia sites.... it wasn't easy.

Here is where I found Brown v. Smallwood, it is from an anti-IRV site:

http://www.mnvoters.org/Brownvsmallwood.htm

At the end of this page is a link to a pdf of the actual decision, 
which is well worth looking at in detail.

On the Range list I posted another election example from Ohio, a very 
interesting election... In both these elections the plurality winner 
ended up losing, and the outcome was pretty good, that is, the method 
did select what appears to have been the pairwise winner.

What I'm finding is that there is lots of material about Bucklin, 
written and published when it was popular and spreading, lots of 
enthusiastic promotion of it, etc.

And then silence. What happened? Where did it go? How was it killed? 
I'll probably need to go to a law library and look for repeal 
statutes, and to newspapers for accounts from the time. Why was it 
dropped? The FairVote explanation is essentially an IRV-warped piece 
of propaganda against Bucklin and Approval Voting and other Condorcet 
methods, and it does not make sense.

It's as bad as the IEEE justification for dropping Approval, or the 
more recent decision by the Dartmouth Alumni Board similarly. In the 
IEEE case, the given reason was "Most voters were not using 
additional approvals." Of course, if they are not, there is no cost 
(unless you have a Yes/No implementation, which has double the 
counting cost). But Bucklin, if voters were not using the additional 
ranks, there was no additional cost, unless they wanted to count them 
for fun. In fact, voters *did* use the additional ranks, and we know 
of elections that were turned. Smallwood reversed one of them, quite unfairly.

But consider what would happen if we had Bucklin today, in a strong 
two-party system. Most voters would not use additional ranks, for 
they prefer a frontrunner *by definition*. Unless they prefer a third 
party -- or *almost* prefer a third party candidate -- there is no 
reason to use the additional ranks. Or, with Approval, additional 
votes. Thus we would expect bullet voting to be the norm.

This is *not* an argument against the methods! Approval or Bucklin 
solved the first-order spoiler effect, like IRV. This is the effect 
that flips elections between one party and another, but only by a few 
percent. As you know, Kevin, IRV can spectacularly fail when an 
uppity third party actually grows in support until it is close to 
winning elections. When there are three viable candidates, the center 
squeeze effect kicks in and becomes *common,* according to the simulations.

What about real elections? Well, there is top-two runoff, which is 
similar to IRV in how it works, and which does show center squeeze. 
How about the last two French presidential elections, if I have it 
right? And then there is the fact of countries which use STV 
single-winner being all strong two-party systems. That should be a 
clue! In any case, then, we have no substantial election experience 
with IRV in three-way contests; but the theory is strong and so are 
the simulations.

We do know what happened to the Ann Arbor IRV law. It was repealed 
through a special election with very low turnout, put up by the Republicans....

I think we need to focus and stay focused on the Wikipedia article, 
and we need to get our act together. I've been saying for some time 
that we need a mechanism for determining consensus in the EM 
community, for if we can get it, and document it, it then becomes a 
source for Wikipedia. It's needed.