[EM] New MN court affidavits by those defending non-Monotonic voting methods & IRV/STV

[Abd ul-Rahman Lomax]

At 01:23 PM 11/6/2008, Kathy Dopp wrote:

>The third doc is by the Minneapolis, MN City attorney.

11SuplementaryReplyMemoinSupportofMotionforSummaryJudgment.pdf

This document follows the same errors that Austen-Smith promoted, not 
surprisingly.

It notes that plaintiff has asserted monotonicity failure as a 
problem of IRV. I'll note, right away, that, unless this is tightly 
correlated with the language of Brown v. Smallwood, this may have 
been poor legal strategy.

As you may know, I think that Brown v. Smallwood is ripe for 
reversal, and that public policy would indicate that we should 
support total reversal, and not what defendants have been seeking, 
which is partial reversal, or specification of the BvS result to 
other than sequential elimination methods which satisfy Later-no-Harm 
and which only consider, at most, one vote at a time, during the 
process. (This would clearly be a reversal of the original thinking 
in Brown v. Smallwood; FairVote has been arguing from dicta that can 
be read as referring to Later-no-Harm, but there is other language in 
BvS that very clearly proscribes all forms of alternative vote.)

As distinct from Bucklin Voting -- or many other possible reforms, 
such as Approval, Range Voting, or Concorcet methods, which still, in 
the end, only *apply* one vote from each voter, but which may 
consider all of them in the process. A result as desired by the 
defendants would damage election method reform in Minnesota for many 
years to come, I'd expect; the only silver lining is that Brown v. 
Smallwood has no application outside of Minnesota and was, 
essentially, a rogue decision, out of step with other states.

Brown v. Smallwood clearly stood for the principle of one person, one 
vote, totally and completely, as to any single election. It did not 
allow any "alternative" votes. IRV flies directly in the face of BvS, 
and FairVote propaganda, prior to the challenge, simply dismissed 
this, encouraging Minneapolis, for example, to disregard the advice 
of its own attorney (in advice given prior to the adoption).

There is a possibility that monotonicity failure can be connected to 
the specific arguments given in Brown v. Smallwood by the court 
majority. I'd have to study that in particular, but I consider it 
only a thin possibility.

As to immediate practical result, I'm with FairVote. IRV should be 
allowed in Minneapolis, providing only that it was properly adopted 
(BvS aside). Brown v. Smallwood should be tossed, and, once again, 
Bucklin voting, which is far, far simpler to count than IRV, and 
which has superior performance to IRV (slightly), should be on the table.

Note that a finding for the plaintiffs in this will, in my opinion, 
damage future voting reform in Minnesota. I very much dislike IRV, in 
comparison to top-two runoff, a vastly superior method, which it is 
replacing. (If that is what Minneapolis had, I haven't checked, but 
this is what has been replaced in every other election held, I think, 
since San Francisco started up a few years ago.).

My concern is that FairVote will prevail on their arguments such that 
sequential elimination preferential voting will be considered legal, 
but not any other form of preferential voting, thus giving us 
Later-no-harm, a minor and quite controversial "benefit," instead of 
such basics as Condorcet compliance (which, if you are going to use 
pure preferential ballots, is truly fundamental, and we only go 
beyond Condorcet compliance when we consider preference strength, 
where, in the end, the Condorcet criterion turns out to be shallow, 
making sense only when preferences are strict and equal.)

The attorney attacks Ms. Dopps' competency as an expert witness. 
That's probably cogent; that is, it seems reasonably likely to 
prevail. The matters at issue here are constitutional ones and, as 
the city attorney notes, such problems as cost or election security 
are irrelevant.

I write a lot, and, believe me, were I writing for an affidavit to 
file with a court, you'd see far less text. What I might think about 
contributing, if there is time, is an analysis of Brown v. Smallwood 
and why it was defective, as well as the deficiency of FairVote 
arguments attempting to discriminate between Bucklin voting (then 
called "preferential voting," same name as IRV was known as) and IRV 
based on sequential elimination.

This would be an argument for the defense, though.

But credentials, I have none. I think the monotonicity issue, though, 
is a huge red herring. Monotonicity failure by IRV is a *symptom* 
that shows erratic behavior, not of great practical consequence, 
because the circumstances where it would affect the result seem to be 
rare; there are much more serious problems, specifically Center 
Squeeze. But Plurality, of course, exhibits Center Squeeze.

IRV is really a fancy, expensive, form of Plurality, unfortunately, 
once we start to look at how it actually works in real nonpartisan elections.

There is an interesting method being used, now, in some states, which 
FairVote is calling IRV, though it is really a form of Contingent 
Vote. It's not IRV at all, except that a contingent vote is cast by 
absentee voters, to be used in a runoff election, if it is held. That 
is a *separate* election, it has a different actual voting 
electorate, with a separate campaign.

I think it may be unconstitutional, on totally different grounds than 
are at issue in Minnesota, if it deprives absentee voters of the 
right to later change their votes, as afforded to all other voters. 
(However, if the process allows voters to specify whether or not they 
were voting in the runoff or not, presumably on the enclosure 
envelope, if they can keep the absentee runoff ballot and send it in 
separately, then it would merely be a form of early voting in a 
contingent election, affording them a right without a cost. I have 
not read the details.)

What if this right, to cast a contingent ballot, or, alternatively, 
to vote later, were afforded to all voters? That might look like IRV, 
but it would actually be quite different. You would still see those 
"comeback elections," which have been eliminated by IRV.