IRV: A Grain Of Truth, A Ton Of Bulloney

Could IRV (Instant Runoff Voting) avoid recounts?  A slew of stories from Minnesota claim that recounts are a headache and that IRV is the cure.  Two examples:

The Star Tribune <read>

Instead, the $40 million-plus campaign continues to permeate our headlines and limit our forward momentum. The Coleman-Franken race is now in a contentious recount and is almost certainly headed to the courts from there. The recount and its aftermath will be a protracted and high-priced affair, and no matter the outcome, most voters will be left wondering if there is not a better way to express our preferences.

Instant-runoff voting (IRV) would have produced an entirely different election.

The South St. Paul Examiner <read>

In a contest so close as the one that has unfolded in Minnesota, a hand recount is likely the only alternative to determine who is number one and who is number two to figure out how number three’s (Dean Barkley) votes are alloted. The election between Landslide Al Franken and Landslide Norm Coleman is an anomoly. It may go down as the closest race in the history of the United States, but IRV would save us the indignity of staring at Big Goof and Little Goof’s ballots and the money a hand recount costs in almost all other instances.

IRV Defined

From the Examiner:

First choices are counted (one, two, or three). If no candidate receives a majority, the candidate with the fewest votes is defeated, and those votes are transferred to the next ranked (one or two) candidate on each ballot. The votes are recounted.The process continues until one candidate has a majority of the votes and is declared the winner.

Many believe that if IRV was used in Florida in 2000 then Ralph Nader would have received many more votes and that Al Gore would have easily won the state and the Presidency.   If employed nationwide many believe that George Bush I would have been elected President in 1992.

A Grain of Truth

The articles are likely correct, that if Minnesota had used Instant Runoff Voting for the Senate race and if there were only three candidates, then the winner would likely have been determined much sooner, on election night, and a recount would have been unnecessary.

Several Doses Of Reality – The “Cure” Makes Makes The “Problem” Worse

There are several issues that are glossed over in these one-sided simplistic articles:

• Recounts are necessary when counts are close. Minnesota has dramatically demonstrated that the careful counting of votes by hand is crucial to actually approaching the ideal of counting every vote and following the voters’ intentions in the declared result.
• IRV does not eliminate recounts. Ultimately, an IRV race comes down to two candidates, if the final margin is 0.5% or less, the voters deserve a careful recount.
• IRV actually makes recounts more likely.   In an IRV election at each stage, there is the possibility of a close margin demanding a hand recount.   Worse there are multiple reasons a hand recount may be necessary at each stage:   The difference between two candidates to be potentially eliminated may be close, the order of elimination can effect the final result.  The difference between the required winning margin of 50% and the highest candidate count may be less than 0.5%, the victor may change if another round of elimination is required.
• IRV will change the nature of the election. One of the differences with IRV would be more candidates competing and more votes for third parties.   Like many people, I see this as a potential advantage of IRV.    However, glossed over is that this means that voting and counting of elections gets more complex.  Today we have, including write-ins, three to five candidates for President in each election.   With IRV, expect six to twelve or more!  In Connecticut we can expect similar numbers for Senate races.  These numbers also add to the rounds and add to the frequency of recounting.
• IRV will make recounting more complex and expensive. The order of voter preference will be critical – recounting an IRV race is straight-forward, but much more time consuming than recounting under the current voting system.

Recounting Is Not A Problem – Its A Symptom Of Election Integrity

CTVotersCount  does not share the concern with the time and cost of recounting.  In the overall scheme of things the expense of recounting is small compared to what is spent in running elections, what is spent on campaigns, and the billions at stake in the decisions and actions of a Senator.  Who cares if the winner is declared on Nov 4th or Jan 31st?   If its important that Minnesota have two Senators on Jan 6th.   Then its even more important that Minnesota have the voters choice as Senator for the six years after Jan 31st.

From the Examiner:

Minnesota and its reputation for good government have taken some hits from the Senate brouhaha. It’s too late to duck that punch or the ones we can expect in the weeks — and maybe months — yet to come. It’s not too late, though, to make sure we don’t get into this situation again.

It’s time for Minnesota to consider adopting IRV and preserve its tradition as a leader in electoral integrity and good governance.

The author has it all backwards.   Minnesota’s leadership in electoral integrity is why   manual recounts are required.  Looking objectively at the care taken with every aspect of the recount only enhances Minnesota’s reputation for good governance.

As A Cure, IRV May Be Worse Than The Problem

IRV sounds great.  We want a vibrant democracy.   Wouldn’t it be great if we could vote for our favorite candidate and still have our preference expressed in the result.   We could vote for Pat Buchannan, Ross Perot, Ralph Nadar, or Dennis Kucinich, yet still see Al Gore or Bob Dole elected President.  Who knows, perhaps in a few years we would have a third viable party.

I live in a condominim, IRV might be a very good way for us to elect directors.

How could anyone possibly object?    Consider that the cure may be worse than the disease:

• IRV means a more complicated, huge ballot. Consider a ballot with an average of four candidates for each of five offices (similar to November in CT, with a few candidates added because of IRV).  Instead of five bubbles to fill in, a contentious voter would need to fill in fifteen bubbles (A ranking of three choices out of four for each of five races).   In a municipal election, voting for six out of say 18 candidates for town council, instead of 6 bubbles, we would fill in 17 bubbles – instead of 18 bubbles  on the ballot there would be 324 (18 x 18), for just this one race.   The ballot would be large, voting would take more time,  more knowledge, and more voter education.
• IRV means more complicated and slower election accounting. For my condo or the local mayor’s race it may be worth it, its not that complicated.   How would we handle a state wide race for Senate with ten candidates?   We would have to know the order of votes on each ballot or have a count from each precinct of each possible combination (4,838,400 not including various combinations of over votes and under votes) – every vote would have to be maintained and accumulated electronically, votes that are counted by hand today would have to be input electronically as well.  Once again, this only considers the complexity added for just one race, the whole process is even more involved.
• Recounting would be complex and time consuming. It would be straight-forward.   One way would be to count each round separately – each round costing about the same as the Minnesota recount.  Perhaps there is a way to speed it up?  Even to count all the 1st and 2nd choices of each voter in our ten candidate race would mean 10o possible combinations (10 x 9 + 10 more for votes for only one candidate)
• Post-Election Audits would be more complex and time consuming. Once again, it would be straight-forward.
• There would be pushes for more electronic voting, less recounting, less auditing. Pre-election testing would be less adequate, more complex, and more expensive.  The whole process of election integrity would be more difficult for election officials to comprehend and execute well.

So there we have it.   Instant Runoff Voting – a ready cure for many real election concerns – yet, risky, complex, and time consuming, quite likely causing more and greater problems than those it is intended to cure.

PS: Here in Connecticut there is no requirement for a hand recount.  From what we saw in the Courtney/Simmons 2006 recanvass, we question if hand counting votes here would result in the same level of scrutiny of each ballot and each rejected absentee ballot we have seen in Minnesota.  Would we come anywhere close to the detail and objectivity apparent in Minnesota?

**************
Update 3/23/2009:   Party Looks At Sample Ballot, Reverses Course On IRV <read>

“Anyone who could look at that and not think that the average voter is going to find that totally frustrating is totally out of touch with the average voter in the City of St. Paul,” Repke said.

On another flier distributed by IRV opponents: mucked-up ballots from the contest between Al Franken and Norm Coleman. During the Senate recount, goofy ballots such as one endorsing “€œflying spaghetti monster” got most of the attention. But much more common were routinely botched ballots in which a voter’s intent simply couldn’t be discerned because of unusual markings.

Thune believes IRV would only compound such problems and disenfranchise voters. “While this may seem like a wonderful thing in Cambridge for a bunch of Harvard professors, we’ve got a general population that has trouble filling out one oval in a Coleman-Franken race,” he says.