Because of a court ruling this morning, it looks very likely that Minneapolis voters will have their first experience with Instant Runoff Voting (IRV) in the 2009 municipal elections.

I'm familiar with the main arguments for and against IRV and, on balance, I'm all for giving it a try. IRV assures that the ultimate winner of an election will have a majority mandate, and it frees voters whose real preference is for a smaller party from having to decide whether to vote for their true choice or vote for the lesser of two evils among the top two candidates in order to avoid "wasting" their vote.

But in the context of the Franken-Coleman recount, it has often been argued in various discussions threads I've read that IRV is the way to avoid such expensive, time-wasting and agonizing recounts. As a matter of math and logic, this is not so. IRV is at least as likely to cause a recount as it is to prevent one, and perhaps slightly more likely to cause one.

Jeanne Massey, executive director of FairVote Minnesota, the leading advocate of IRV, agrees that IRV would not preclude recounts and this is not an argument that FairVote makes on behalf of IRV.

Before I expand on that argument, if you are unfamiliar with IRV here are the basics. (If you already know the basics, please skip to the subhead below, "The logic and the math."

Here's a paragraph from FairVote, with a couple of adjectives removed, on what IRV is:

"IRV is a ... system by which voters rank candidates in order of preference (1st choice, 2nd choice, 3rd choice, etc.), ensuring majority winners in single-winner races where there are more than two candidates on the ballot. Under IRV, voters cast their vote for their favorite candidate knowing that if he or she doesn't gather enough votes to be one of the top two finishers, their votes will count toward their second choice. Votes cast for less popular candidates are redistributed to more popular candidates, based on the voters' second choices, until one candidate emerges with a majority of votes."

In a 2006 referendum, Minneapolis voters approved the use of IRV for the 2009 election. IRV opponents filed suit to block the experiment from taking place. This morning, Hennepin County District Judge George F. McGunnigle summarily dismissed the challenge. Unless it is appealed and a motion to prevent Minneapolis from proceeding is granted by the appeals court, Minneapolis will adopt IRV for the November city election. (In case you missed it, Minneapolis Mayor R.T. Rybak announced yesterday that he would seek a third term. City Council seats will also be up.)

The St. Paul City Council has said that if the courts concluded (as McGunningle did this morning) that IRV is constitutional, it would give St. Paulites a chance to vote on whether to try IRV there. A bill before the Legislature this session would provide a standardized set of rules and procedures for any of the home rule cities in Minnesota (it's not most cities but there are 107 that have home rule charters) that decide to experiment with IRV. FairVote's long-range goal is to adopt IRV for Minnesota statewide races, which brings us back to how the system might affect close races like Coleman-Franken.

The logic and the math
If IRV had been in place for this last election, the votes cast for Independence Party Senate candidate Dean Barkley, as well as the Libertarian Party and Constitution Party candidates and write-in candidates (462,590 votes were cast for someone other Coleman or Franken), would have been assigned to Coleman or Franken based on the voters preference between those two.

Presumably (but, as a matter of logic and math, not for sure), the allocation of that many additional votes would have produced a clearer results between the top two vote-getters, perhaps enough to avoid a recount, a court challenge and a delay in seating the winner in the Senate. Conventional wisdom, based on exit polling, is that this system would have advantaged Franken. Conventional wisdom is that the presence of an Independence Party candidate on all recent Minnesota ballots has more often helped the Republican than the Democrat, which is almost certainly a key reason that there are a lot more DFL sponsors for the IRV bill in the Legislature.

But, as a matter of math and logic, if the votes had broken slightly differently, it would have been just as likely that IRV could have created the recount situation. Franken or Coleman might have been ahead by 20,000 votes when all "first-preference ballots" were counted. But when the 462,590 votes of the lower-finishing candidates were distributed based on IRV, the margin might have shrunk to a level that would have had to be recounted.

(It's quite possible, I would say likely, that under IRV the number of first-preference votes for the third, fourth and fifth-place finishers would have risen significantly, because IRV reduces the incentive for "strategic voting," in which a voter who prefers a minor candidate votes for the lesser of two evils.)

Massey, of FairVote, agrees with this logic. A close outcome requiring a recount could occur with or without IRV, there is no way to insure against such an outcome, and Minnesota should have in place a mechanism for the fairest and most efficient recount in such circumstances.

Massey and I had a slight disagreement about whether IRV makes a recount slightly more likely or slightly less. I say slightly more and here's why.

Imagine a case where, based on first-preference counting, there is a clear leader in the race, but the margin between the second- and third-place finishers is very small. Before IRV can be used to assign the second-preference votes to the top two finishers, you might need a recount to determine who finished second. That possibility doesn't exist in the traditional system.

Massey agreed, but noted that under the IRV legislation that FairVote supports, the provision by which a very small margin automatically triggers a recount, at public expense would apply only to the final count between the top two finishers. In what you might call the tie-for-second scenario I just described, the candidate finishing third in the original count would have to ask for and pay for the recount.

Massey believes there are statistical reasons that IRV makes a recount less likely. The more candidates you have receiving votes (and therefore the smaller the number of votes per candidate), the more likely it is statistically that the top two will finish in a tie or a virtual tie. IRV increases the sheer number of votes to be divided between the top two finishers, and the higher the number the lower the probability of a tie.