Yesterday we ran a piece by political scientist Tony Gierzynski, pointing up what Gierzynski feels are serious flaws in the IRV system now in place for Burlington’s mayoral race. Email response was sharp and swift, but no letter longer or more detailed than this from Fairvote’s Terry Bouricius. This is an issue VDB is interested in seeing from all sides, especially after this past Town Meeting Day. With that said, the floor is yours, Mr. Bouricius. — PB

Dear VDB,

Prof Gierzynski has it wrong about IRV — and in amazingly big ways. His analysis is deeply flawed.

Firstly, he believes IRV is more complex and discriminates against some voters as a result. Note that in the 2009 mayoral election, the recount showed that there was only ONE defective ballot in the entire city (the original machine totals showed four such ballots but the manual recount found three of those to be valid ballots, two of which went to Wright and one to Kiss).

That is, 99.99% of IRV ballots cast were valid. No sign of IRV leading to more errors.

Also, voters in the low-income renter wards were as likely to use additional rankings as voters in the more affluent wards. (81.9% of voters in Wards 2 and 3, compared to 81.8% of voters in Wards 4, 6 and 7.) There is no evidence of any class bias in the actual use of ranked-ballots.

It is also worth noting that for MANY voters it is much EASIER to rank candidates than to vote in a vote-for-one election, which requires making the often extremely difficult decision of whether to vote for a true favorite choice, or a “lesser of two-evils” candidate with a better chance of defeating a hated candidate.

This is particularly true in a city election where the lack of polling numbers forces voters to do their own research or guess. If a voter in Burlington using plurality voting wanted to make sure a particular one of the four front-runners was stopped, she wouldn’t have any easy way of knowing which candidate stood the best chance of blocking that hated candidate. Talk about putting a huge burden on voter research.

Plurality can be MUCH more complicated for the voter to figure out than IRV.

Also note that two scholarly studies of IRV elections in San Francisco found that districts with high numbers of low-income and African American voters had a LOWER rate of uncountable ballots with IRV, than with non-IRV races in the same other district (this combines both undervotes and overvotes). One of these studies also found that the rate of voter participation in the ultimate runoff round among voters in such disadvantaged districts increased by as much as three-fold with IRV over the old system of separate runoff elections — far more than the increase resulting from the use of IRV among voters in more affluent districts.

Secondly, Gierzynski uses the terms “paradox” and “perverse outcomes.”

He uses the term “thwarted majorities paradox” to describe an example of a “Condorcet-winner” (the technical term) such as Montroll, who would beat any of the other candidates in one-on-one race, to attack IRV. He fails to acknowledge (or maybe didn’t even understand) that this “paradox” is far MORE prevalent in the case of plurality elections and separate runoff elections. Montroll came in third in terms of first choices, and by plurality rules this “Condorcet-winner” would lose badly.

In a traditional runoff election he also would have been eliminated and couldn’t win. IRV gives such “Condorcet-winners” the BEST chance of winning the election of any voting method used by any government anywhere in the world, yet Gierzynski perversely twists this into an attack on IRV.

Vote-for-one ballots simply don’t gather the needed alternate preference information from voters that reveals the paradox under plurality and two-round runoffs.

Gierzynski also refers briefly to the non-monotonicity issue (in unique scenarios raising a candidate’s ranking can hurt that candidate), without acknowledging that this non-monotonic dynamic is significantly MORE prevalent in two-round runoff elections, because voters can vote strategically to advance a weak opponent for their true favorite choice in the first round and then switch their first choice to their true favorite in the runoff.

Thus if this “paradox” is of concern to someone, moving away from separate runoffs to IRV is an improvement.

Finally, Gierzynski attacks IRV for not forcing citizens into a two-party system (what some call a duopoly).

Fundamentally, he does not favor multi-party democracy — the norm throughout most of the democratic world. He suggests plurality elections promote political coalition building, as if we should just pretend the “spoiler” dynamic and animosity between Democrats and Progressives didn’t exist.

In contrast, IRV has been proven to promote coalition building across partisan lines in Australia, San Francisco, and even in Burlington (you may recall that 2006 working-class-Republican mayoral candidate Kevin Curley urged supporters to rank Kiss second in order to stop the upper-class Hinda Miller).

In sum, I think his is a very weak analysis.

Yours,

Terry Bouricius

_______

Terry Bouricius is a former Progressive Party city councilor and state legislator, past state board member of the League of Women Voters, and a current policy analyst with the national nonprofit FairVote: the Center for Voting and Democracy.

Late Update, Monday 6:09 am:

Of course, this back and forth could go on forever, and we have other fish to fry (like Joe Lieberman, who now says he may attempt to weasel his way back into the Democratic Party). But we wanted to give Tony Gierzynski a chance to rebut Terry’s comments, and so here briefly is that rebuttal:

“The extent of the perverse outcomes for the 2009 mayoral election in Burlington is actually much worse than I indicated. A number of mathematicians with interest in these sort of electoral systems extended my previous analysis and found that the 2009 election suffered from not only the “thwarted majorities” or Condorcet’s paradox, but also the “no-show paradox” that shows that Wright voters who preferred Montroll over Kiss (that is, ranked Montroll 2nd) would have been better staying home and not voting at all.

“The election also evinced the property of nonmonoticity—additional votes for Kiss could have made Kiss lose. It is unequivocally clear that IRV did NOT result in a majority winner. (I highly recommend the full analysis which can be found at http://rangevoting.org/Burlington.html.)

“Finally, as I stated on the Mark Johnson show, I am not arguing that the current system with a runoff is a better system; I am instead arguing that IRV is not superior to the current system as argued by the IRV supporters. There are better than ways to solve the majority problem than both of these (for a discussion of one alternative see a discussion of “range voting” at http://rangevoting.org/).

“What I think works best is for the factions who share a common ideology (Progressives and Democrats here in VT) to work out their differences in primaries or caucuses and then put forward one candidate.”