Some Analysis of the 2009 Burlington IRV Election
The polls closed at 7:00 p.m. on March 3rd, and at approximately 8:15 p.m. that night the IRV tally had been completed and the winner announced. In the initial count of first choices, no candidate received a majority. As the ward results were posted on a large screen in the auditoriumin of city hall, incumbent mayor Progressive Bob Kiss had a plurality lead as the first five wards were posted. but when the votes from the final two, more conservative-leaning, New North End wards reported, the lead swung to Republican councilor Kurt Wright. With all first choices tallied, Wright was in the lead with 33% (2951), Kiss was in second with 29% (2585), Democratic city councilor Andy Montroll had 23% (2063), Independent Dan Smith was in fourth place with 15% (1306), Green Party candidate James Simpson (35) and write-ins (36) combined were less than 1%.
Since there was no initial majority winner, the instant runoff tally followed immediately. Since a record of every ballot's set of rankings was generated by the optical scan voting machines, it took less than a second for the computer to complete the IRV tally. Note that all of these individual ballot records are posted on the city's web site (http:/www.burlingtonvotes.org) so that anybody who wishes can manually, or with a spreadsheet, verify the IRV tally for themselves.
In the first round of the runoff, all of the write-ins, as well as Simpson and Smith were defeated, since none of them had a mathematical possibility of winning. Democrat Andy Montroll was the biggest beneficiary of the alternate choices of voters who had ranked these eliminated candidates first. Bob Kiss received the second most, and Wright received the fewest. This narrowed the gap between Wright and Kiss, but left Wright ahead, and Montroll still in third place.
In the final runoff round, Montroll (now in last place) was defeated, and his supporters favored Kiss over Wright by nearly two to one. This catapulted Kiss into the lead and victory. Among voters expressing a preference between these two finalists, 52.5% (4313) favored Kiss and 48.5% (4061) favored Wright. In addition, four ballots were spoiled and 602 did not indicate any preference either way between Kiss and Wright (sitting out the final runoff).
Wright and some of his supporters have complained in the news media, talk radio and Internet about this outcome, suggesting that maybe the candidate with the 33% plurality in the initial count should have won. But IRV clearly worked as intended to avoid the "spoiler" dynamic. Wright has since partially backed off his complaint and said he has no intention of challenging or seeking to overturn the IRV method, but may seek a recount just to make clear how the system works.
Since all of the ballot data is made publicly available by the city, it is possible to run a variety of "what if" scenarios. For example, What if Montroll had been more popular and had made it into the final runoff against Wright instead of Kiss? If we remove Kiss from the tally altogether (as if he hadn't run), and transfer all of those ballots to the next choice indicated by each voter, we again have no majority winner in the initial round, but Montroll would also have defeated Wright in the final runoff tally.
While Sore losers in Burlington are complaining about sour grapes, instant runoff voting has proven itself again as a bulwark of democracy.