User:RobLa/Burlington2009

From electowiki

As readers of Electowiki know, I'm Rob Lanphier, aka User:RobLa. This page is my point of view about what happened in Burlington in 2009. Why do electoral reform advocates obsess about the 2009 mayoral race in Burlington, Vermont? Well, I have a few theories. See the #Burlington2009 section below to learn more....

Burlington2009

hashtag Burlington2009

This page discusses the reason why I think that Burlington eletion was a mess in 2009.

An outline for this page:

  • First, I'll bring up a couple of electoral criteria:
    • #Monotonicity - There are many people that obsess about the monotonicity criterion failure that happened in 2009. However, that obscures the biggest problem with what happened in Burlington, which was the Condorcet failure
    • #Condorcet criterion - this was the indisputable failure of instant-runoff voting in that election. It was clear immediately after the election that IRV/RCV picked the wrong candidate.
  • Then, there's my diagram, which I'm still pretty happy with:
    • #Diagram - my diagram that I made shortly after the election

Monotonicity

main article: Monotonicity

There are many people that obsess about the monotonicity criterion failure that happened in Burlington. It's a red herring, though. The real problem was with the failure of the #Condorcet criterion

Condorcet criterion

main article: Condorcet criterion

The Condorcet criterion is not that hard to explain. In short, if a simple majority of voters prefer candidate A over candidate B (and express that preference on their ballots), then candidate B should not be elected.

Infographic

hashtag "#SVGbyRobLaBurlington2009"

Back when the election first happened, many of us on EM-list came to realize that what we feared would happen with IRV/RCV actually happened. We thought that this would be our "I told you so moment". I put a lot of work into an infographic about the election, for which you can find the SVG here: File:Burlington2009election-infosheet.svg

We thought it was going to be a learning moment, but alas, it wasn't meant to be. But it serve as a good recruitment tool for Electorama.com, and led to RBJ joining the mailing list.
The following section is an interview about with RBJ about the Burlington election in 2009.

Interview with RBJ in 2020

email interview with Robert Bristow-Johnson ("RBJ"), started in May 2020 and still happening...
Below is an interview with Robert Bristow-Johnson, conducted via email. Robert has lived in Burlington for a number of years, working in a variety of roles as an audio engineer, having written "The Audio-EQ Cookbook". Back in 2009, ValhallaDSP had this to say about him[1]:
If you program audio DSP effects, you have undoubtedly heard of Robert Bristow-Johnson. If you haven’t heard of him, get yourself a copy of his Audio-EQ Cookbook. This is a succinct little document, that provides coefficient generation formulas for pretty much any second order filter building block you would ever need: lowpass, highpass, allpass, peak EQ, notch, shelving filters, etc. Today, I am implementing these formulas for my own code, but I know I’ve implemented them in at least 3 prior DSP environments. Really good stuff. A quick Google search doesn’t turn up a ton on Robert Bristow-Johnson (or RBJ, as he is often abbreviated). He worked at Eventide, Fostex, Young Chang (Kurzweil), and Wave Mechanics. RBJ has a few AES papers as well, all of which are worth reading, covering topics ranging from PSOLA pitch shifting (made famous by AutoTune) to filter design to interpolation for oversampling.
Robert is a longtime citizen of Burlington, and has long been troubled by the persistent drive toward Instant-Runoff Voting (IRV) in Burlington. In 2014, he was invited to write "Having it Both Ways" for the local "VTDigger" publication from The Vermont Journalism Trust, where he pointed out a strategic mistake (or two, or maybe more) that the Republicans in Vermont made. I'm grateful for the opportunity to interview Robert via email[2]:

Question #1: introduction

Q: (User:RobLa) - It's my understanding that you were living in Burlington back in 2009, when this election took place.  Is that correct?  If so, can you tell me what it was like?

A: (Robert) - My family and I moved to Burlington in 2000 and it was always my permanent address and voting address since 2000. I've voted in nearly every election there in the 21st century and have served as an election official in Ward 7 for three years. Ward 7 is where Bernie and Jane Sanders live. There's a photo of Bernie voting in Super Tuesday 2016 and me standing in the room that someone snapped offa the TV screen. I've done some other activism affecting Burlington elections in that I drew the map that was eventually adopted in defining Burlington's 8 wards. [See Robert's fuller account in a 2014 exchange].

Both Vermont as a state and particular towns like Burlington are pretty far left of center, in an authentic and wholesome way. Bernie Sanders, in becoming mayor back in the 1980s, essentially founded the nation's most successful third party (in terms of getting people elected to government), the Vermont Progressive Party. There are a lotta state legislators and city councilors with "P" after their name, yet Bernie never put a "P" after his name, nor a "D" until he ran for President in 2015. Instant-Runoff Voting (IRV) was happily adopted in 2005 and used first 2006. Now the label Ranked-Choice Voting (RCV) is used for the very same method as IRV, although it's a neologism and a misnomer. Ranked-Choice Voting originally means any election system using ranked-order ballots where voters mark their candidates #1, #2, #3...

In March 2009 (first Tuesday, what we Vermonters call "Town Meeting Day") Burlington held the second mayoral election using the "standard" Instant-Runoff Voting method of RCV. This election had 5 candidates on the ballot, 4 candidates were credible with a serious financial investment into publicity and promotion. Several candidate forums with all 4 candidates. Going into Town Meeting Day it was a toss-up between three candidates that were all plausible winners.

The result was that the Republican Kurt Wright had the plurality of first-choice votes, the Democrat Andy Montroll was the Condorcet winner (a.k.a. the "Pairwise Champion"), and the Progressive Bob Kiss, the incumbant, was the IRV winner and therefore was re-elected mayor of Burlington for another 3 years.

A lot of people expected Andy to win. He was the candidate in the center, had a lot of first-choice votes (but came in third) and far more second-choice votes than any other candidate. Almost immediately University of Vermont (UVM) Legislative Research Shop (led by Prof. Anthony Gierzynski) and outside analysts Warren Smith and Brian Olson examined the individual ballot data and reported that Andy Montroll was the Condorcet winner, the candidate who would beat any other candidate in the IRV final round, had Andy been in the final round.

Right away many people were suspect of the legitimacy of the election result and called for the repeal of IRV. That repeal effort was successful and Instant Runoff Voting was repealed the following year, at Town Meeting Day 2010 with a 4% margin of defeat. I voted to retain IRV, because it was better than First Past The Post (FPTP) and I had hoped that it could be reformed into a Condorcet-compliant RCV, but I was on the losing side.

Supporters of Kurt Wright, the plurality winner (of first-choice votes), thought he was robbed. A few people like me understood that it was Andy Montroll who was robbed. The Progs and the IRV happy talkers became entrenched and insisted that their candidate was legitimately re-elected (and he was, according to the law) and that nothing went wrong with the election but supporters of losing candidates were sore losers. But, just like the 2000 or 2016 presidential elections, in a democracy there is a real problem when the majority of voters are sore losers.

When I found out (from the news) that Andy was the pairwise champion, I obtained the same ballot data myself from the city clerk's office and wrote my own computer program to scan the ballot rankings and tally the election using Condorcet rules and confirmed the results although my numerical results are slightly different than either Warren Smith's or Brian Olson's numerical results. But the qualitative results, who beats whom, are exactly the same. The ballots I counted were exactly consistent with those counted in the IRV tally so my subtotals agree exactly with the official round-by-round tallies of the IRV election.

Question #2: knowledge of Condorcet

Q: (User:RobLa) - That's interesting! How did it occur to you to try tabulating the pairwise results? How well did you understand the Condorcet methods prior to the 2009 election?

A: (Robert) Well, it was because the local news reported, within a week or two, that in some sense Andy really won the election. How was that? And I knew that the only sense it could be was that Andy was the Condorcet winner (CW) and I knew that if the CW gets to the IRV final round, the CW will beat anyone there and win the IRV election. I knew who was, locally in Burlington, responsible for introducing IRV to Burlington, Terry Bouricius, who is, I believe, a member of the EM Mailing List because it was he that suggested that I join it. There were dueling commentaries in the local papers between Terry and UVM Prof. Anthony Gierzynski about whether something actually screwed up in the IRV election. It wasn't until I actually got on the EM list that I figured out who Warren Smith is and that it was Warren who was in command of the technical details and was actually the fact source behind Prof. Gierzynski. I didn't see Warren's page at https://www.rangevoting.org/Burlington.html until later but somehow found out that there was raw ballot data available from the Burlington City Clerk voting website burlingtonvotes.org .

Question #3: knowing about Condorcet before 2009

Q: (User:RobLa) - So you knew about Condorcet winners before the 2009 election? How well did you understand electoral methods before 2009?

A: (Robert) - Before 2009, I remember reading a good article in Scientific American by Partha Dasgupta and Nobel laureate Eric Maskin ( https://www.scientificamerican.com/article/ranking-candidates-more-accurate/ ) where the authors dub Condorcet rank-order voting as "True Majority Rule". I realized that the IRV rules were not the same as the naive Condorcet rules, but I could tell that if the CW gets into the IRV final round, the Condorcet Winner (CW) will win IRV. So I wasn't worried. In fact, later, the paper Risk-limiting Audits for Nonplurality Elections by Sawarte et. al. ( https://hovav.net/ucsd/dist/irv.pdf ) has shown, in Table 1, results that all of these other IRV elections, including Burlington 2006, ended up electing the CW. Only Burlington 2009 failed to elect the CW and my feeling about that is exactly the same as regarding the Electoral College in the U.S. The Electoral College does a pretty good job of electing the popular vote winner, except when it doesn't. And when it fails to elect the popular vote winner, the elected President never has voter legitimacy and is handicapped in their role in office. Similarly, when IRV fails to elect the CW, the elected IRV winner will never carry the voter legitimacy that the CW would simply because more of us voted for the CW.

Question #4: The 2010 Repeal

Q: (User:RobLa) - With your math background and having read about Condorcet, you had a bit of head start on your fellow Burlington citizens when it comes to understanding the difference between a Condorcet winner and an IRV winner. When voters repealed IRV in 2010, do you feel like they were doing it because they understood that distinction, or was the 2010 repeal of IRV more of a referendum on Bob Kiss's performance in office?

A: (Robert) It is true that there was a lot of confusion and disillusionment regarding the 2009 IRV result. Like the three parties, there were at least three different positions regarding IRV.

  • Progressives (the party of the reelected mayor) and "IRV Happy Talkers" denied that anything went wrong with the election parroting FairVote's position. A parody of their position: "Of course it was a huge success! No voting machines exploded or burst into flames. A majority of voters did not suffer from paper cuts." Their actual claims: "The Burlington election was a model of clean, open debate without 'spoiler' concerns..." and "Burlington's instant runoff voting (IRV) election went off without a hitch in 2009. If anything, it was even more successful (than in 2006). IRV clearly worked as intended to avoid the "spoiler" dynamic..." are technically and decidedly false. I knew that, I tried to be clear about it, but I was also opposed to the repeal question. I had hoped to reform ranked-choice voting rather than get rid of it.
  • Republicans (and conservative Democrats) considered the GOP candidate, who had the plurality of first-choice votes, as the candidate who was robbed. But that is just rejecting the whole notion premising ranked-choice voting that the plurality winner is not always the candidate truly preferred by the majority of the electorate when there are more than two candidates. Republicans in Vermont found out in 2014 in the gubernatorial election when the Libertarian candidate for governor was most likely a spoiler and caused the GOP candidate to lose. https://vtdigger.org/2014/11/11/robert-bristow-johnson-ways/ Sometimes people get hoisted by their own petard.
  • Democrats were divided on IRV. Many of us really expected the Democrat candidate to win because he has broad (although not as deep) support. He was overwelmingly the second choice of voters who voted for someone else as their first choice. His loss on March 3, 2009, was a bitter surprise.

Some IRV opponents gathered on the internet some false beliefs regarding what happened. They made a category error at first claiming that people were "disenfranchised" with IRV drawing from some complaints of the San Francisco and Oakland RCV elections. In those elections, there were more than a dozen candidates and only three levels of ranking. So you could mark your 1st, 2nd, and 3rd and none of your votes counted in the final round and you were not able to weigh in on the actual selection of the mayor. But Burlington 2009 had 5 candidates and 5 ranking levels. No one were denied an opportunity to weigh in on any of the candidates. Ballot access requirements are important. We don't want just any old schlub to get on the ballot but only those who demonstrate some prior base of support.

So, ironically, the choice we had in 2010, was between Dumb and Dumber, which is the same problem as with the two-party system. Dumb was IRV and Dumber was FPTP, and in 2010, Dumber won and Burlington Vermont lost its ranked-choice voting by about a 4% margin. I was a voice in the wilderness calling for a third way of looking at it, but the two sides were so polarized and so unwilling to separate themselves from their own talking points, that fewer voters understood, despite some newspaper analysis that did make clear that the Democrat candidate would have beaten either the GOP or the Prog in the IRV final round, had he been in the final round. Some responses to that point was simply "so what?". They didn't understand that this clearly spells out the spoiler scenario.

Question #5: Reforming RCV

Q: (User:RobLa) - You said in your previous answer that you opposed the 2010 repeal initiative, and that you preferred to "reform ranked-choice voting rather than get rid of it". What reform would you propose to ranked-choice voting?

References and footnotes

  1. Costello, Sean (2009). "DSP Hero: Robert Bristow-Johnson". ValhallaDSP.
  2. I also had the opportunity to have a "30 minute" conversation with Robert that became an enjoyable two hours for both of us, but that part was off-the-record.