Talk:2009 Burlington mayoral election

Condorcet winner?

Unlike Burlington's first IRV mayoral election in 2006, the IRV winner, Bob Kiss, was not the same as the Condorcet winner, Andy Montroll.

I'm very surprised by this claim. How do we know the Condorcet winner? Where was this count documented? Are the ballots public? Is it even legal to do a Condorcet count? Tom Ruen (talk) 06:57, 12 August 2009 (UTC)

There is no citation so I'm inclinded to delete it after a month if there is no facts to back it up Gang14 (talk) 15:28, 18 August 2009 (UTC)

Condorcet should not be applicable to IRV voting, unless there is some weird oddity added into the IRV system implemented (however, there is no mention of anything different about the IRV system used mentioned in the article). You might also look at the IRV in the United States article which also mentions condorcet regarding this election. That probably should be removed in that article, as well. — al-Shimoni (talk) 12:01, 17 November 2012 (UTC)

Not applicable? If the ballots are available, IRV – almost uniquely – makes it possible to know the Condorcet winner (if any). —Tamfang (talk) 06:12, 10 December 2012 (UTC)

Reason for Controversy?

The article does not exactly explain why there was a controversy. It mentions the first round plurality, but that only means IRV worked as it was supposed to, id est not electing someone on plurality alone when there is voter splitting. If you look at the candidates as "Left Wing" and "Right Wing," then even from the first round it is obvious that the actual plurality was for a "Left Wing" candidate, a plurality that was split, giving the illusion that the "Right Wing" candidate had a plurality (a real life example of the "Hitler Hyperbole" used in advocating IRV). The IRV voting corrected that in the subsequent rounds, thus choosing the candidate that was most preferred. So, why the controversy and later repeal of the IRV system? Is this just an example of "sour grapes" by the RW supporters, or is it a lack of understanding by the voters on how IRV plays out and its purpose? An explanation of the fundamental whys of the controversy added to this article would be appreciated. — al-Shimoni (talk) 12:12, 17 November 2012 (UTC)

It is a well-established fact that the pairwise champion, a.k.a. Condorcet winner, is Andy Montroll, who came in third with respect to 1st-choice votes. The Warren Smith page that is cited spells it out. Although the supporters of the plurality winner claim their guy was robbed, the fact is that their guy lost to the pairwise champion by 930 votes and to the IRV winner by 252 votes. This is also quite well known, locally in Burlington. Deleting this information from the article is removing relevant and cited content and is contrary to Wikipedia guidelines. I hope you guys can refrain from doing it again in the future. 71.169.179.223 (talk) 02:44, 14 January 2014 (UTC)

External links modified

Hello fellow Wikipedians,

I have just modified 3 external links on Burlington mayoral election, 2009. Please take a moment to review my edit. If you have any questions, or need the bot to ignore the links, or the page altogether, please visit this simple FaQ for additional information. I made the following changes:

• Added archive https://web.archive.org/web/20110725111051/http://www.burlingtonvotes.org/20090303/2009%20Burlington%20Mayor%20Round4.htm to http://www.burlingtonvotes.org/20090303/2009%20Burlington%20Mayor%20Round4.htm
• Added archive https://web.archive.org/web/20111201060318/http://www.ci.burlington.vt.us:80/ to http://www.ci.burlington.vt.us/
• Added archive https://web.archive.org/web/20110725111632/http://www.burlingtonvotes.org/20090303/2009%20Mayor%20Reports.zip to http://www.burlingtonvotes.org/20090303/2009%20Mayor%20Reports.zip

When you have finished reviewing my changes, please set the checked parameter below to true or failed to let others know (documentation at {{Sourcecheck}}).

Y An editor has reviewed this edit and fixed any errors that were found.

• If you have discovered URLs which were erroneously considered dead by the bot, you can report them with this tool.
• If you found an error with any archives or the URLs themselves, you can fix them with this tool.

Cheers.—InternetArchiveBot (Report bug) 00:38, 11 November 2016 (UTC)

External links modified

Hello fellow Wikipedians,

I have just modified 2 external links on Burlington mayoral election, 2009. Please take a moment to review my edit. If you have any questions, or need the bot to ignore the links, or the page altogether, please visit this simple FaQ for additional information. I made the following changes:

• Added archive https://web.archive.org/web/20071129081028/http://www.ci.burlington.vt.us/mayor/ to http://www.ci.burlington.vt.us/mayor/
• Corrected formatting/usage for http://www.ci.burlington.vt.us//

When you have finished reviewing my changes, you may follow the instructions on the template below to fix any issues with the URLs.

Y An editor has reviewed this edit and fixed any errors that were found.

• If you have discovered URLs which were erroneously considered dead by the bot, you can report them with this tool.
• If you found an error with any archives or the URLs themselves, you can fix them with this tool.

Cheers.—InternetArchiveBot (Report bug) 17:59, 27 July 2017 (UTC)

Most-preferred

(The first few posts in this section were copied from my talk page) North8000 (talk) 21:48, 16 February 2018 (UTC)

Regarding this edit , "most-preferred" in this context isn't a subjective term; it's the definition of Condorcet winner. All the voters' expressed preferences are tallied up and one candidate is literally preferred over all the others. I've already provided references explaining this, but we could add more academic refs for the definition, or change the wording to "the candidate preferred more often than the others" or something more explicit, if you think that's better. — Omegatron (talk) 13:53, 16 February 2018 (UTC)

As background, I was just trying to help out a bit....there was a notice somewhere asking for more eyes. Maybe it was overly simplified for me to simply say that "most preferred" is simply a subjective term. There can be several ways to determine "most preferred", with a simple standard election being one of them, and the Codorset definition being another of them. So selection between which of those many definitions to use is subjective. IMO, if "most preferred" went back in there, it should specify the context e.g. "most preferred according to the Condorcet criteria". But I think that "Cordorcet winner" expresses that in a more succinct manner. Sincerely, North8000 (talk) 19:47, 16 February 2018 (UTC)

None of the references are advocates for Condorcet voting systems, though; they advocate Score, IRNR, etc. They're just criticizing IRV for not selecting the most-preferred candidate. Saying "it wasn't the Condorcet winner" doesn't explain why that's a bad thing.

Also I agree that there's redundancy in saying both "was neither the same as the plurality winner (Kurt Wright) nor the Condorcet winner" and "prevented the election of the presumed winner under a plurality system"/"because the condorcet winner", but couldn't think of a better way to say it. Deleting the criticism of IRV along with the duplication is not justified, however. — Omegatron (talk) 00:37, 17 February 2018 (UTC)

My point was very narrow. "Most preferred" has a very broad reaching meaning, where "most preferred according to the condorcet method" has a much more limited meaning. Merely fulfilling the latter falls short of fulfilling the former. Similarly, sourcing that establishes only the latter does not establish the former. North8000 (talk) 11:42, 17 February 2018 (UTC)

I agree with North. If the article explains that different methods would have elected different people I have no objection. If the article characterizes different methods with adjectives so that the article is not just talking about the election but is now assigning these characteristics to the voting methods then we have a problem. First the pros/cons of different methods and the debate over which is best belongs in articles about voting methods. But if its still going to be advocated for inclusion here then those adjectives will have to pass WP:Verification by citation to WP:RSs with appropriate levels of WP:WEIGHT. Blogs and webpages for groups that advocate one or the other, such as Scorevoting(dot)org are self-published WP:PRIMARY sources. We can still use them, but only for saying in the text that the advocacy group has position X about method Y. NewsAndEventsGuy (talk) 12:45, 17 February 2018 (UTC)

I can see how "most-preferred" sounds like a broad term at first glance, but it's really not. More voters expressed a preference for Montroll over other candidates than vice versa, so it's objectively true that he was "the most-preferred candidate".

Even if you disagree that this should be called the "most-preferred candidate" because it sounds biased, it's still objectively truthful to say "X criticized the election because the candidate X considers to be the most-preferred was not elected."

"Most preferred according to the condorcet method" isn't quite right either, since there are several Condorcet methods, and it still implies that the critics are advocating that we adopt a Condorcet method, which they're not. They're not criticizing the election because they like Condorcet methods; they're criticizing the election because the candidate who received more preference votes than anyone else did not win. How else can we word this? — Omegatron (talk) 14:56, 25 February 2018 (UTC)

My strength and weakness here is that I don't have in-depth knowl3ege of the technical details and many facets here. "Strength" comes from going by common meanings of a term to a typical reader. That said, any changes should recognize that, in that context, saying only "most-preferred candidate" without qualifiers or attribution a very big broad-reaching statement.North8000 (talk) 23:45, 25 February 2018 (UTC)

@Omegatron,

(A) We don't report on any editor's opinion of what is WP:TRUTH, we report on what is said in reliable and ideally WP:Secondary sources. There is an exception called WP:SKYISBLUE, but since you're having to explain and defend the desired language this is obviously not an example of such a fact.

(B) It's super-easy to say the bloggers at ScoreVoting didn't like the outcome because the Condorcet winner didn't win, because that is indeed what the PRIMARY source from ScoreVoting says.

(C) In my view, you're trying hard to preserve "most-prefered" which sounds a lot like math-and-voting-system geekspeak. See MOS:JARGON.

SUM - Like North, I'm not an expert in comparative voting methodologies either. But since Wikipedians follows sources, for most topics expertise isn't necessary if you provide inline attribution for primary sources, and simple English.

NewsAndEventsGuy (talk) 02:10, 26 February 2018 (UTC)

@Omegatron "Majority-preferred candidate" is a better way of phrasing "Condorcet winner." The issue with "most-preferred" is it can be misinterpreted as saying "most-preferred, including strength-of-preference information."

You're 100% correct that when you only have ordinal information, the standard meaning of "(uniquely) most-preferred" in social choice theory is the Condorcet winner, but when speaking to a lay audience they may not know the difference between cardinal and ordinal methods.

The criticism can be phrased more clearly by saying IRV "elected a candidate who was opposed by a majority of voters, as X would have lost 54–46% against Montroll." Link to Condorcet winner can clear up ambiguity here. Closed Limelike Curves (talk) 19:07, 20 March 2024 (UTC)

How about "pairwise champion" as described above, or "pairwise winner"? — Omegatron (talk) 17:10, 17 March 2018 (UTC)

Following sources

The text asserts FairVote called it a success for just one reason, citing a primary source from Fairvote that in fact listed three. I added all three in this edit and plan to put all three in again unless someone can convince me why all three do not belong here. NewsAndEventsGuy (talk) 12:49, 17 February 2018 (UTC)

All three reasons are already there. Read the whole section. You duplicated the "99.99%" content that I had already added in the following paragraph.

I included point 1 because it's a legitimate opinion.

I left out point 2 because it's not particularly meaningful and could equally be claimed about any alternative voting system, but you added it and no one has removed it since.

I included point 3 as a separate paragraph because everyone agrees that people can fill out the ballots correctly, but that's not an argument for or against a particular system. Everyone can fill out the ballots correctly in North Korea's voting system, too. That doesn't make it a good idea. — Omegatron (talk) 14:39, 25 February 2018 (UTC)

Your desired text is problematic because you created an Although X said success (reason 1), B said failure. But A also said success for Reasons 2 and Reason 3. You exported those away from the verb phrase (consider it a success) so the sentence doesn't comport with the source, and it sets up a false comparison. It's much easier to just say what Fairvote thought. (stop) New sentence, then say what scorevoting.org etc thought. Problem solved .NewsAndEventsGuy (talk) 02:19, 26 February 2018 (UTC)

Wikipedia is not a soapbox for promotion of FairVote's platform; we combine and distill information from multiple sources. Here are some examples as a list:

1. IRV was a success because it avoided vote splitting.
2. IRV was not a success because it failed to elect the most-preferred candidate.
3. IRV promotes competition between more than two candidates
4. IRV doesn't have a significant problem with invalid ballots
5. IRV is simple for voters to understand
6. IRV avoided the need for expensive runoff elections
7. etc. — Omegatron (talk) 16:36, 17 March 2018 (UTC)

Visualizing the pairwise matrix

(copied from Talk:Instant-runoff voting) I've also taken a look at the pairwise matrix shown at Burlington mayoral election, 2009, and am wondering whether it could benefit from visualization as well. I'll leave that for another discussion.Raellerby (talk) 14:33, 8 April 2018 (UTC)

Good point. I've made some edits to Burlington mayoral election, 2009 over the past couple of days that I hope clarify the pairwise matrix, and are hopefully good raw material for someone with better data visualization skills than I have. -- RobLa (talk) 17:39, 10 April 2018 (UTC)

Is the Burlington Telecom scandal notable enough for the summary?

In their last edit, User:RRichie changed the last sentence of the article summary from this:

The results caused a post-election controversy regarding the IRV method, such that IRV was repealed in March 2010 by a vote of 52% to 48%.

...to this:

The results and a post-election controversy involving Kiss and Burlington Telecomm, contributed to a ballot initiative that resulted in the repeal of IRV in March 2010 by a vote of 52% to 48%.

...and put this in the edit summary:

This is factually true. Wright backers had abandoned their signature drive, then restarted it months later after the Burlington Telecom controversy.

The article itself goes into detail which includes mention of the controversy surrounding Kiss and Burlington Telecom. From what I've read and understand about the election, Kiss was a beleaguered mayor embroiled in scandal, and his political weakness probably played a big role in IRV's repeal. The article itself lays it out in (what I hope is) a well-cited and NPOV way:

In late 2009, a group of several Democrats (who supported Republican Kurt Wright) led a signature drive to force a referendum on the election method. According a local columnist, the vote was a referendum on Mayor Kiss, who was a "lame duck" because of a scandal relating to Burlington Telecom and other local issues. However, in an interview with Vermont Public Radio, Mayor Kiss disputed that claim. IRV was repealed in March 2010 by a vote of 52% to 48%.

The citations in that section dispute the idea that the Burlington Telecom scandal was the dominant reason behind the repeal. I'm going to undo User:RRichie's change from today, but a better summary of the article text would be welcome. -- RobLa (talk) 19:54, 9 June 2018 (UTC)

Agreed that it doesn't belong in the intro. IRV promoters claim it was only about the mayor, while IRV opponents claim it was only about IRV. The detailed section in the article is the appropriate place for that discussion. — Omegatron (talk) 17:31, 23 June 2018 (UTC)

IRV fell to vote-splitting in this election

The analysis section states: "The election is considered a success by IRV advocates such as FairVote, since it prevented the election of the presumed winner under a plurality system by avoiding the effect of vote-splitting between the other candidates". This is FairVote's assertion, and is untrue.

In an election between Andy Montroll and Bob Kiss, Andy Montroll wins. By adding Kurt Wright as a candidate, we split the vote: Andy Montroll loses votes to Kurt Wright. Under Plurality, Andy Montroll and Bob Kiss split the vote, and Kurt Wright wins; under IRV, Andy Montroll and Kurt Wright split the vote, and Andy Montroll is eliminated, transferring votes strongly to Bob Kiss. Kurt Wright is the spoiler candidate. John Moser (talk) 14:51, 11 November 2018 (UTC)

@John Moser:: yup, you're right. Kurt Wright's introduction into this race demonstrates how IRV can violate independence of irrelevant alternatives (IIA) (more conventionally known as "allowing spoilers"). Are you aware of any solid citations that document how Burlington 2009 is a clear demonstration of IIA? The best citation I was able to find in my cursory search this morning is a 2017-01-27 politics.stackexchange.com answer by endolith. It's an incredibly thorough answer, but having a more conventionally-reputable source would be even better. -- RobLa (talk) 18:23, 11 November 2018 (UTC)

Nope. I've done some additional mathematical analysis of IRV since that shows other interesting properties, such as that when a mutual majority exists all voters who mutually rank at least one candidate above all mutual majority candidates are irrelevant so long as the number of voters |m| in the minority is greater than than the number of voters |M| in the majority divided by 2. It's a lot more complex than that but I'm not pasting unreadable set theory here in latex; and that's not going in the article right now because WP:NOR. John Moser (talk) 02:13, 5 July 2019 (UTC)

Summary Table “Share”

Ballots that do not include a vote in an election are, in common practice, excluded from vote totals because they are abstentions from voting. In the same way, ballots that become exhausted because not all candidates are ranked represent voters’ abstention from the next round of voting. For that reason, when calculating percentages for a round, the total number of “active” ballots, where votes are actually cast, is the most representative divisor.

The summary table shows the votes received in the last round in which a candidate participated, along with the “Share in Maximum Round” which was calculated as the fraction of all ballots rather then the percentage of votes actually cast. The other day I changed it to the latter since that is more representative and therefore better to use in this compact diagram. Today RobLa changed it back to what they called the “correct percentages”.

These two sets of values are equally “correct”, but those based on actual votes rather than ballots are more representative.

Opinions: which should be used here?

— Andy Anderson 05:16, 19 March 2019 (UTC)

@AndyAnderson: For the summary table, it might be reasonable to display both the percentage of the total vote, and the percentage of the "% Active" (like you added in the lower, more detailed table). However, it would be misleading to only include the "% Active" in the summary table as the only percentage. 8980 votes were cast in this election, so percentages attributed to each candidate should be the percentage of the total votes cast. For example, Bob Kiss received 4313 out of the 8980 votes cast in this election (and only after getting votes transferred to him from Dan Smith and Andy Montrol). 4313 out of 8980 is 48.029%, not 51.5%. The two sets of percentages are not equally correct. -- RobLa (talk) 03:01, 23 March 2019 (UTC)

I agree, "percentages attributed to each candidate should be the percentage of the total votes cast". But a ballot is not a vote. There were 8980 ballots, but not all of them had first-choice votes, that number is 8976. In the final round there were only 8,374 votes cast (the "active" ballots). And that’s why Kiss ended up with 4,313 / 8,374 = 51.5% of the votes cast. — Andy Anderson 03:44, 23 March 2019 (UTC)

"But a ballot is not a vote." I believe that, in common practice, a ballot is a vote, and I'm pretty sure that the voters that cast "exhausted" ballots in the 2009 election would concur. -- RobLa (talk) 20:21, 23 March 2019 (UTC)

addressing partisan motivation of the repeal

This explains some edits I'm going to make to this article.

I know anti-RCV folks like Rob L like to worship at the altar of the Burlington story and want to make this story one of disgruntled backers of the Condorcet candidate, Democrat Andy Montroll, driving the repeal. Thus, the current version of the article rather hilariously says "A group of several Democrats led a signature drive to force a referendum on the election method (supported by Republican Kurt Wright)."

But that's a false narrative. The repeal campaign was in fact led by Kurt Wright, the Republican candidate who lost after leading in first choices. The folks active in the campaign were primarily people who backed his 2009 campaign, a-nd the repeal measure only carried 2 of the the city's 7 wards - the two ward with the council's only Republican councilors and where Wright had done well in 2009. Democrat Andy Montroll backed keeping RCV, as did Bernie Sanders, prominent Democrats like Howard Dean and nearly the entire city council.

Wright convened the repeal campaign kickoff news conference. A story on that news conference [1] reports "Several of the IRV opponents, including Ewing, backed Republican Kurt Wright for mayor this past March. Another Wright supporter, David Hartnett, also collected signatures for the group. "

That article highlights the centrality of Kiss's problems as mayor: <<The question for March 2010, however, is how much of the anti-IRV effort is truly a referendum on IRV, and how much is a referendum on Mayor Kiss. His term isn't up, technically, until 2012, but a spate of leadership crises related to Burlington Telecom, the Moran Plant redevelopment and the loss of General Dynamics practically put Kiss in the lame duck category. And to think it's only the guy's first year of a three-year term.>>

See other stories on how John Ewing backed Wright's campaign in 2009 [ https://www.sevendaysvt.com/vermont/right-as-wright/Content?oid=2136269] and Dave Hartnett was Wright's 2012 campaign spokeperson [2]

Another story after the campaign summarizes things accurately - including howhow the Burlington Telecom scandal revived a stagnant repeal campaign and how the repeal vote broke down, with only the Wright strongholds backing repeal.

https://www.sevendaysvt.com/vermont/irv-repeal-signed-into-law/Content?oid=2178268

...The IRV repeal effort seemed stagnant until news of Burlington Telecom's financial troubles went public in the fall. Then, it seemed to gather steam and by March the anti-Kiss sentiment appeared to crest and the vote in March was seen as both a referendum on the voting system itself, as well as one on Kiss and his administration.

Overall, IRV lost by a 52-48 margin, but it was defeated resoundingly in the New North End Wards 4 and 7, and barely survived in Ward 6. Support was strong in the other four wards, but turnout there was much lower. ere's how the IRV vote broke down by ward:

Ward 1: 405 to keep, 264 to repeal
 Ward 2: 428 to keep, 185 to repeal
 Ward 3: 510 to keep, 292 to repeal
 Ward 4: 1203 to repeal, 606 to keep [ GOP ward' Ward 5: 793 to keep, 545 to repeal Ward 6: 490 to keep, 477 to repeal
 Ward 7: 1006 to repeal, 437 to keep [ GOP ward] RRichie (talk) 02:51, 24 March 2019 (UTC)

As a tag, the main signature gathering was Linda Chagnon, according to this article about the news conference convened by Kurt Wright: "Linda Chagnon of the New North End. Chagnon alone netted roughly 500 signatures."

Chagnon also was very active in the Tea Part the same time. See this article on Tea Party rallies that same spring: ".... In Montpelier, Vt., where 60 people mocked the government in song and word, Linda Chagnon, a 57-year-old caregiver from Burlington, carried a sign that read: "Give Me Liberty, Not Debt."

http://www.nbcnews.com/id/36560551/ns/politics-more_politics/t/tea-party-ends-tour-capital-it-loves-hate/#.XJfBUihKhPY

In addition, the campaign slogan was "Keep It Simple." Silent was the last word "Stupid," but the subtext clearly the "KISS principle" -- all very clever, but underscoring how much of it was anti-Kiss.

RRichie (talk) 17:50, 24 March 2019 (UTC)

User:RRichie is the CEO of FairVote and has a rather significant conflict of interest. Mind you I'm also an elections reform person, and have to take my own care when I'm working on these.

The first of the recent series of edits removes most reference to "controversy" and talks about "focusing attention" instead. It begins directing focus on Kurt Wright as a sore loser, and the string of edits continues this theme.

User:RRichie even states on this talk page: "The IRV repeal effort seemed stagnant until news of Burlington Telecom's financial troubles went public in the fall. Then, it seemed to gather steam and by March the anti-Kiss sentiment appeared to crest and the vote in March was seen as both a referendum on the voting system itself, as well as one on Kiss and his administration." Nevertheless, he leans heavily on Kurt Wright and his campaign team being the first ones to rally against IRV—even if they were apparently unsuccessful.

Even so, the narrative is sufficiently informative, and I've edited it to undo replacing "other voting reform advocates" with "opponents of IRV" (hey, look, score voting is clownshoes and those people have all these fantasies about people scientifically measuring candidates and not voting strategically, but if we're calling them dumb we still have to admit they're honest about their misguided ideas), and added in the terminology of "post-election controversy" because that's an accurate description of what happened.

The most important part of all of this is the technical analysis: Why did IRV fail? Well it failed because of vote splitting. It also failed because of a mutual majority lock-in property—when too many voters in a minority coalition all share in ranking at least one candidate above all mutual majority candidates (Bob Kiss and Andy Montroll, in this case), those voters's ballots (no matter their contents!) have zero impact on the election—but that's new research and pointedly *doesn't* go in the article…yet. What we do know is those voters who voted Wright above Montroll and Montroll above Kiss ultimately didn't count, and the election behaved as if Montroll and Kiss were in the same party and all remaining voters were party voters voting in a party primary: those voters as a closed and associated group selected Kiss, the others selected Wright, and the runoff was between Kiss and Wright.

The rewrite is still polished up to heavily emphasize a narrative of Kurt Wright and his flunkies instrumenting a repeal as sore losers, and downplays that their efforts failed until Bob Kiss was suddenly unpopular. I suspect the Republicans—typically against ranked voting in the current political environment—would have voted for repeal anyway; can't prove anything about that either way. At the end of the day, the repeal was bipartisan. Seeing as I can't figure on how to rewrite it without slanting it the *other* way (which is just as bad) and the historical context is informative, it stands for now, even if it reads as a series of inferences to justify a conspiracy theory and persuade the reader of an association between the public opinion of the system and the actual performance of the system. John Moser (talk) 02:50, 5 July 2019 (UTC)

Participation and Mutual Majority Lock-Out

This counts as too much OR, but someone can have fun with this.

IRV fails the participation criterion in interesting ways. In particular, because dividing a set $S$ into two subsets always results in a subset with no more than ${\frac {|S|}{2}}$ elements, if given the universe of all candidates $U$ and a finite set of all voters $N$, if there is a mutual majority $U'\subset U$, and if majority $M\subset N$ and minority $m=N-M$ such that all voters in $M$ prefer all candidates $U_{M}$ to all other candidates, and all voters in $m$ prefer a set of candidates $U_{m}$ to all other candidates, the result of IRV is independent of the votes cast by $m$ if $|m|>{\frac {|M}{2}}$.

The actual theorem is more-complex than that and expands into a mutual plurality kind of thing. That behavior is *extremely* useful when selecting multiple candidates for proportional representation: even before reducing the ballot power (as per Meek's Method et al), sets of ballots mutually favoring subsets of candidates sort of decide among themselves without input from other voters. As candidates are elected, the voting power of those ballots is diminished so that others have primary power in selecting mutually-representative candidates, but large coalitions retain more voting power so as to get representation appropriate to the coalition size.

When selecting one candidate, as per IRV, this behavior locks out voters, and is able to disenfranchise large coalitions of voters, particularly when a majority party or other majority coalition represents less than ${\frac {2}{3}}$ of all voters. Irrelevant voters who don't rank candidates in $U_{M}\cup U_{m}$ are excluded from the count of all voters in these considerations, which exemplifies the generalization.

The simple explanation for 2009 is vote-splitting between Montroll and Wright. The complete explanation is really that Wright captured more Montroll>Kiss voters than Kiss>Montroll voters, and that Wright's coalition was smaller than the Kiss-Montroll coalition but larger than half the Kiss-Montroll coalition. As such, Wright's coalition had zero impact on the election and may as well have not voted. The election behaved as a party primary between a theoretical party made up of voters who voted all but one of Kiss and Montroll over Wright (yep, the generalization is complex), followed by a general election between the nominees.

Because the state changes at every round and the voters whose ballots impact each subset of candidates changes, the generalization actually functions more like a PRNG than a social choice function: the outcome is a function of the ballots, but it's not the computation of the relationship of the nature of the voters's exposed preferences as a whole to the candidates and the selection of a winner based on that relationship. Still, it follows certain rules and can be predictable and manipulated: the interesting part of Burlington, 2009 isn't that it elected a non-representative candidate, but that the failure mode is so readily analyzed and so easily reproduced by nominating candidates that you can weaponize it and thus a well-resourced political entity can act as a dictator and force the election of a specific candidate.

Condorcet methods generally resist this, although Condorcet-Hare methods (ironically, all based in some way on IRV) both avoid this defect and are highly-resistant to all other strategy.John Moser (talk) 03:26, 5 July 2019 (UTC)

A Commons file used on this page or its Wikidata item has been nominated for deletion

The following Wikimedia Commons file used on this page or its Wikidata item has been nominated for deletion:

• Vermont Progressive Party Logo.svg

Participate in the deletion discussion at the nomination page. —Community Tech bot (talk) 03:06, 20 June 2022 (UTC)

Stub assessment

I removed the "stub" assessment for this article, since it's clearly more mature than that. I'll leave it to someone else to decide what assessment the article deserves. -- RobLa (talk) 02:40, 29 June 2023 (UTC)

Including exhausted/indifferent ballots

@AndyAnderson Typical practice is to include truncated ballots in totals; see 2011 Irish presidential election for an example.

A reversed convention wouldn't make sense, because indifferent or truncated ballots affect the outcome of many elections. Lots of Condorcet methods are affected by the inclusion of truncated ballots. Truncated IRV ballots affect earlier rounds as well. Handling truncated ballots plays a major role in STV elections (multiple-seat IRV), and truncated ballots are always included in quota calculations. Generally, if a ballot can affect the outcome of an election, it has to be included in calculated voting shares. Closed Limelike Curves (talk) 16:06, 20 March 2024 (UTC)

Better imperfect source than no source

@Jannikp97 I think it's sensible to try looking for a better source than rangevoting.org (which I admit has its problems). That being said, could we hold off on removing the citations until we can find better ones? Even if the source is unreliable, it's better to include it so users can verify the reliability themselves—with no citation, it's easy to think the information probably comes from a reliable source that just hasn't been included in the text. The tag is very helpful for these cases. Closed Limelike Curves (talk) 16:35, 2 April 2024 (UTC)

Hi Closed Limelike Curves,

I left one of them in, replaced one of them by a more suitable citation. For the third one, regarding "Analyses suggested Montroll also would have won under most rated voting methods, including score voting, approval voting, majority judgment, or STAR voting.", the citation actually does not contain these analyses, so for this case I think it is better to actually delete it. Might also just be correct to fully get rid of this sentence. Jannikp97 (talk) 06:56, 4 April 2024 (UTC)

That's reasonable, thanks! Closed Limelike Curves (talk) 00:57, 5 April 2024 (UTC)