Massachusetts Reformers Launch Campaign for Statewide IRV

Comments currently closed for Massachusetts Reformers Launch Campaign for Statewide IRV

• Adam says" Instant Runoff Voting ... has proven itself in the many cities and countries in which it is used." In Burlington Vermont in March 2009, Instant Runoff Voting has proven itself to have failed. It elected a candidate when more voters have explicitly agreed that a different candidate was their choice. Then as a consequential result, it has lost the confidence of voters except for the Progressives from whom their minority candidate was elected by use of IRV. Chances are better than even that IRV will be repealed in 2010, and all this would not likely have happened if the True Majority winner (a term coined by Nobel Laureate Eric Maskin, synonymous with "Condorcet winner") had been elected instead. I write computer code for a living (in audio signal processing applications). I've heard that same cliche: "Don't let the perfect be the enemy of the good" as a defense for sloppy code, quickly thrown together, that superficially appears to do the job. At least at first. Sounds pretty good, until an unforeseen failure that the sloppy programmer didn't anticipate. A better cliche: "A job worth doing is worth doing right." IRV is only half-right. And the criticisms of Condorcet or True Majority Rule, are pretty bogus: "The Condorcet winner might not be *anyone's* first choice." Not bloody likely. Far less likely than IRV electing a candidate that is less preferred than another candidate as explicitly indicated by the voters. "Condorcet is less transparent than IRV." Bulls**t. Both require computers to tabulate and computers report the result. The method that Condorcet tabulates (which candidate has voter preference) is far more straight forward than IRV (a Kabuki dance of transferred votes). Nobody will be scratching their head wondering how the Condorcet winner won the election as they are scratching their heads in Burlington VT. The only *legitimate* issue with Condorcet is the very unlikely possibility of a Condorcet cycle (candidate Rock beats candidate Scissors, candidate Scissors beats candidate Paper, candidate Paper beats candidate Rock). Again, highly unlikely with real voters along a simple political spectrum (Nader voters aren't likely to choose Bush over Gore as their 2nd-choice). But this can be anticipated in advance, rules can be put into place in case there is a cycle. (Perhaps, if there is a Condorcet cycle, then we elect using IRV rules, hell why not? It won't happen anyway.) Unlike what your website says: "Ranked Choice Voting, which is also known as Instant Runoff Voting." Ranked Choice is not synonymous with IRV. Condorcet (and other methods, like Borda) are also Ranked Choice. It's a falsehood to imply to voters that the only way they can express their will in a ranked choice system is with IRV. What happened in Burlington might happen in Massachusetts. A minority candidate squeaks out an IRV victory over the explicit preference of the voters, then you MA folks will be having the same argument and you might see a referendum that repeals IRV and along with it, Choice Voting. That would be too bad. Why not anticipate the problem and fix it before it has an opportunity to sew trouble?

  Posted by Robert Bristow-Johnson, 2009-09-25 14:32:31 (4 years ago)

• Oy, let the cynics talk themselves into a corner about the imperfections of voting system X or voting system Y. No voting system is perfect! But don't let the perfect be an enemy of the good. Instant Runoff Voting's time has come. It has proven itself in the many cities and countries in which it is used. It is the only game-changing electoral reform with serious traction, besides, perhaps Clean Elections. Thank you FairVote for helping us promote our campaign here in MA. Readers please visit our official website for more info: http://www.voterchoiceMA.org Come one and all, let's get 100,000 signatures and set a record-breaking precedent that will help evolve democracy in this country like nothing before!

  Posted by Adam Friedman, 2009-09-23 22:02:19 (4 years ago)

• I dunno Jack. I am not sure that non-monotonicity that affects the non-Condorcet winner is such a bad thing. Like the Condorcet cycle, it's hard to imagine it happening in reality. for an example, i do not see the 2009 Burlington mayoral race as being affected by non-monotonicity. At the end of "Round 2", there were 3 standing candidates and the plurality winner, Republican Kurt Wright, was still on top in terms of 1st-pick votes and transferred votes mostly from 4th-place Independent Dan Smith. Now since the IRV final round is just like one of the hypothetical "mano-a-mano" Condorcet race (it doesn't matter 1st-pick, 2nd-pick, etc, it only mattered if a voter ranked Wright ahead of Kiss or if they ranked Kiss ahead of Wright), then we know that the Condorcet winner (in this case Montroll) would beat *anyone* in the IRV final round if he can make it into the final round. Now, if it was that Kiss was ahead of Wright with a plurality of votes in Round 2 then non-monotonicity issues may have been possible if more of these Wright voters had instead voted for Kiss sufficiently to keep Wright out of the final round, Montroll goes into the final round and supposedly beats Kiss (as he would have with the vote totals we have). But if Kiss was ahead of Wright to the point where Wright was close to missing getting into the final round, it's not too likely that Montroll would have been the Condorcet winner in the first place. Kiss might have been. The way it was is that even though Montroll was the Condorcet winner, therefore preferred over either Kiss or Wright in a one-on-one race, there is no way that additional votes for Kiss (and taken away from Wright) would have possibly taken Wright out of the final round without also simply putting Kiss on top. It is clear with these Wright voters that promoting their Kiss vote up past Wright, it would also promote their Kiss vote past Montroll. There were no monotonicity issues in that race, as best as I can tell. There is no scenario where adding votes for or switching votes to Kiss would have caused him to lose the election in IRV. But if the vote totals were different, it's conceivable how that could happen. I really consider the possibility of non-monotonicity to be very very low, as I consider the likelihood of a Condorcet cycle messing things up. But, in 2005, when I voted for IRV at town meeting day, I didn't think it was likely that IRV would elect a different candidate than Condorcet would. And in fact, in 2006, Bob Kiss was both the IRV winner (which is how he became mayor) *and* the Condorcet winner. In 2006, clearly no candidate had majority voter preference over Bob Kiss, no matter how you looked at it. But that is not the case in 2009. That's why there was this big fuss about IRV to get it repealed.

  Posted by Robert Bristow-Johnson, 2009-08-22 01:13:28 (4 years ago)

• The unanswered question is this: why is monotonicity more desirable than other properties of voting systems? Incentives for coalition-building and programmatic campaigning are arguably more important to the health of a democracy.

  Posted by Jack, 2009-08-21 12:17:21 (4 years ago)

• A typically bizarre claim by Rob Richie: "As to ?spoilers,? I suspect everyone in Burlington?s election this year voted their sincere preferences - in that critically important sense, IRV is overwhelmingly likely to be ?spoiler free,? unlikely plurality voting." 1) Rob Richie doesn't know what percentage of voters there cast sincere ballots versus strategic ones. He's just assuming that to be the case. Consider that there was a group of voters in Burlington who preferred Republican to Democrat to Progressive, who would have seen an improvement (getting the Democrat instead of the Progressive) if they had insincerely ranked the Democrat in first place. If they weren't strategic in _this_ election, we can certainly expect a lot of them to be in the next one. But Rob will just keep saying absurd things like, "100% of voters in this election were sincere". 2) Strategic voting is a totally separate problem from spoilers. In this Burlington election, Kurt Wright was a spoiler in that if you had removed him from the ballots, then the Democrat would have won instead of the Progressive. And since the Dem was preferred by a majority of voters to the Progressive _and_ to the Republican, many voters would call that an improvement. The problem with spoilers (which exist when a voting method fails independence of irrelevant alternatives), and non-monotonicity, is that you elect the wrong candidate. When faced with these problems, Rob Richie always confuses the issue by bringing up strategic voting. He'll say something like, "this election wasn't non-monotonic, because the voters didn't strategically exploit the non-monotonicity" or something bizarre to that effect. And I don't think it's ignorance. I think he genuinely understands the concepts, and deliberately confuses people to make IRV look good.

  Posted by Clay Shentrup, 2009-08-20 14:08:17 (4 years ago)

• The debate about theoretically derived advantages and disadvantages of single-winner systems has gone on for years. I have yet to see a major empirical contribution to that debate's resolution. Instead we have a constant rehashing of old arguments. What would be interesting for all partisans and useful for reformers in general is systematic, real-world testing of whether, how, and under what conditions voters and candidates respond to the various good incentives that IRV, ranked pairs, et cetera present.

  Posted by Jack, 2009-08-20 10:09:42 (4 years ago)

• Another correction I missed in proofing: I said: "... it is likely that IRV will be repealed in Burlington in 2009." I meant to say that it is likely that IRV will be repealed in Burlington in 2010. And that's too bad. Condorcet would have elected the candidate not only with the clearest popular support (in that the Condorcet winner is explicitly preferred by the majority of voters over any other specific candidate), it would have elected a centrist that was palatable to the widest subset of voters in Burlington. IRV elected a candidate that pleased the Progressives, was palatable to the Democrats, but was not palatable to the Republicans. And it was the Republican candidate who secured the plurality of first-choice votes. That indicates a subset of the electorate of significant size that is not just disappointed in the election result, but very disappointed since for the great majority of these voters the IRV winner was marked as their least-preferred candidate. The Condorcet winner would have both satisfied the Democrat voters, but would have been palatable to both the Republicans and the Progressives. I doubt if a repeal effort would have gained as much traction against the Choice Voting method with Condorcet as it has against Choice Voting with IRV. And it's too bad that the result may very well be that we lose Choice Voting in Burlington.

  Posted by Robert Bristow-Johnson, 2009-08-19 15:58:08 (4 years ago)

• I failed to finish a sentence: Other than the possibility of a ?cycle? or Condorcet paradox .. there is no election principle where Condorcet receives a "demerit" and where IRV does not. Other than the possibility of the Condorcet cycle, there is no non-partisan reason where Condorcet is inferior to IRV in reflecting majority preference. And there are many reasons why it is better.

  Posted by Robert Bristow-Johnson, 2009-08-19 04:29:03 (4 years ago)

• So, it's considered "too nerdy" that an election principle that embraces the value of "majority rule" be that if a majority of voters agree that Candidate A is a better choice than Candidate B, then Candidate B should not be elected? That's pretty fundamental. And in Burlington Vermont, in 2009, IRV failed that fundamental principle. It's considered "too nerdy" that if any loser is removed from the race, that this should not change who the winner is? That also seems pretty fundamental, and in Burlington Vermont in 2009, IRV failed to abide by that principle too. It's considered "too nerdy" that voters should not be called upon to do "strategic voting"? That voters should not have to consider forsaking their favorite candidate in order to avoid helping to elect their least favorite choice? Supporters of the Republican candidate in Burlington in 2009 found out that their primary support for their candidate has actually aided their least desirable candidate (the Porgressive) win over a more desirable candidate (the Democrat). So assuming IRV survives the refererdum, what will savvy Republicans be thinking in 2012 when they go to the polls to vote in the mayoral election? They may have to consider ditching their candidate to bump up support for the candidate not of their choice, but one they can live with. All three of those principles are cited on this very site as reasons to support IRV. But IRV failed those principles in Burlington Vermont in 2009 and Condorcet would have fulfilled those principles. Both IRV and Condorcet require identical ranked-order ballots. Both require computers to do the tabulation of the vote and the computer reports the results. How is it that Condorcet is less transparent than IRV? IRV requires a little Kabuki dance where votes are transferred from one candidate's pile to another. And the basis for doing so: elimination of the "weakest candidate" is based on the arbitrary value that *only* first-choice votes count while second-choice votes count for nothing. In IRV, second-choice votes are as good as last place in evaluating the popular support a candidate has from the electorate. Who's idea is that? From what consistent and non-arbitrary principle does that come from? Condorcet says simply, if some Candidate A is preferred by the majority of the electorate over Candidate B, Candidate C, Candidate D, and all other candidates, then Candidate A is elected. There is no other candidate that the majority of voters would prefer to see take office than that Candidate A. The process of counting these votes is straight-forward and transparent and much simpler conceptually than what IRV does. Other than the possibility of a "cycle" or Condorcet paradox (which is a helluva lot less likely than the possibility of the IRV winner not being the Condocet winner, which means that the candidate who won was explicitly rejected by the majority of voters in favor of another candidate). Because of voter alignment (Nader voters would not likely choose Bush over Gore in 2000), a Condorcet paradox is not likely at all. But even so, similar rules to eliminate the weakest candidate (and re-run another round) can be adopted to deal with the case of a Condorcet paradox. It would be no worse than the rules used in IRV to eliminate candidates. So, let's see, both IRV and Condorcet require ranked-order ballots and computers to tabulate votes. But, by definition, Condorcet will not elect a candidate when it is clear that the majority of the electorate preferred someone else. IRV cannot make that claim and, indeed, failed to do that in Burlington in 2009. That's so basic, so fundamental: Let's not elect a candidate when it is unambiguously clear that the majority of voters have expressed on their ballots that they prefer someone else. How can there be any question as to which reflects the popular sentiment. In this way, IRV is much like the Electoral College. Most of the time the Electoral College "has no effect". The electoral vote selects the same candidate that the popular vote does. The Electoral College does pretty good, when it elects the same candidate that the popular vote does. But when the electoral vote elects a different candidate than is indicated by the popular vote, then this *never* brings more legitimacy and confidence in the election. It only makes it worse, the popular majority has good reason to believe that the convoluted electoral math has stolen the election from them. Then it naturally leads to the question: What good is the Electoral College? If it agrees with the popular vote, it does no harm, but there is never a time that it disagrees with the popular vote that a majority of people say "Whew! Boy, am I glad we got this Electoral College to make sure that these elections are properly decided." It's only the minority who say that. Those who *should* have been the losers, because they were the minority, are the only ones thanking the Electoral College. The same comparison can be made with IRV and Condorcet. Only if the IRV winner is the same as the Condorcet winner that people can look at the results and say that the winning candidate unambiguously has popular support. IRV does pretty good when it selects the same winner that Condorcet does. But when it does not, voters are scratching their heads wondering how this candidate was elected when more of these voters had explicitly expressed that they preferred someone else. Then, for the same reason, the question must be asked how is it that IRV *ever* reflects the popular sentiment than Condorcet when, by definition, the Condorcet winner is *never* less preferred than any other single candidate? Just like the U.S. popular vote is the "gold standard" by which we judge how well the Electoral College is doing, in multi-candidate elections, the Condorcet criterion is the gold standard by which we judge how well other methods have worked, whether those other methods are plurality, IRV, Borda count, approval, Range voting, Coombs', etc. By electing the popular choice, we are trying to minimize disappointment in the election results. Otherwise, why not just give it to the minority? Let the loser be the winner! In Burlington Vermont in 2009, IRV failed to minimize the measure of disappointment in the mayoral election, whereas it is clear that Condorcet would have chosen a centrist candidate that had more popular appeal. Both the Progs and the Dicks could have felt "okay" with the Condocet winner, but the Dicks certainly did not feel "okay" with the IRV winner. And enough Democrats feel the same way that it is likely that IRV will be repealed in Burlington in 2009. Too bad. It could have been avoided if it was Condorcet instead. Because there would have been far fewer really disappointed voters. FairVote.org seems so invested in IRV (as opposed to other choice voting methods) that they seem unable to back out, and when IRV repeats this failure and loses support as a result, I wonder how long it will take Rob and company that maybe they should have promoted other methods, at least along with IRV, instead of just pushing IRV. That was and is a mistake, assuming the goals that FairVote espouses.

  Posted by Robert Bristow-Johnson, 2009-08-19 04:20:04 (4 years ago)

• It is not at all clear to me that "something better" than IRV is available "for the taking", if by "available" Robert mean a voting rule that voters and opinion leaders can understand and accept, and that is transparent enough to meet legitimate concerns about election integrity. Whatever the conceptual merits and demerits of Condorcet-compliant voting rules (they have both), the justification for them is just too nerdy for them to ever be adopted for public elections, and the good ones are too complicated for election officials to count in a way the public can trust. Condorcet methods can be used effectively in some professional organizations, but that's about it.

  Posted by Bob Richard, 2009-08-17 20:03:48 (4 years ago)

• It's at the very core of the discussion to decide who *was* in first place, second place, or third place. It's a stretch to say that Andy Montroll was "in third place" (by what standard?) when, in fact, the majority of voters have expressed that they preferred Montroll over either Kiss or Wright. The same voters expressed, by majority, that they preferred Kiss over Wright. In fact Wright (the plurality holder) was third, Kiss (the IRV winner) was second, and Montroll came in *first*. But then, even by your defintion, then you must concede that candidate you claim "should" win (and did, in fact, under IRV rules) came in second. Does it not create "political difficulty of suggesting a candidate who is in [second] place in a three-person race 'should' win"? What consistent rule or ethic do you appeal to for why the candidate who came in second (in terms of 1st-choice votes) should win the election? Why should that rule apply only to the comparison of Kiss to Wright? Why shouldn't the same rule apply to the the side-by-side comparison of Kiss to Montroll? Or Wright against Montroll? What consistent reason can you appeal to that supports that majority rule of preference should only apply to one pair of candidates but not the other? How can that be considered fair when this rule applies to one set of candidates and not another? Re: "spoilers", I agree with your suspicion that everyone in Burlington voted their sincere preference. But what about the next election? In 2006, Bob Kiss not only won the IRV, but he was also the Condorcet winner. In 2006, there was no single candidate that could claim they had majority support over Bob Kiss. His election in 2006 reflected the popular sentiment. But that was not the case in 2009. In 2009, the great majority of voters that supported Progressive Bob Kiss as their first choice, marked Democrat Andy Montroll as their second (or at least ahead of Republican Kurt Wright). The majority of voters that supported Montroll as their first choice marked Kiss as their second choice (and it is these transferred votes that Kiss can thank for being elected over Wright, otherwise he would have lost). What about the voters that supported Wright as their first choice? Who do you think their second choice was? Many didn't mark a second choice (they are the strident IRV opponents now). Many marked independent Dan Smith as their second choice. Many marked Montroll as their second choice, but almost none marked Bob Kiss as any higher than last place (discounting the fifth and very weak candidate, Green Party Simpson). So now, these Republican Prog-haters have to be thinking if they regret supporting their first choice, Kurt Wright, as their first choice, because the result of that support of their favorite candidate is that their least favorite candidate won. If IRV survives (I, too, do not want to see it reverted to 40%+ plurality rule) what might these voters be thinking as they enter the polls in 2012? "If I vote for my favorite candidate, I will be helping to elect my least favorite candidate." That is precisely what we were trying to avoid with ranked-order ballots and IRV *failed* to avoid that in Burlington in 2009. Because this failure is simply being ignored by the Progs and other IRV supporters, they may miss that it's not only Republicans that are signing the anti-IRV petition, but also many Democrats. Because of the already expressed positions of many Democrat city counselors, we can safely say that the IRV referendum will appear on the 2010 ballot on Town Meeting Day. And it may very well succeed and we'll be stuck with with 40%+ plurality rule (and non-instant runoff if 40% is not attained by any candidate). If, instead, Condorcet was the law, the reaction against it would be *less* than it is with IRV. Progs would have had more satisfaction than if the Republican plurality winner had won. Democrats would have been happy with te results, and Republicans would have been more satisfied with a centrist Democrat than the more fringe (from their perspective) Progressive. The Democrats (and others, including the Burlington Free Press) are wondering how their centrist and popular candidate (who *did* get a lot of 1st-choice votes, but whose real popular support is reflected when voters compared him to any other specific candidate) *appeared* to do so poorly, when in fact, he really *didn't* do poorly in this multi-candidate race. With IRV or plurality it would "appear" so. But there would be much less dissatisfaction with the election outcome if Condocet was the law. And minimizing the dissatisfation with the outcome is the *whole* *point* of elections. Otherwise, why don't we just give it to the minority candidate? Rob, you should ask Terry about this, but certainly the IRV referendum will appear on the ballot in 2010. And, whether it deserves it or not (IRV *is* better than plurality rule, but like the health care debate, there is a lot of vitriolic disinformation flying about here in Burlington), I believe there are better than even chances that IRV will be defeated. Too bad it will be replaced with something worse rather than something better. Especially when that something better is available, just for the taking.

  Posted by Robert Bristow-Johnson, 2009-08-15 13:43:03 (4 years ago)

• Note that Paul is sharing a press release done by Masaschusetts reformers, not FairVote. In addition, Condorcet voting certainly is worth considering, but presents its own issues. Besides the political difficulty of suggesting a candidate who is in third place in a three-person race "should" win, some believe it would create incentives for candidates simply trying to avoid offending anyone -- hardly the basis of executive leadership. As to "spoilers," I suspect everyone in Burlington's election this year voted their sincere preferences - in that criticaly important sense, IRV is overwhelmingly likely to be "spoiler free," unlikely plurality voting. We also believe that IRV in Burlington upheld the principle of majority rule. To be sure, one can never force the majority to __like__ the winning candidate, but you can prevent election of the candidate who is __least__ liked, unlike with plurality voting.

  Posted by Rob Richie, 2009-08-13 14:31:41 (4 years ago)

• Why does FairVote continue to repeat two canards? That IRV: 1. "ensures that the winning candidate is the one with majority support..." 2. "eliminates the spoiler effect..." It *helps* to elect the candidate with majority support but does not "ensure" that goal, and it *mitigates* against the spoiler effect but does not "eliminate" it. Please stop making such an absolute claim when the facts do not support it. Again, the case study is this year's mayoral election in Burlington Vermont that was decided using IRV. There were four solid candidates: Progressive Bob Kiss (the incumbent mayor) Democrat Andy Montroll Republican Kurt Wright Independent Dan Smith As a matter of public record, we know that on March 3rd 2009, of the 8976 ballots that were properly marked, 7541 (84%) expressed an opinion of the merits of Bob Kiss vs. Andy Montroll for mayor. The majority (4064, nearly 54%) clearly expressed a preference for Montroll, yet the mayor was re-elected with IRV. Andy Montroll prevailed in voter preference in a head-to-head comparison to any other candidate, whether that be Kurt Wright or Dan Smith or Bob Kiss. In regard to the spoiler effect, the presence of Kurt Wright in the race should not affect the relative merit, in terms of the majority of Burlington voters, of candidates Bob Kiss and Andy Montroll. If the majority of voters think that Montroll was preferable to Kiss in a ?mano-a-mano? contest, the outcome of that contest should not be affected if Kurt Wright is a candidate or not. But what happened? In effect, with IRV rules, Montroll was identified as the biggest loser in the second round when he was actually preferred over either Kiss or Wright when compared directly to either candidate. Montroll was eliminated prematurely and would have beaten any candidate, including Kiss, in the IRV final round had he been in the final round. If Wright had not been in the race, Montroll would have been elected over Kiss in the IRV final round if the same 8976 voters came to the polls March 3rd and expressed the same preferences that they had on March 3rd. That is a matter of public record. (file: 2009 Burlington Mayor Final Piles Report.txt in http://www.burlingtonvotes.org/20090303/2009%20Mayor%20Reports.zip) When IRV came up to vote in 2005, I voted for it knowing about this potential problem with it, but not expecting that IRV would actually elect a candidate when a majority of voters actually expressed a preference for a different *specific* candidate. I was mistaken and quite surprized in 2009 when IRV failed to elect the majority winner. It seems that FairVote has completely committed IRV when a better "choice" method has always been around. This better method, named after 18th century political philosopher and mathematician Condorcet, chooses the candidate that is preferred by the majority, in a one on one race, against any other single candidate. It uses the same ranked ballot that IRV uses, but tabulates the votes differently and more directly in reflecting that popular sentiment of the voters. Perhaps Massachusetts election reformers should look into plugging Condorcet rather than just jumping on the IRV bandwagon. I can send anyone an analysis of the 2009 Burlington mayoral election and description of Condorcet vs. IRV in a pdf document. E-mail me at rbj@audioimagination.com and I'll send it you.

  Posted by Robert Bristow-Johnson, 2009-08-12 13:40:55 (4 years ago)