Thursday
04Jun2009
IRV -- Non-monotonicity Examined in Council Race
June 4 The worksheet linked below demonstrates the sensitivity of the outcome to therankings of fairly smallrange of voters (70-79 voters)and their relative ranking of Jack and Michael, and three differient resulting winners.
The principle being demonstrated is that sometimes giving a higher ranking to your favorite candidate may have a negative effect on his chances of winning. (see the case of the 71 Michael voters.)
http://theredant.squarespace.com/storage/Aspen%20city%20council%20non-monotonicity_b.pdf
And the background story:
http://theredant.squarespace.com/storage/MichaelMonotoneViolation..pdf
Take the post-election survey:
http://survey.constantcontact.com/survey/a07e2imsis1fumvwxa7/a0121fuunhl0z/greeting
View Printer Friendly Version
Email Article to Friend
Reader Comments (28)
This happened in the recent Burlington Vermont election for mayor.
Voters hurt their preferred candidate by voting for him.
From a report on the Burlington election:
Burlington Vermont 2009 IRV mayor election
Thwarted-majority, non-monotonicity & other failures (oops)
By Anthony Gierzynski, Wes Hamilton, & Warren D. Smith,
March 2009. (skip to summary)
this election also featured nonmonotonicity.....In other words, Kiss won, but if 753 Wright-voters had switched their vote to Kiss, that would have made Kiss lose!
...pretty much every voting method mankind ever invented would elect MONTROLL
– making this a pretty easy election to call – except that IRV elects KISS and plurality elects WRIGHT.
http://rangevoting.org/Burlington.html
Joyce McCloy writes that "pretty much every voting mankind ever invited would elect Montroll."
Come again? Montroll was in last place when the Burlington was reduced to three candidates. There in fact is no election method used in any public election in the world or any office that would have elected him.
On the other hand, systems that McCloy apparently supports like approval voting would allow a candidate to lose even if 55% of voters supported that candidate as their first choice.
As to non-monotonicity, IRV is no different than the two-round system Aspen used to have - except that the old system was easier to game. See a good discussion of this reality in the Minneapolis Star Tribune by Professor Jack Nagel of U-Pennsylvania:
http://www.startribune.com/opinion/commentary/47071097.html?elr=KArksc8P:Pc:U0ckkD:aEyKUiD3aPc:_Yyc:aULPQL7PQLanchO7DiUr
Correcting post above:
Rob Richie mistakenly said: "Joyce McCloy writes that 'pretty much every voting mankind ever invited would elect Montroll.'"
No, Rob. That wasn't my statement, it was a statement on a report by the following election method experts:
Anthony Gierzynski, Wes Hamilton, & Warren D. Smith
(Prof Gierzynski headed the Vermont Legislature Research Workshop that analyzed the I2009 Burlington RV election). http://www.uvm.edu/~vlrs/IRV.html
Rob Richie mentions "non monotonicity" but immediately distracts the reader away from that dangerous topic to a discussion on "gaming the system" Richie can't deny that IRV is non monotonic - that IRV causes voters to hurt their preferred candidates so he changes the subject.
READERS: non monotonicity means -that you can cause your favorite candidate to lose by voting FOR him/her. it has nothing to do with gaming the system. IRV can cause a winning candidate to lose by ranking him higher, or cause a losing candidate to win by ranking him lower.
Aspen voters who wanted to elect Michael Behrendt actually caused him to lose by ranking him as their 1st choice. - That is what "non monotonicity" is.
Michael Behrendt, Council candidate, got defeated by 75 of his own supporters, doing their best to support him by ranking him #1 on their ballot.
BIn the 2009 Burlington Vermont mayoral election, Bob Kiss won.
But if 753 Kurt Wright's voters had voted for Kiss (instead of Wright) Kiss would have lost! See the analysis by Gierzynski, Hamilton, & Smith - "non monotonicity"
http://rangevoting.org/Burlington.html#summary
IRV is a well intended election method, but it truly does not meet expectations.
It harms election transparency because it is not additive - it cannot be tallied at the polling places like regular ballots can. It utilizes a sorting and reshuffling of the ballots - often with bizarre results.
If the goal is to enfranchise voters, then the election process must be as simple and transparent as possible.
IF the goal is to empower third parties, then there are several election methods that can be utilized that do not have the perversities that IRV has.
IRV. IRV actually hurts third parties - unless - they are already strong in the jurisdictions they exist in.
Other options include approval voting or even Fusion Voting for example. Fusion Voting has proven to strengthen third parties in jurisdictions that used it, and it is transparent, it can be tallied at the polling places, and does not require a change in voting software or machines.
more info on IRV problems at www.instantrunoffvoting.us
Hi, Joyce.
Any system that eliminates candidates between rounds or counting is non-monotonic. Aspen's previous system was as well. But unlike IRV, the non-monotonic quality of the old runoff system had tactical significance.
I appreciate you saying that IRV advocates can be "well-intended," as sometimes your posts suggest otherwise IRV advocates think your concerns can be addressed, but agree that it's good to debate a full range of voting methods.
It would help me if you could clarify whether you support approval voting and range voting over current voting methods. Both of those systems can result in a candidate losing who is the first choice of a super-majority of voters. Are you recommending one of those systems for Aspen or anywhere else? You might be interested in my blogpost on approval voting today at:
http://www.fairvote.org/blog/2009/06/voting-method-debates-go-way-back-electing-the-pope/
I''m sure the dialogue will continue, but it would help me understand your concerns more clearly if I knew what method(s) you were for as opposed to what you are against. Thanks.
Readers. Do not let anyone distract you from the issue at hand.
The issue is NOT what election method I or other prefer.
The issue here is whether IRV can cause some voters to hurt their preferred candidate by ranking him 1st.
The answer is YES.
It happened in this Aspen IRV election, and it happened in the recent Burlington Vermont election.
Would this have happened with Aspen's traditional election method?
NO.
Don't be fooled.
The issue is - do you want an election method - IRV - that can cause you to hurt your preferred candidate?
It's true: with score voting, even if more than 50% of voters prefer you to another candidate, you can still lose to that candidate.
But for that to happen, those voters would have to EXPLICITLY state their willingness to accept your challenger as a good compromise.
Score voting, put simply, gives the win to the candidate with the highest average rating among all voters. And anyone who made it through high school arithmetic can construct a situation where more than half the voters give a higher score to A than to B, but where the average score of B is higher than the average score of A.
For instance: 3 voters, two give A 9 points and B 5, one gives B 9 points and A 0. A is prefered 2-to-1, but A's average score is 6, while B's is 6.33. The crucial point though is that the voters get to make the choice about who is an acceptable compromise and who isn't. The A-prefering voters would only ever vote 5 points for B if they honestly believed B to be a fair compromise (which they would only find necessary to do if there were more than two candidates; obviously I've left these extra candidates out of this example for simplicity, but you can imagine the A-first voters giving a candidate C a zero, and the B-first voter giving them a 9, i.e., as high a score as they gave to B.)
Now, if your second-choice wins because you EXPLICITLY stated a willingness to accept them as a compromise candidate, you will find little to complain about; it's not as if you could have given MORE than the maximum points to your first favorite, and accepting the compromise may very well have kept your LEAST favorite candidate from winning.
Whereas under IRV, you can be shocked to discover that your first choice would have won, and your last-choice would have lost, if only you had voted COUNTER TO your true intentions by ranking your first-choice lower than first-place on your ballot!
I'm honestly surprised that anyone could take IRVs negative and completely non-intuitive results which arise because of its non-monotnoic behavior, and try to favorably compare it versus score voting's positive ability to allow explicit compromises when the voters so choose.
Why Burlington was a disaster is clear when you see the actual results. Rob Ritchie rejects non-monotonicity as an argument, but it is clearly important that changing your vote to favor a candidate can make him lose, when without your vote he would WIN.
In addition, he points out that Montroll came in third in Burlington, but what he did not mention was that Montroll was the clear second choice of everyone whose first choice was someone else. IRV elects EXTREMISTS and makes it impossible for compromise candidates to win.
Wright's voters would want Montroll rather than Kiss.
Kiss's voters would want Montroll rather than Wright.
Because Montroll was eliminated early, Wright's voters' second preferences were not counted.
IRV works great in Australia, where there are two major parties and the third-largest is a satellite of one of the top two. It will not work in a Burlington-like situation, where there are three fairly strong candidates.
When will Ritchie confess to this?
My response IRV/STV proponent Jack Nagel's OpEd:
Accounting of IRV's flaws had its own There's no perfect voting system, but instant-runoff foes avoid comparisons.
http://www.startribune.com/opinion/commentary/47071097.html?elr=KArksc8P:Pc:U0ckkD:aEyKUiD3aPc:_Yyc:aULPQL7PQLanchO7DiUr
By JACK NAGEL
Last update: June 5, 2009 - 6:10 PM
NAGEL: Instant-runoff voting is attracting growing support both in the United States and abroad, but Andy Cilek (Opinion Exchange, May 29) contends that the system has fatal flaws. The Minnesota Supreme Court is now weighing such arguments as it considers whether to uphold a Fourth District Court decision ruling that IRV is constitutional. A Supreme Court decision in favor of IRV will clear the way for the voting method (known officially as ranked-choice voting) to be used in Minneapolis elections, in accordance with the desire of nearly two-thirds of voters in the 2006 referendum.
Some of the flaws that IRV's critics allege are simply incorrect. For example, Cilek claims that IRV would give some citizens more votes than others. Back in 1915, the state Supreme Court used this reason to reject a different method of ranked-choice voting called the Bucklin system. That analogy is misleading. Under Bucklin, electors could have votes counting for more than one active candidate at the same time.
MY RESPONSE: The Bucklin method, unlike IRV, is a fair voting method because every voter has all of his/her choices counted equally. Nagel neglects to mention that in IRV only some voters have their 2nd and 3rd choices counted, even when their 1st choice loses the election, so that IRV allows some voters to vote for more candidates and have more influence on the election than other voters.
NAGEL: If no candidate received a majority of first preferences, then both the first and second choices of those voters who expressed second preferences were added to candidates' totals. In contrast, IRV counts second preferences only after the voter's favorite has been eliminated. Thus IRV allows only one vote per voter at every stage of counting.
MY RESPONSE: Nagel neglects to be precise. He should state that "IRV allows AT MOST one vote per voter at every stage." In IRV some voters have LESS than one vote at each stage, some voters have ZERO votes at each stage and some voters never have their 2nd and 3rd choices counted, even though their 1st choice eventually loses the election.
Also Nagel neglects to mention that in IRV/STV some voters have votes for more than one candidate counted, and other voters do not (even though both groups of voters may have voted for more than one
candidate.)
NAGEL: There is no perfect voting system.
MY RESPONSE: Precisely what Nagel should say to convey the meaning of this argument is "There is no perfect voting system, SO IT IS OK TO USE ONE OF THE WORST KNOWN VOTING METHODS - IRV".
Nagel's claim that "There is no perfect voting system" is based on Arrow's theorem that only applies to rank choice ballot voting methods
- not *all* voting methods.
NAGEL: Every method has practical advantages and disadvantages, and every method is subject to one or more logical imperfections. The only sensible way to decide about an electoral reform is to compare its defects -- and strengths -- with those of realistic alternatives.
MY RESPONSE: The above statement is a good idea because there are many fair, simpler voting system without all the flaws of IRV/STV including Condorcet methods, approval voting, range voting, top-two runoff and many others, that are much better choices than IRV/STV.
NAGEL: IRV's opponents usually avoid such comparisons, probably because they are an alliance of strange bedfellows who could not agree among themselves on what voting method they would advocate instead of IRV.
MY RESPONSE: Precisely what Nagel should say to convey the meaning of this argument is "because they are an alliance of strange bedfellows who could not agree among themselves on what voting method they should advocate instead of IRV, THEN WE SHOULD USE ONE OF THE WORSE KNOWN VOTING METHODS - IRV."
Submitted by Kathy Dopp by email.
Joyce -- you are wrong when you say that in non-monotonic results were not possible in Aspen's former runoff system. Please read Jack Nagel's commentary in the Minneapolis Star Tribune so that you understand this issue better.
Bruce -- I respectfully disagree that the Condorcet winner should always win and do not believe that Burlington was a "disaster" because of fhe defeat of the candidate in last place when the field was reduced to three. To believe Montroll should have won in Burlington, you also have to believe that someone should win who is everyone's compromise choice in a race that is something like 46%-45%-9%.
IIf you had such a result in Burlington, the system would be repealed.... instantly. We're facing a potential repeal effort of IRV there primarily because supporters of the plurality leader in the first round are unhappy with the result. Imagine what would happen if you had backers of two candidates with collectively more than 90% of first choices feeling aggrieved.
I also hope you step back from this debate enough to consider what it means to call the Burlington results a "disaster." I would assume you would agree that it would have more of a disaster for the plurality leader in the first round to win, as the candidate in second had more overall support. But that's the typical system in the USA that we're seeking to reform. IRV in faxt gave a better chance for Montroll to win than a traditional runoff as well. So IRV is better than every other system used in any U.S. election for what you seek, but you're calling it a "disaster" because it didn't elect someone who in fact wouldn't have won under any method used for any governmental election anywhere in the world.
Perhaps you could give IRV backers a bit off a break and believe that there are alternate points of view on this about the goals of a voting method? And... as I mention in my blogpost today on approval voting, I would hope that you would try harder to try to win your preferred reforms to see if they can't be won. If they in fact can't be won, by spending so much of your time criticizing IRV, you're in fact just defending the status quo.
I suspect that's what Joyce McCloy wants (Joyce, correct me If I'm wrong), but is that what you and Clay want?
Rather than Rob Richie try to use red herrings, straw men or other diversions, lets just talk brass tacks.
IRV is like playing Russian Roulette with your vote.
By ranking your most favorite candidate first, you may cause him to lose.
IRV has that perversity - you just don't know if you are hurting or helping your preferred candidate by voting for him.
Burlington and Aspen are two examples of how this played out, voters played Russian Roulette and lost, all in 2009.
Rob Richie's misdiretion to gaming in Nagel's op-ed is lame. Sure, Lim-butt told Republicans to cross over and vote for the lesser Democratic candidate in the 2008 Presidential primary elections. That was because McCain had the race sewn up early, and the Dems were still battling it out. Lim-butt could afford to "game" the election by having some Republicans cross over and vote for Dems. So it's really an apples and oranges.
The Presidential primary election takes place over several months. I have always objected to holding those primary elections over several months because the race is almost always all but over well before my state (NC) gets to vote in May.
If there was a one-day Presidential primary election nationwide, there could be no gaming of the election. Since the race wouldn't be determined in advance in some states, no voter could afford to take a chance and vote for anyone other than their first choice.
With a one-day national primary (in the spring of the year so people aren't freezing their asses off campaigning in the dead of winter in Iowa and NH) everyone would have the same info on the candidates at the same time instead of stuff coming out about one candidate not being able to keep his pants up, or one candidate not getting enough votes by the first Super Tuesday, so they have to drop out.
And we pretty much already have enough information about the candidates to do a one-day national Presidential primary election NOW! And doing so would be a whole lot easier than doing IRV. Can you imagine how much of a national nightmare an IRV presidential primary would be - with state-wide central counting? Imagine how much of a nightmare it would be with nationwide central counting?
Because there is no reason to even have a IRV presidential election after a separate IRV presidential primary. I am sure RR and the FairyTaleVote people really want us to do away with primaries altogether - and just have one big IRV electoral love-fest in November and sing "Kumbaya" instead of "Hail to the Chief"!
This is a healthy debate, and without being judgmental, I respect everyone's input.
I tend to be more practical than theoretical. So the question I have is, what is the practical impact of all this discussion on the IRV voting method Aspen just used with nine candidates for CC? Here's what I mean.
Weren't the basic ground rules to rank the candidates for office in order of preference? Now I'm going to restate that basic instruction for what I believe they should be given the practical realities of voter behavior - Rank your favorite candidates (#1, #2 and maybe #3) and rank your least favorite candidates (#9, #8 and maybe #7). For candidates you are indifferent about (in other words, you don't have a strong opinion or you would be just as happy no matter which prevails) rank them from #3 to #7 the best you can.
The important point here is that the vast majority of voters are not going to immerse themselves in the election process and come up with clear reasons why they ranked one person #6 and another person #5. They will know why they ranked their favorite candidate #1 and their least favorite candidate #9, I feel.
Once voters properly fill out their ballot following this instruction, isn't it true that the results will be a proper reflection of how the voters voted, using the method of tabulation that Aspen selected?
I fully appreciate that after the fact an analysis can be done that shows various "permutations" of the results - along the lines that MM has done with the Michael Behrendt results. But that isn't relevant, in my opinion, for what should be obvious reasons, not the least of which is that these various permutations cannot be predicted beforehand. They are nothing more than the mathematical result of voter ranking actions on that particular day with that particular group of voters.
Consider this. If the same nine candidates ran again one month later, with more information available to voters and a different set of voters - more or less just because of the passage of time - the permutation that has been identified for Michael would probably not recur. There would be other post analysis permutations identified, but again, it doesn't matter, does it? The permutations, like Michael's, don't matter when it is quite clear they can't be predicted beforehand. And there is nothing about the Michael permutation, as far as I can see, that invalidates the results of the Aspen CC election we just had.
What am I missing? No need for anyone to restate points they've already made and please focus on the practicality of the Aspen vote that just happened. For what it's worth, I think bringing this conversation back to addressing the practicalities of what just happened here would help many readers.
Thanks
Mr. Louras has a good question. The practical impact is that Aspen can expect these anomalies to occur because it seems that Aspen, unlike many less fortunate cities, draws many candidates into the race, and the larger the candidate pool, the higher the likelihood of such anomalies. It undermines voter confidence when voters are afraid to fully rank their ballot or second-guess whether they should vote “sincerely.” Perhaps some voters suspected or knew this and it explains why they bullet voted to much in Aspen, and many missed participating in the runoff if their candidate did not make the runoff. It appears from Marks’ data that the drop off in participation in the runoff vote was greater than in traditional runoffs. So those are two practical effects.
Another practical effect is the confusion that this method creates. If the voters of Aspen keep the system in place, it is likely that candidates in the future will hire mathematicians to help them achieve the greatest chance of success, although anomalies such as Mr. Behrendt’s can’t be predicted. That would be a sad state. The more complexity and confusion, the less practical transparency there is.
Practical effect 4.-- One of the reasons the method is in the Supreme Court in MN is that there are serious questions as to the legality of IRV, founded in U.S. Constitutional principles of Equal Protection. The non-monotonic risk in IRV is one of the key issues being argued. Colorado State law appears to be more specific than MN and prohibits a method where lower rankings count against higher rankings. Therefore, any of the Aspen mayoral or town council candidates or voters could have likely tied up the election in courts for months if they had chosen to do so. Aspen is lucky that no one chose that route. The next batch of candidates or supporters may not be so passive about their rights.
Practical effect 5—the effect that this analysis demonstrates was for a narrow range of voters, (in the 70’s). Far more perverse results are certainly possible, which could find many voters very unhappy with outcomes that did not make sense. If you read the mathematician’s referenced papers, and this one in particular, http://rangevoting.org/IrvPathologySurvey.html, as well as Kathy Dopp’s paper http://electionmathematics.org/ucvAnalysis/US/RCV-IRV/InstantRunoffVotingFlaws.pdf the practical problems you ask about come to life.
The practical effect that most non-mathematicians might present is the fact that IRV does not allow voters, particularly with big fields of candidates, the opportunity to hear candidates’ positions debated thoroughly. An instant run-off can never simulate a real runoff and all that is learned in a more focused campaign. There is nothing more practical than that for democracy.
Pete
I don't think anyone is talking about invalidating any Aspen results. It sounds like plenty of people could have, but simply didn't. My thought is that this discussion/debate over IRV is centered on whether to have a responsible (read: fair) system used, when the results are so easily challenged, especially in terms of voter intent, such as in the Michael B case. To distract readers by discussing allegations/suspicions of invalidating the election results only detracts from the greater and more important FACTUAL debate of whether these debates about IRV should have occurred BEFORE the IRV question was put on the Nov 2007 ballot.
pete lurous,
you could read
http://www.mnvoters.org/IRV.htm
to help answer your question on practical problems.
the authors said it well:
"The (STV) IRV method has a 3rd fatal flaw: In elections with more than two candidates, voters are placed in the precarious position of not knowing whether they are helping or hurting the cause of their favored candidate. (see NONMONOTONICITY Paradox)http://www.mnvoters.org/nonmon.htm
IRV is NOT as easy as 1-2-3 as proponents suggest. Voters shouldn't need a calculator to figure out if they are helping or hurting their cause by how they rank candidates. Instead of IRV, this method of voting should be called SITD for “Shot in the Dark” or PIAP for “Pig in a Poke.” This dwarfs the hanging chad problem in Florida."
No voter wants to harm their favored choice. Telling him that the permutatoins are so complex that they can't figure it out, is not going to help make the voter feel more comfortable. That is a practical problem no matter which way you look at it.
quoting Louras:
>The important point here is that the vast majority of voters are not going to immerse themselves in the election process and come up with clear reasons why they ranked one person #6 and another person #5. They will know why they ranked their favorite candidate #1 and their least favorite candidate #9, I feel.
Once voters properly fill out their ballot following this instruction, isn't it true that the results will be a proper reflection of how the voters voted, using the method of tabulation that Aspen selected?
PeteLouras.
--REPLY BY W.D.Smith: not necessarily.
Just because a voting method solicits rank-order ballots, that does not imply the election results will "be a proper reflection of how the voters voted."
For example, consider this voting method:
1. everybody submits a rank-order ballot.
2. they all are ignored and Hitler is elected.
QED.
Does IRV (instant runoff voting) improve over this? Yes. But for example,
in a situation where IRV elects X, but a big majority of the voters preferred Y over X, and indeed a big majority of voters prefer Y over EVERY rival, I think it is legitimate to question whether IRV's election result ("X wins and Y doesn't") is a "proper reflection of how the voters voted."
That is exactly what happened in Burlington Vermont in its 2009 mayoral election, see http://rangevoting.org/Burlington.html
where Montroll defeated every rival by majority, but IRV elected Kiss.
For another example,
suppose IRV elects X, but we happen to know that if MORE voters had
voted X top, that would have made X LOSE. That also can happen with
IRV (it's one kind
of "non-monotonicity"). And indeed non-monotonicity happened in both
Burlington and in Aspen's council election. Whenever this happens,
again it often is quite questionable
whether IRV provided "a proper reflection of how the voters voted." I mean, was X really the right winner? And after the change makes X lose, was X really not the right winner?
It's impossible to believe both. In at least one of these two situations, we know IRV failed to do the right thing.
There are many other weird illogical pathologies IRV can suffer, see http://rangevoting.org/IrvPathologySurvey.html
So, what can we conclude from all this?
Does IRV always provide "a proper reflection of how the voters voted"?
I would answer: definitely not.
Does that by itself mean IRV should be rejected? Not necessarily.
Plain plurality
voting also has problems, for example. For example "vote splitting."
Con: IRV exhibits more kinds of weird
illogic problems than plurality, and it is more complicated (which incidentally is probably related to why all of Aspen's rank-ballot elections were incorrectly counted!), and it has to be counted centrally and cannot be counted in precincts, which is a problem in terms of fraud and security risks.
Pro: IRV solves plurality's "spoiler" problem in situations where there are only 2 strong candidates (but not if there are 3 or more), and tends to produce better results on average in computer simulations using honest voters. (But using "strategic" voters they perform the same.)
What I personally advocate is RANGE VOTING and perhaps its simplified
(but not as good) form APPROVAL VOTING. See
http://rangevoting.org
and
http://rangevoting.org/Approval.html
This as far as I can tell is
better than both IRV and plain-plurality, e.g:
* approval is simpler than both. It delivers better results on average than plurality in computer simulations, and comparable results to IRV. It can be counted in precincts and using ordinary "dumb"
voting machines without special software. It experimentally has the
smallest voter-error rates of any voting system.
* range voting delivers better results on average than either approval, IRV, or plurality; can be counted in precincts; is simpler than IRV but not as simple as plurality and approval. It can also be counted in principle using "dumb" voting machines, but in practice that may not be a good idea (depends what kind of machines we are talking about) in which case you'd need new machines/software, or would need to count manually.
Range and approval voting are much freer of weird illogic problems.
Both also can allow
voters to express ignorance (if they want to).
So that's a sketch of my views. If you find yourself agreeing, please ENDORSE rangevoting.org by pushing the "endorse" button on its front page then filling the form.
I think it would be great if some places like Aspen adopted range and/or approval voting in single-winner elections. The world needs better quality democracy.
Thank you.
I think that one other important point to remember is that IRV benefits the majority in power. Dr. Steven Brams quote, as is on our website, says: "a runoff election is designed to keep a strong minority candidate from winning." That INCLUDES both a "runoff" election and an "instant runoff" election.
I've put a lot of thought into that and believe it is absolutely true. And, that is the reason that in places that IRV is being promoted, you'll notice it is the (political) majority behind it.
It stands to reason that if a city is overwhelmingly Democrat, for example, that most of the mayoral candidates would be from that side of the aisle, and once the secondary votes are tallied, they will end up going to the majority candidate. Also, in a runoff situation, with the top two candidates, a strong minority candidate would have little chance. In Mpls, if you look back to past elections, you will find that this is true.
I should also note that in Montana, it was the mainly the Republicans, and those right of center who had pushed for IRV.
Do we really want a voting system that is weighted so it favors those in power, or one that provides the most level playing field, the widest diversity of viewpoints and strongest most open elections for all involved, candidates and voters? I would suggest the latter creates a stronger democracy.
We believe its good to have 3rd and 4th party candidates as well as Indepentents, and that the so called spoiler effect is nonsense. It is healthy in a democracy to have more than two candidates. The people complaining about the spoiler problem are those who stand to lose from it, in spite of their calling themselves nonpartisan. Besides, who are we to tell someone they are wasting their vote? And, each candidate has just as much right to run for office as anyone does to vote.
Andy Cilek
www.MNVoters.org
Marylin,
I think it would be difficult being an activist AND a candidate at the same time. But, if anyone can do it you can. :-)
As you expose this serious flaw, as you apparently have been trying to warn of since February, those who mindlessly defend the City will try to beat you down.
Illegitimi non carborundum!
IRV IS A KIND OF VOTING VOODOO WITH ELECTION OFFICIALS OR VOTING VENDORS IN THE ROLE OF WITCH DOCTORS PRODUCING THE MAGICAL RESULTS. Not for the average layperson to understand for sure.
From Voting Matters Blog
"The counting of IRV is complex — the elimination of some candidates at the end of the first round means that second choice votes are transferred to other candidates. If a third round is required the elimination and transfer process continues. The average voter has to place great trust in the reliability of the counting algorithm in a way far beyond what is necessary in plurality voting. So the counting is opaque and non-transparent — a kind of voting voodoo with election officials in the role of witch doctor producing the magical results. If one believes strongly that the average voter should be able to understand and observe the counting of votes in a democracy, then IRV fails to meet this standard."
http://votingmatters.wordpress.com/2009/05/17/cary-nc-tries-irv-then-says-no-more/#more-175
Thanks to all for the many thoughtful responses.
"Ken Liles" - This is the first time Aspen ever used IRV, so I believe the reason we saw so many "bullet votes", and other voter ranking imperfections, is the lack of adequate voter education and experience with this voting method. I think these imperfections can be minimized in the future with better education, etc. So I wouldn't throw out IRV on that count, but would encourage our election officials to get this fixed in future elections.
The "anomalies", and assuming you're right that the more candidates there are the more anomalies we'll get, don't bother me at all. And thank you for agreeing with me that the "Michael effect" can't be predicted. I think this is a critical point that no one else seems to want to address.
On "transparency", I think transparency is critical in the post election tabulation of results. This idea that there's a "black box" and no one knows how the black box is tabulating the results, is a red-herring in my opinion (I know you didn't say this, but others have). The Aspen tabulation methodology is not that complicated and scaring voters with claims of black boxes gets us nowhere.
Practical effect #4 - thanks for the info on MN Supreme Court case. We'll see how it turns out - and I hear you on your Colorado point.
Practical effect #5 - can't get the link to work, but I did find my way to the RangeVoting.org site and will spend time reading. But I think your point continues to be the non-monotonicity effect on voters' psyches, which again I don't find very compelling. Using our recent nine candidate CC race, are voters really going to lay awake the night before trying to anticipate the "Michael effect"? The best thing a voter can do is follow the instructions - i.e. rank their choices - and let the chips fall where they may with the voter tabulation methodology we've chosen. The "what ifs" after the fact, are meaningless and I think are being exaggerated by some anti-IRV types.
Your closing pp - I understand your point, and it is valid. But if a large majority of voters say they don't want to go through the time, cost and "pain" of a protracted run-off, aren't they in effect making the judgment that IRV, or some other voting technique, is good enough for our local elections, and they will live with the results? That's what it feels like to me.
Thank you for your thoughtful comments.
"No IRV Fan" - sorry that you picked out one small phrase in my long blog and decided that somehow I was trying to "distract readers." For what it's worth, I think my blog was one of the most open-ended and inviting blogs that has been posted.
Your question about whether more debate should have been had on IRV earlier is certainly valid. But we are where we are, so the question I think CC will be asked to deal with is "where do we go from here?"
"Ann Brewer in MN" - you seem to be hung up by the monotonicity effect, but chose not to comment on my main point that in our nine candidate race for CC here in Aspen, it can't be predicted beforehand, so how much of an impact can it really have on voter behavior? I continue to believe, that the best thing a voter can do for his favorite candidate is rank her/him #1. If there is a post-election analysis that shows some unpredictable "anomaly", should we be bothered? I say no.
"Warren Smith" - you started your post by quoting two pp's from my post, but then proceeded to miss one key point. I posted "......isn't it true that the results will be a proper reflection of how the voters voted, USING THE METHOD OF TABULATION THAT ASPEN SELECTED?(emphasis added).
So your example of Hitler being elected is off point, because the Aspen tabulation method does not reject or "ignore" the ranked ballots that are submitted.
I've read your post a few times, but you take me on a journey of hypotheticals and some theory that just doesn't apply to what happened here in Aspen with 9 candidates for CC. I'm not sure how familiar you are with what we did or how our tabulation method works, so I'm having trouble responding. I will remain open-minded to "rangevoting" if that can be proven to be a better method.
"Andy Cilek" - thanks for your point about majority parties. I understand what you are saying, though I'm still trying to think through the effect of this here in Aspen, and whether the results would be very different with our traditional run-off methodology.
Thanks again for all the thoughtful comments. I remain unpersuaded by the practical reality of the non-monotonicity effect for the simple reason that it can't be predicted beforehand. If voters follow the instructions, rank their favorites and least favorites, and do the best they can ranking the "in betweens", I think they will be very satisfied with the results - which I sense is the case here in Aspen today.