L.A. NOW

Southern California -- this just in

« Previous Post | L.A. NOW Home | Next Post »

L.A. County to study using instant runoff elections

In an effort to avoid multiple, costly runoff elections, Los Angeles County supervisors unanimously voted today to investigate whether the county can save money and boost turnout through instant runoffs.

Instant runoffs prevent repeated special primary elections by allowing voters to rank candidates instead of just voting for their top choice. If no candidate wins a majority, the candidate with the fewest first-choice votes is eliminated and his votes redistributed based on ballot rankings until someone wins a majority.

The process, also known as “ranked choice voting,” has been used in San Francisco, Burlington, Vt., Takoma Park, Md., Cary, N.C., and Pierce County, Wash.

Supervisor Mark Ridley-Thomas proposed that the county consider instant runoffs to combat “voter fatigue,” noting that by the end of May, residents in his 2nd District will have participated in six elections since January 2008. That included the March 24 special primary to fill the state Senate seat he vacated when he was elected supervisor in November; it cost the county registrar-recorder $2.2 million and saw voter turnout of just 6%.

“The costs are serious, and instant runoff voting is a more efficient way to expedite the voting process,” Ridley-Thomas said today.

-- Molly Hennessy-Fiske

 
Comments () | Archives (11)

I would urge anyone considering a push for Instant Runoff Voting to take a serious objective look at its common shortcomings, such as in the recent IRV mayoral election in Burlington, Vermont. It actually featured a case of non-monotonicity, where the winner would have become the loser if more people had ranked him HIGHER. It also featured the no-show paradox, where a group of voters would have been better off by staying at home than by showing up to cast a sincere vote. You can find out a lot more in this analysis by Warren Smith, a Princeton math Ph.D. who is an expert on election methods.

http://scorevoting.net/Burlington.html

I would recommend that the county investigate other alternative election methods which can also avoid the cost of expensive runoff elections, such as "score voting" (scorevoting.net). Particularly since IRV has some problems with respect to cost:

- IRV may require the purchase of new voting machines, whereas score voting can be run on all voting machines currently in use in the country.

- Since IRV proceeds over many discrete rounds (potentially more than two), there are more chances for tied or nearly-tied tallies, increasing the chance of an expensive recount (or lawsuit). Score voting does not have this problem.

- IRV needs to be counted centrally; it cannot be counted in precincts like our current system. Centralizing a higly parallelized process can increase costs. Whereas score voting, like the current system, can be counted in precincts.

More information is availabe at the Center for Range Voting website: scorevoting.net

I have a masters' degree in mathematics and am executive director of The National Election Data Archive, a group of mostly statisticians devoted to fair, accurate elections. Also, I am currently an expert witness for plaintiffs in a legal case in Minnesota that is challenging the constitutionality of IRV on the basis of its violations of the requirement for equal treatment under law.

Before voting to adopt or voting to test Instant Runoff Voting, LA should investigate the truth about its flaws. For instance, many false claims have been made about IRV that are simply false. For instance IRV does *not* find majority winners, does *not* solve the spoiler problem, and is *not* less costly than top-two runoff elections.

Worse, IRV is a fundamentally unfair method for counting ranked choice ballots because it counts the 2nd, 3rd, etc. choices of only *some* voters, and out of the 2nd, 3rd,... choices that IRV considers, only some are considered in a timely fashion when those choices could still help that candidate's chances to win.

Due to IRV's unequal treatment of voters' ballots, IRV is non-monotonic (I.e. a 1st choice vote *for* a favorite candidate can cause that candidate to lose an election, whereas ranking that same candidate last or not at all, can cause that candidate to win!), and IRV will often elect extremist candidates that are opposed by a majority of voters and eliminate moderate candidates who are supported by a majority of voters. This exact situation recently occured in the Burlington,VT IRV election.

For more information about IRV, please read these documents:

North Carolina develops method to count IRV with precinct-opti-scanners
http://kathydopp.com/serendipity/index.php?/archives/41-North-Carolina-develops-method-to-count-IRV-with-precinct-opti-scanners.html

The Additional Costs of an Instant Runoff Voting Election
(In an email from Chris Walker, Jackson County Election Official, WalkerCD@jacksoncounty.org)
http://electionmathematics.org/em-IRV/IRV-Cost-Considerations.pdf

The 18 Flaws and 4 Benefits of IRV
http://electionmathematics.org/ucvAnalysis/US/RCV-IRV/InstantRunoffVotingFlaws.pdf

Burlington Vermont 2009 IRV mayor election
Thwarted-majority, non-monotonicity & other failures (oops)
http://rangevoting.org/Burlington.html

The Legal Case Against IRV
http://electionmathematics.org/em-IRV/MNcase/MNSupremeCrtOrder-3-17-09.pdf

LA would be wise to wait until after the question of IRV's legality under the US constitution is decided before investing the resources that it would take to switch to such a fundamentally unfair, costly, and inauditable voting method.

There are numerous other alternative voting methods that are far more fair and easily administered than IRV is, including methods that use rank choice ballots or approval ballots or rating ballots.

Almost any voting method is more fair and less costly and more auditable and finds more majority winners than IRV including your existing plurality/top-two runoff method.

IRV is one of the most flawed election methods there is:

Instant Runoff Voting is associated with LOWER voter turnout in San Francisco California and Takoma Park Maryland, home of Fair Vote Director Rob Richie. (see The activist city that didn’t vote Will there ever be enough candidates in a Takoma Park election triggerInstant Run-off Voting? by Tamra Tomlinson 12/07/2007)http://www.takoma.com/archives/pdfs/2007/1207pdfs/tp1207/TKP_MN_120107_00_00_017_c.pdf

San Francisco’s 2007 IRV mayoral contest had 100,000 fewer voters than did San Francisco’s one on one runoff mayoral contest in 2003. (See http://www.instantrunoffvoting.us/turnout.html)

IRV is associated with LACK of racial minority representation.
in Takoma Park Maryland. Takoma Park is 34 percent African-American..” yet “….Takoma Park, in Montgomery County, has no minority representative. ” (See (See Greenbelt mulls changes to its voting system Thursday July 24, 2008 Gazette.net)
http://www.instantrunoffvoting.us/minorities.html

IRV is also associated with lack of racial minority representation in Australia, where IRV has been used for decades. The Center for Range Voting advises that: “in the top IRV-using country (Australia) and the top two plurality-using countries (USA & UK) in years 2005-2008. We find, e.g, that there are zero racial-minority IRV seatholders… ”
(See “Instant Runoff Voting (IRV) and racial minorities” http://www.rangevoting.org/IRVraceMinorities.html )

IRV costs significantly more than plurality elections
http://www.instantrunoffvoting.us/costs.html
yet IRV in most elections provides a plurality result
http://www.instantrunoffvoting.us/majority.html

The instant runoff election in Burlington,Vermont suffered from nearly every pathology in the book! Non monotonicity - where with instant runoff, a voter can hurt their preferred candidate by ranking them first. A spoiler effect - in this election, Kurt Wright was the spoiler. The "no show" paradox - Wright supporters who also supported Montrol would have helped him if they hadn't shown up to vote at all. Majority failure -the candidate supported by the most voters did not win. Incumbent protection thanks to name recognition. Centrally counted votes - instant runoff opened up the election to fraud because votes were not counted where cast.
http://instantrunoff.blogspot.com/2009/03/burlington-instant-runoff-election.html


IRV requires more complex technology to count, because IRV is not additive and cannot be summed up at the polling places.
http://www.instantrunoffvoting.us/votingsystems.html

CARY NC REJECTED A 2ND IRV EXPERIMENT. March 12, 2009 Cary North Carolina turned down second bite of Instant Runoff Voting Pilot, process still too flawed The City Council chose plurality by a vote of 6-1 instead, citing problems with IRV and noting that the one IRV election they had produced a plurality winner, not a majority winner.
http://irvbad4nc.blogspot.com/2009/03/instant-runoff-voting-pilot-remains.html

See www.instantrunoffvoting.us for facts, reports, news.

"For instance IRV does *not* find majority winners, does *not* solve the spoiler problem, and is *not* less costly than top-two runoff elections."

IRV greatly reduces the spoiler problem.

IRV does find majority winners. Not always a Condorcet winner, but still a majority winner. How does it find majority winners less than top two runoff?

Yes, it is less costly than top-two runoff elections. San Francisco has saved money with avoided runoffs. So did Burlington.

"IRV will often elect extremist candidates that are opposed by a majority of voters and eliminate moderate candidates who are supported by a majority of voters. This exact situation recently occured in the Burlington,VT IRV election."

Plurality would have elected Republican Kurt Wright, who was the LEAST favored of the three strongest candidates. Andy Montroll, who was the Democratic candidate, would have been the spoiler to Bob Kiss, who was the Progressive candidate and the IRV winner. Kiss was favored by a majority over Wright, and Montroll got fewer first place votes than either Wright or Kiss.

Top two runoff would have given us an election between Kiss and Wright. A longer and more expensive campaign for the candidates, greater expense for the city to hold the runoff, probably lower turnout than IRV, and the possibility of electing Wright if his turnout declined by less than Kiss' turnout.

Under range voting, Montroll would have the best chance at victory, but Kiss or Wright could have also won, depending on how strongly voters preferred their particular candidates. Smith, the fourth place Independent candidate, had a more distant shot at winning.

"Worse, IRV is a fundamentally unfair method for counting ranked choice ballots because it counts the 2nd, 3rd, etc. choices of only *some* voters, and out of the 2nd, 3rd,... choices that IRV considers, only some are considered in a timely fashion when those choices could still help that candidate's chances to win."

It's very simple. Each voter has the right to one vote in each round. Nothing against the Constitution there.

We find, e.g, that there are zero racial-minority IRV seatholders… ”

Ross Mirkarimi, who is very proud of his Iranian heritage, is on the San Francisco Board of Supervisors. Ed Jew served on the Board of Supervisors, and he is Chinese.

"IRV requires more complex technology to count, because IRV is not additive and cannot be summed up at the polling places."

They use paper ballots in Australia and Ireland. Sounds pretty low-tech to me.

"San Francisco’s 2007 IRV mayoral contest had 100,000 fewer voters than did San Francisco’s one on one runoff mayoral contest in 2003."

Lower turnout had more to do with the weakness of the challengers to the Mayor than the voting system. No method of voting is capable of making elections perfect.

Do you blame plurality voting every time there is a decline in turnout?

"The instant runoff election in Burlington,Vermont suffered from nearly every pathology in the book! Non monotonicity - where with instant runoff, a voter can hurt their preferred candidate by ranking them first. A spoiler effect - in this election, Kurt Wright was the spoiler. The "no show" paradox - Wright supporters who also supported Montrol would have helped him if they hadn't shown up to vote at all. Majority failure -the candidate supported by the most voters did not win. Incumbent protection thanks to name recognition. Centrally counted votes - instant runoff opened up the election to fraud because votes were not counted where cast."

Incumbent protection: If the incumbent winning is a pathology, then no voting system will every satisfy you! Fusion is out the window because Hillary Clinton won re-election in New York in 2006. So is top two runoff, because Saxby Chambliss won re-election in Georgia last year. There are too many incumbents that have been re-elected under plurality to list here.

Centrally counted votes: Plurality is not fraud-free, even when it is not centrally counted. Furthermore, there is no evidence of fraud in Burlington's election.

Spoiler: IRV prevented Montroll from being a spoiler to Kiss under plurality. Wright was a strong candidate, hardly someone I would call a spoiler.

No-show paradox: The no-show paradox didn't show up in this election! If 753 Wright voters who had Montroll as their second choice stayed home, Montroll would have been in the final round with Kiss. However, their strategic slacking would have been for naught as Kiss still would have won. Top two runoff was actually more vulnerable to the no show paradox in this election.

We find, e.g, that there are zero racial-minority IRV seatholders… ”

San Francisco Board of Supervisors:

Eric Mar (District 1), David Chiu (District 3), and Carmen Chu (District 4) are Asian. Sophie Maxwell (District 10) is Black. See previous post for Ross Mirkarimi (District 5).

These comments are in response to Donna Miller's several comments:

First, the fact that there are a handful of minority representatives in San Francisco is interesting, but not conclusive, particularly if their districts are majority-minority; although I do not know if that is the case. If you have that information, I would be interested to see it. If that /is/ the case, than IRV's tacit encouragement of racial segregation would not be a mark in its favor; or perhaps racial differences are simply less defining to San Franciscans than they are to African Americans in Maryland?

Second, while it's true that "nothing is perfect", central counting of votes does make a less-than-perfect situation even worse, in, I think, an obvious way. Furthermore, central counting may also be more expensive.

Third, there is a definition for spoiler. A spoiler is a non-winning candidate who, if removed from all ballots, causes the winner(s) of the election to change. By the definition, Wright was a spoiler for Montroll. It is true that IRV makes spoilers _less_ likely (to be a spoiler, a candidate needs at least 25% of the votes in a 3-way contest, unlike our current system which requires as little as one vote), but it is often claimed that IRV eliminates them, which is not true. However, there are other alternative voting methods which ARE completely immune to spoilers, such as score voting.

Fourth, there most assuredly was a no-show paradox in Burlington; if those 753 Wright voters had stayed home, you are correct that Montroll would have faced Kiss in the final round; but you are incorrect when you claim that Kiss would still have won. Montroll would have won; which for most of those 753 voters, would have been a better outcome (as they preferred Montroll over Kiss). I'm curious why your calculations seem to come out differently from mine.

Fifth: IRV claims to always choose a "majority winner", but this is left as a vague term in IRV literature. Personally, I would consider a majority winner one that is preferred by the majority when compared to any other candidate; i.e., a Condorcet winner. However I do not know what IRV advocates are using as a definition of "majority winner"; could you please tell me how you define this term? Keeping in mind that Montroll was a Condorcet winner in this election, so please be precise enough to clarify how Kiss was a majority winner but /not/ the Condorcet winner, and why the IRV definition of "majority winner" could be considered superior. I would very much like to know.

I know there is some friction between IRV supporters and supporters of other election reforms, and I am trying to be informative and to honestly seek information, and avoid antagonism; but this being the internet, I want to state that clearly. We're all interested in the same thing; finding the fairest democratic system.

Thanks to Dale Sheldon for his thoughtful reply. I like IRV, but I'm also in favor of allowing localities to experiment with other methods like score voting. I'd like to give a more complete reply, but I'll be busy over the weekend. I'll be happy to write more on Sunday evening.

An analysis (from the point of view of voting-methods theory) of the Los Angeles Mayoral elections over the years is available at
http://rangevoting.org/LAmayors.html .

The 2001 election is interesting. Instant Runoff
would have yielded a different winner and different
2nd-place finisher too. It can be claimed to have
malfunctioned. "Range voting" is a better and simpler 1-round single-winner voting system. I would suggest trying it.

http://rangevoting.org

As an election-method expert I'll state that IRV is an improvement over plurality voting, which is what's currently used throughout the U.S. That comparison is clear because plurality voting is so unfair that nearly any reasonable alternative (including IRV) is an improvement. Another way to say this is that the "single-mark ballots" we currently use are too primitive. Election-method experts agree about the need to collect more information on ballots.

As seen in these comments, election-method experts disagree about how to do the calculations using increased voter-preference information. That's because the mathematics isn't simple. I've written a book titled "Ending The Hidden Unfairness In U.S. Elections" that clarifies (in pictures, tables, and clearly written explanations) many of the complications. The book also explains VoteFair ranking, which is a fairer election method I developed.

In this L.A. election context, the most important point to consider is that IRV weaknesses don't show up clearly in our current two-party system, but if IRV is too widely adopted then the many weaknesses of IRV will become obvious. At that point it will be difficult to shift to a fairer method.

If fairness is really the goal, then the biggest gains will come from reforming _primary_ elections to use 1-2-3 ballots and nearly any calculation method (including IRV). That approach will, at a minimum, dramatically reduce the vote-splitting that special interests use to eliminate reform-minded candidates. The result would be candidates in general elections who are closer to what voters want. Then the common reaction of "I don't like any of the candidates" will become less common.

Richard Fobes
Author of "The Creative Problem Solver's Toolbox"


Connect

Recommended on Facebook


Advertisement

In Case You Missed It...

Video

About L.A. Now
L.A. Now is the Los Angeles Times’ breaking news section for Southern California. It is produced by more than 80 reporters and editors in The Times’ Metro section, reporting from the paper’s downtown Los Angeles headquarters as well as bureaus in Costa Mesa, Long Beach, San Diego, San Francisco, Sacramento, Riverside, Ventura and West Los Angeles.
Have a story tip for L.A. Now?
Please send to newstips@latimes.com
Can I call someone with news?
Yes. The city desk number is (213) 237-7847.

Categories




Get Alerts on Your Mobile Phone

Sign me up for the following lists: