UPDATE NOV 16 From the Chronicle: Ranked-choice voting sets stage for new tactics

UPDATE NOV 12 FairVote, ranked-choice-voting proponents, has issued a statement rebutting Don Perata’s criticism of ranked-choice-voting in his concession speech Thursday.

UPDATE NOV 11 3:30 p.m. Joe Eskenazi of The Snitch writes that “Don Perata Deserved to Lose,” focusing on the erstwhile Oakland mayoral candidate’s lack of a ranked-choice voting strategy.

His statement today that he’d have won easily in a conventional election was telling. Sure he would have. And if my mother had wheels, she’d be a bicycle.

Quite simply, you don’t sign up to play football and show up with a strategy befitting rugby. And you don’t get into an RCV election without an RCV strategy. Quan and others knew the benefit of picking up second and third votes. Perata didn’t try or didn’t care…

While the manner in which ranked-choice voting is tabulated is difficult to comprehend, the mindset of a voter is not: Vote for your favorite, second-favorite, and third-favorite. Simple.

RCV is not the cure for all of the political system’s shortcomings. But it’s hardly antidemocratic. Quite simply, there is no ideal situation to decide who should win a situation like the District 10 field, in which more than 20 candidates split some 17,000 votes. How is it less democratic for RCV to divvy up second- and third-place votes than to pick the top two vote-getters — each of whom amassed barely over 1,000 votes and outpolled other candidates by 100 or so tallies — and then run them against each other? How is it acceptably democratic for Tony Kelly to make a runoff with 1,200 votes, but undemocratic for Malia Cohen to beat him out with 3,700 RCV-adjusted votes? Read the full post.

UPDATE NOV 11: A column by the Chronicle’s C.W. Nevius Thursday is called “Ranked choice a rank choice for elections,” and asks the question: “What is wrong with a two-candidate runoff?”

UPDATE Nov 10: The Oakland blog Zennie62.com scores an interview with Dave Macdonald, Alameda County Registrar and the most sought-after vote-counter since Katherine Harris (no other connection or similarities implied.)

Yesterday we had a little back and forth here over whether the delay in releasing the final vote count in the Oakland mayoral election was due to ranked-choice voting. Here’s what Macdonald says in the above interview:

“We’re still processing ballots. It is a close contest. We’ve had more ballots to process in this election than any other election in our history, and the reason is of course because with ranked-choice voting, every vote in Oakland, Berkeley, and San Leandro got 3 ballot cards, and so it’s triple the amount of paper we have to handle.”

Macdonald did say later that even if ranked-choice voting weren’t in play, this election is particularly crowded and would have necessitated two ballot cards.

UPDATE Nov 9 2:37 p.m. From the East Bay Express:

  • Perata May Have Blown It On Ranked Choice Voting

    …Perata’s strategy, which essentially was to show disdain for the new voting system, may have backfired. By telling voters to just pick him, he may have alienated supporters of Kaplan and Joe Tuman, who is currently in fourth place. He also sent an unspoken message that if he was not a voter’s first choice, then they should just leave him off their ballots.

    It was a head-scratching maneuver, considering that none of the pre-election polls showed Perata receiving a majority of first-place votes. That meant he knew he had to get lots of second- and third-place votes to win. Yet he didn’t go after them. And it may cost him dearly. Perata has received 34% of the first-place votes, compared to Quan’s 25%, but she trounces him on second and third-place votes, propelling her over the 50% mark, according to the most recent results.

UPDATE 1:05 p.m. Here’s an interview KQED’s Cy Musiker conducted Monday with FairVote.org’s Steven Hill, the architect of Oakland’s ranked-choice voting system.

UPDATE 12:47 p.m. San Francisco Chronicle columnist Chip Johnson seems to have some big reservations about what’s happening in Oakland’s ranked-choice mayoral vote.

When the final votes are tallied, Oakland residents may wind up with the mayoral candidate who received less first-place votes than her closest rival, but tallied more second-place votes than anyone has seen before, to become the city’s next mayor.

The transfer of votes in such bulk is unprecedented in the history of ranked-choice voting in the nation, political experts said. No one saw this coming…

While the system worked well for the underdog candidates in the Oakland mayor’s race, the results do not represent the first choice of most Oakland residents by a fairly wide margin…

And if there is anything an aspiring Oakland politician can take away from the city’s first ranked-choice voting experience, it’s this: In an election where second- and third-place count, you are far, far better off being the least-hated candidate than the most popular.

And Oakland residents may get to find out if a mayor-elect who won the job on the strength of second-place votes has the mandate needed to carry out the city’s No. 1 job. Read the full column here.

UPDATE 12:07 p.m. Alameda Registrar spokesman Guy Ashley said he didn’t think the delay of the results in Oakland was due to ranked-choice voting, but because of the closeness of the race and the fact that reps from the campaigns are looking over the backs of the people scrutinizing the provisional ballots.

On the other hand, isn’t the election only close because of ranked-choice voting? If it had been a regular election, wouldn’t Perata have won outright because he received more first-place votes?

This doesn’t speak to the overall benefit of ranked-choice voting as a system. If you like the concept, you’re probably willing to sacrifice a week or so in receiving final results.

Wikipedia points to arguments for and against instant-runoff (aka ranked-choice) voting.

10:07 a.m. We received this email from Rob Richie, executive director of FairVote, an electoral reform organization that supports ranked-choice, or instant run-off, voting, and whose Senior Analyst, Steven Hill, is the architect of Oakland’s ranked-choice voting system. Mr. Richie took issue with my attributing the delays in reporting the final election results in the Oakland mayoral and San Francisco supervisor races to ranked-choice voting.

The delay in finalizing results is not due to ranked choice voting. It is due to the fact that Oakland, San Francisco and other communities had to open up absentee and provisional ballots and count them — a process that takes time no matter what system is being used. That’s why Ron Dellums’ election in June 2006 wasn’t assured into well into the second week after the elections.

Note that Dave Macdonald and Jon Arntz can push the button and update the RCV tallies anytime they want to — that’s easy. The hard part for them right now is counting ballots, and that would affect any close election.

The Oakland Tribune reported this today in a story about the Oakland mayoral race:

…However, for the (ranked-choice) system to work, all votes must be verified before the process can begin — thus necessitating the current delay in knowing the winner of a race.

So perhaps there’s some confusion here. Well, not perhaps. We have a call out to Dave Mcdonald, the Alameda Registrar, now.

For what it’s worth, outside of any delays that may or may not be occurring due to its implementation, I like the concept of ranked-choice voting, as it seems to me to eliminate the classic voting dilemma that attained its Nader uh I mean nadir in the presidential election of 2000. You know — you’re a die-hard liberal and you want to vote for Ralph, except you know there’s not a snowball’s chance in Florida of him winning. But you go ahead and vote for him anyway, then spend the next four years being vilified by your George W. Bush-hating friends for personally causing the Iraq war. If you could have ranked-choice that vote, you could have specified: 1) Nader 2) Gore 3) Your Uncle Ernie, and your Nader vote would have eventually accrued to Gore.

The same goes for votes on the right side of the political spectrum. You may want to vote for a Tea Party favorite that has no chance of winning because she, uh, oh I don’t know, once said she dabbled in witchcraft. If you know that vote would eventually kick in for the more mainstream candidate, you could do so without fear of contributing to the victory of a Democrat.

Not bad, in my book…

Here’s a video from Alameda County explaining ranked-choice voting.

The Ranked-Choice Voting Debate 16 November,2010Jon Brooks

  • james k. sayre

    Ranked election theft is what we are now watching in the Oakland mayoral election. My understanding of our American democracy is that elections are won by the person with the most votes. A simple and straight forward way of picking an election winner: just count all the votes and the candidate with the most votes is declared the winner. But election losers in the City of Oakland apparently tired of that simplicity. No, they decided that no one could be an election winner unless they had a majority of all the votes cast. Now this presented a small problem: what if no one candidate got a majority of all the votes cast in an election? So the powers that be invented a run-off election staged between the top two candidates. But the perennial election losers tired of this. No, they wanted to devise an election system that turned losers into winners. So we now have a rank voting system where the second and third choices of voters who voted for the least popular candidate can now be as important as the original votes for the leading candidate. This is insane. This is voters in Wonderland. It is illegal and unconstitutional to “distribute” the votes of one candidate and give them to another candidate. This rank election counting scheme violates my rights to have my vote counted in an honest election.

  • Edward Yu

    I beg to differ. “Vote theft” would be if the votes were being actually taken from another who needed them but as in a standard election in these localities, if no vote getter got a majority (over 50%), then a run-off election would ensue.. It appears to me that this process is fantastic as a cost effective way of eliminating the costs of a run off election. It is not theft nor dishonest, just a different implementation of an old process. But as I get older, I can see how it really is hard to accept changes like this. Nonetheless, it saves the taxpayers money (no run off election) and save voters the time to get in what they really feel by voting their conscious and their practical side.

  • Ranked-choice voting does not, as you implied, fix the Naderesque “spoiler” problem; it just delays it. Example: consider a race, A vs. B, where B is leading 55% to 45%. Then add a new candidate, C, that appeals primarily to B’s supporters; most prefer C over B, some still prefer B, but a few would rather have A than C. (In other words, a more popular Nader.) Who wins?

    45%: A > B > C
    10%: B > A > C
    15%: B > C > A
    30%: C > B > A

    It’s not the original winner, B; they’re eliminated first. And it’s not the new candidate, C; they lose in the final round. It’s A. That means that C spoiled the election.

    But there ARE voting methods that DO fix the spoiler problem. Approval voting is one such method, and it would be both faster to count and cheaper to administer than RCV.

  • Oh, one more thing: In all likelihood, RCV will also fail on it’s promise of majority preference. Since many voters didn’t list either Quan or Perata anywhere on their ballots, neither one will end up with 50% of all ballots cast; rather, one will end up with 50% of only the ballots that were not exhausted.

  • Mike LaBonte

    RCV is not like non-ranked elections, where for example you can hand count provisional ballots after election day and just add to the machine totals from election day. RCV requires all ballot data to be in one place so that the successive elimination process can take place. This throws off people who are used to getting approximate initial counts, and then just adjusting them later. You can’t do that.

    One question I have about Oakland’s new election rules is if a runoff is held when the RCV tally does not produce a majority. If not enough voters fully rank their ballots, someone will be declared winner with less than a majority. Note that in the final round Quan won with 43,825 votes out of 97,970 ballots cast, 44.7%. Will there be a runoff now?

  • Mike LaBonte

    Correction: the total number of ballots for the final round should not include the ones overvoted (228) or undervoted (1,304). So actually only 96,438 ballots count, and Quan won with 43,825/96,438 = 45.4%.

    Of course her official percentage is 51.09% because they don’t count the 10,664 ballots with votes for candidates who did not make it to the final round. Funny how that many people can show up, vote for anyone but Quan or Perata, and they are treated as though they never voted. The way RCV achieves majority is by redefining it.

  • Rob Richie

    Thanks for covering this.

    @Labonte — Mike is an opponent of RCV who knows that RCV elections are decided in one round. A December runoff this year would have had far less turnout — declines in turnout aren’t universal, but are the norm, especially when the first round has all the excitement of what was on the ballot this November in California. (In 2006, for example, far fewer voters participated in Ron Dellums’ election for mayor in Oakland decided in June — a open seat race, with Dellums barely securing a majorit — than this November.

    @Hess — Dale is another RCV opponent. He supports voting systems where your ballot will count for more than one candidate at the same time and a candidate who would win an absolute majority of 51% of first choices could lose to a candidate with barely any first choice support, so he has a hard case to make. What he neglects to say here is that among voters who indicated a preference for Perata or Quan, Quan as of Friday had won a majority.

    This last point speaks to James K. Sayre — as of the Friday count, more voters support Quan over Perata. Winning only 35% of the vote in the past would have required a separate runoff. It’s not enough to win in the first round.

    • FactCheck

      Using RCV’s logic, if you eliminated Perata because he had the lowest votes in round 10 –

      Quan would have won with 100% of the votes!

  • Hello again Mr. Richie. My last name is Sheldon-Hess; it’s hyphenated.

    You and I have, of course, been down these roads before, so I’ll try to keep this short for everyone else’s benefit.

    As you well know Rob, no voting system is perfect. It’s true that, in the case where over 50% of voters prefer one candidate over all others, RCV will always elect that candidate, while approval voting will, occasionally but not often, fail to do so. But you will note that such was not the case in this election: there was no majority among first-preferences. Nor was it the case in this year’s Governor elections in Connecticut, Florida, Illinois, Maine, Minnesota, Ohio, Oregon, or Vermont, or the Senate elections in Alaska, Colorado, Florida, or Illinois. What’s most important, I think, isn’t rigid adherence to often-inapplicable criteria (not that I couldn’t rattle off a half dozen equally-reasonable criteria that RCV fails), but rather the expected performance of a method weighted across all likely outcomes, so that all criteria are incorporated in proportion to how frequently, and how egregiously, they are failed.

    And approval voting performs much better than RCV by any such measure. In fact, after plurality, RCV performs the worst out of dozens of commonly-suggested voting system reforms. William Poundstone’s superbly-readable “Gaming the Vote” covers this quite nicely, but this image distills the conclusion (note: it refers to RCV by the name it was better known under a few years ago, “instant runoff” (before your organization’s recent re-branding effort)).

    Honestly Rob, I can’t figure out why you’re so adamant about this. The data is against you, and you’re losing cities that figure that out (Aspen CO, Burlington VT) just as quickly as you’re gaining them with your one-sided explanation of things (Portland ME, Minneapolis MN); so I guess we’ll be talking again, in another city paper’s editorial pages, real soon. See you there.

    • Mike LaBonte

      I have to agree with Dale on approval voting. The comparison of methods at rangevoting.org makes range voting seem like the best choice, but approval voting is pretty close in terms of mathematical qualities, and is easier to understand and implement than either RCV or range.

      A question for the RCV experts: in how many places has RCV/IRV (not multi-seat STV) ever been used, and how many still exist? My understanding is that it has been tried and ultimately rejected many times since the 1870s. FairVote has a well-written History of IRV page that is interesting, but it does not easily translate to an accurate tally of adoptions and rejections of IRV.

  • Mike LaBonte

    If RCV were not used the standard approach would have been for Perata and Quan to go into a runoff. The upside is that there would be more campaigning focusing on just two candidates. Rob is correct that a downside of runoffs often is low turnout. But at least the voters KNOW whether or not they are participating. Preliminaries are even better. They are like runoffs but are more definite, and the turnout is usually better than runoffs.

    • FactCheck

      Historically, Oakland Run-offs always enjoy higher turn-out than the first election. Oakland uses a consolidated election, putting their city elections with state/federal elections in June, then a run-off, if needed in Nov.

      Though often repeated by RCV salespeople, turnout doesn’t necessarily drop in Run-offs in high profile races – including the old SF December run-offs prior to RCV.

      SF Turnout:
      Nov. ’09 44.95%
      Dec. ’09 Run-off 48.48%

      Nov. 03 45.67%
      Dec. ’03 Run-off

      • Rob Richie

        Oakland has mayoral elections in even years. Just as with SF’s Board of Supervisor races with runoffs pre-RCV, turnout would drop from the November election with governor, etc on the ballot to a December runoff. And big money of course would be more likely to have its day — “independent” expenditures quadrupled in SF’s old runoffs, usually manifested n the form of negative attacks ads.

        Meanwhile, Oakland had about three-quarters of its pre-RCV elections won without the need of a runoff in smaller, less representative primary electorates.

  • Mike LaBonte

    Just listened to the Steven Hill interview audio, in which he said the single RCV election was cheaper than an election and a runoff. Does anyone have actual numbers? RCV elections usually cost much more than non-ranked elections.

    Hill also compared the 106,000 who voted in this election to the 83,000 who voted in 2006. But there was no runoff after the 2006 election (6 candidates), and the 2010 RCV (10 candidates) went on for 10 rounds! It’s hard to say whether it was RCV, the candidates, or something else on the ballot that brought the higher turnout and closer race this time. Did 10 candidates run only because RCV was used?

    Again, by the official definition “only” 85,774 voters elected the Mayor in the final round this year, not 106,000. But a runoff most likely would have had even lower turnout.

    • FactCheck

      The 2006 mayoral Election occurred in June, when turnout is slightly lower that the Dec. General. Hill is comparing a primary with a general.

      Oakland had June elections, then a runoff between the top two in Nov. if no one got 50%+1 in June. Turnout would be higher in Nov.

    • Rob Richie

      Mike’s homework is always selective.

      Cary (NC) saved money in its first RCV election, as Mike knows. This year, North Carolina ran a statewide election with RCV, with 1.9 million voters. Statewide elections in NC cost several million dollars. the cost of RCV will be much less. Any costs for RCV elections are primarily in the transition to them, not in ongoing costs. Running RCV elections this year in SF is less expensive that what running four December runoffs would have cost.

      On turnout, Mike neglects to mention that ballots are still being tallied. He also overlooks some intriguing statistics like the fact that in the June 2006 election, nearly twice the percentage of people skipped voting in the mayor’s race.

      Note that in the pre-RCV world, most Oakland races were won in lower turnout June primaries. Those electorates were starkly less representative of the overall electorate in many ways – age, race, income, etc.

  • fred

    I totally agree with James k. Sayre that it seems like stealing votes from the winner. I do not like this ranking system and it is a very unfair system to the person with the majority of the votes. The polls showed that Perata had the most votes and he should have been or should be declared the winner automatically. The voting system should be as simple as this: the person with the majority of vote, which in this case it is Perata should be declared the winner and that should be the end of the story.

  • Mike LaBonte

    Fred says the person with the majority of votes should be the winner. Well Perata had votes somewhere on 41,949 ballots and Quan had votes somewhere on 43,825 ballots. So Quan wins.

    Should we care only about first place votes? Then in a race with 10 candidates someone theoretically could win with approval from barely more than 10% of the voters. And you have the spoiler problem too.

    I assume “stealing votes from the winner” means transferring Perata’s votes to Quan. Actually in the final round Perata still had every vote given to him on any ballot. It was the votes for all other candidates that were transferred. The first round is basically saying “People who voted for Fields, he can’t win so who would you vote for now?”. It’s the same thing that happens in a runoff, where some voters have to choose a candidate other than the one they originally wanted.

    If you are against RCV “in principle” then you must be against runoffs too.

  • I’m with Dale on this one. The empirical data and game theoretical analysis all show Score Voting and Approval Voting to be better than IRV in essentially every quantifiable way. (See my “website” link for an enumeration of those differences.)

    Richie is reduced to quibbling over cost savings that, if present at all, are minor compared to the cost of bad government. I have seen the election costs for the past several years here in my home of San Francisco, and there is no clear indication of any substantial cost savings since implementing IRV.


Jon Brooks

Jon Brooks is the host and editor of KQED’s health and technology blog, Future of You. He is the former editor of KQED’s daily news blog, News Fix. A veteran blogger, he previously worked for Yahoo! in various news writing and editing roles. He was also the editor of EconomyBeat.org, which documented user-generated content about the financial crisis and recession. Jon is also a playwright whose work has been produced in San Francisco, New York, Italy, and around the U.S. He has written about film for his own blog and studied film at Boston University. He has an MFA in Creative Writing from Brooklyn College.

Sponsored by

Become a KQED sponsor