Friday, March 26, 2010

Response to Kathy Dopp "Report" on IRV "Flaws"

Kathy Dopp, a Blogger and lone member of what she calls the "National Election Data Archive," has authored a "report" alleging 18 flaws of IRV. Her "report" has been floating around the Internet, but is deeply flawed itself.
Here is a response to each of the points she raises...

1. Dopp: "Does not solve the "spoiler" problem except in special cases..."

Dopp has her "special cases" reversed. In fact, IRV solves the spoiler problem in virtually all likely U.S. partisan elections. Whenever a third party or independent candidate is unlikely to be one of the top vote-getters (true in over 99% of U.S. elections), IRV eliminates the spoiler problem completely. If a third party grows to the point that its candidates out-poll major party candidates, another issue that is related to the spoiler problem, can occasionally arise. This is where supporters of a third party candidate may worry that by supporting their favorite candidate, they risk causing their less-preferred compromise choice to be eliminated from the final runoff, leading to the election of their least-preferred choice. In other words, the issue of whether to vote for your favorite choice, or to rank your compromise choice first can resurface in this unique circumstance. But this is extremely rare and no different than a candidate in a party’s political primary arguing "Vote for me because I am more electable in the general election."

2. Dopp: "Requires centralized vote counting procedures at the state-level…"

IRV creates no need to centralize the counting or the ballots themselves, although that is one possible counting procedure -- and indeed a central count is often sensible for smaller jurisdictions. But all that is required to implement IRV is central coordination of the tally. If ballot images are recorded on optical scan equipment, the data from those images can be collected centrally for an IRV ballot (with an appropriate manual audit to confirm accuracy). If a hand-count is conducted, vote totals need to be reported to a central tallying office in order to determine what step to take next in the count. In Ireland, for example, there are 43 counting centers in the presidential race. Election administrators count ballots and report their totals to a national office that in turn instructs the administrators at each counting center on what to do next. The entire process takes less than a day even though more than a million ballots are cast.

3. Dopp: "Cannot be implemented without modification to the ballots or to the optical scan voting machines or their software."

She is wrong when she says older "discrete-sensor machines cannot accommodate ranked ballots. San Francisco has used such voting machines for IRV since 2004. Obviously one needs to modify the ballot to give voters the option of ranking candidates, and many current voting machines are not programmed to record ballot rankings. While true, this is hardly a "flaw" of IRV.

4. Dopp: "Encourages the use of complex voting systems and …[FairVote promotes] electronic-balloting…"

Most government IRV elections are in fact conducted with hand-count paper ballots, including national elections in Australia, Ireland and Papua New Guinea. FairVote is a leading advocacy organization for IRV, but it is joined in supporting IRV by numerous other organizations and individuals, including the founders of TrueVote Maryland and election integrity leader David Cobb and Anthony Lorenzo.

As to FairVote, it advocates the replacement of all paperless voting machines with paper-ballot systems, such as optical scanners. All three of the major voting machine vendors have created optical scan options for ranked-choice ballots. Not all of these are ideal (some, for example, cannot handle more than three rankings), but FairVote expects IRV elections to be overwhelming run on paper ballot systems in the future. FairVote advocates that all such machines store a redundant electronic record of each ballot, as well as a paper ballot to allow for better fraud detection, and simplify ranked ballot tabulations. Rather than making such elections more complicated, this would simplify the process, while improving transparency and integrity.

5. Dopp: "Confuses voters…"

All the evidence shows that voters are not confused by IRV. The rate of spoiled ballots did not increase in any of the U.S. cities when they switched to IRV. For example, Burlington (VT) used IRV for the first time in a hotly contested race for mayor in 2006, and among those casting votes in the IRV race fully 99.9% of ballots were valid, with the very highest valid ballot rate in the ward in town with the highest number of low-income voters. San Francisco’s rate of valid ballots in the most closely contested race in its first citywide election with IRV was 99.6%. Furthermore, exit polls have been conducted in every city having an IRV election for the first time in the modern era. Each survey shows that voters overwhelmingly prefer IRV to their old method of elections.

6. Dopp: "Confusing, complex and time-consuming to implement and to count…"

IRV certainly is simpler for election officials and voters than conducting a whole separate runoff election to find a majority winner. It is more complicated to administer than a single vote-for-one election, but election officials have adjusted well to their new responsibilities. Note that the winning threshold for an IRV election, as with any election, must be specified in the law.

7. Dopp: "Makes post election data and exit poll analysis much more difficult to perform…"

To date, IRV election can make it easier to do post-election and exit poll analysis. Because optical scan counts with IRV require capturing of ballot images, San Francisco (CA) and Burlington (VT) were able to release the data files showing every single ballot's set of rankings – thereby allowing any voter to do a recount and full analysis on their own.

Exit polls can be done just as well under IRV rules as vote-for-one rules. California requires a manual audit in its elections, which has been done without difficulty in San Francisco’s IRV elections. Manual audits should be required for all elections, regardless of whether IRV is used or not.

8. Dopp: "Difficult and time-consuming to manually count…"

Manual counts can take slightly longer than vote-for-one elections, but aren't difficult, unless many different races on a ballot need to go to a runoff count. As cited earlier, Irish election administrators can count more than a million ballots by hand in hotly contested presidential elections in one standard workday. In most IRV elections the bulk of the ballots have first preferences marked for the two strongest candidates, so these ballots only need to be sorted once. It is only the small stacks of ballots for eliminated candidates that may need to be resorted according to alternate choices.

9. Dopp: "Difficult and inefficient to manually audit…"

IRV can be manually audited just as well as vote-for-one elections, although it does take more effort (since voters must be allowed to express more information on their ballot). Contrary to Dopp's insistence, there is no need to use precinct sums to perform an audit. A manual audit can be done using a random sample of ballots from a random sample of voting machines to confirm that the ballot records are accurate, or by a complete re-tally from a random sample of voting machines. A complete re-tally of all ballots (a recount) is, of course, possible but unnecessary unless a court recount is ordered.

10. Dopp: "Could necessitate counting all presidential votes in Washington, D.C.…"

If the Electoral College were abolished and IRV were then adopted for future national popular vote elections for president, there would need to be national coordination of the tally in order to know which candidates got the fewest votes nationwide and needed to be eliminated – just as in Ireland. But the actual counting of ballots does not need to be federalized any more than if IRV was not used, and could be conducted by counties, states or whatever level is easiest and most secure for that jurisdiction. Note that voters certainly would be pleased to have a majority winner in elections for our highest office.

11. Dopp: "IRV entrenches the two-major-political party system …"

IRV neither "entrenches" nor "overthrows" the two-party system. It simply ensures no candidate wins in the face of majority opposition. If a minor party has the support to earn a majority of votes, it will win in an IRV election.

IRV is a winner-take-all method, like plurality voting and two-round runoffs. However, IRV allows independents and candidates with minor parties to run without being labeled as spoilers. This may reveal a higher level of support for these parties, and if these parties are attractive to voters, their support may grow.

Relating to multi-party representation, any winner-take-all, single seat election method tends towards two dominant parties, at least in any given geographic area. To allow for multiple parties to regularly win office, jurisdictions should adopt a form of proportional representation in which candidates will be able to win office with less than 50% of the vote.

Note that Australia’s IRV elections are often cited as an example of two-party domination. But while the two major parties (one of which is actually a coalition of two parties, with one party running in one particular region of the country) dominant representation, the minor parties contest elections very vigorously, with an average of seven candidates contesting house elections in 2007. That year the Green Party did not win any seats in house elections, but it ran candidates in every district and earned 8% of the national vote. It naturally would prefer a proportional representation system, but supports IRV over alternate winner-take-all systems and uses it to elect its internal leaders.

12. Dopp: "Ranking a voter’s first-choice candidate LAST could cause that candidate to WIN…"

Dopp is referring to what election methods experts refer to as the "non-monotonicity" paradox. The key fact she doesn't state is that this is not unique to IRV, but common to all runoff election systems (whether instant or traditional). While technically correct, her presentation is intentionally miss-leading. The mere receiving of an additional first choice vote can never be the cause of a candidate's losing (although her wording intentionally implies this). Non-monotonicity is the result of the possibility that a change in the order of finishing of the other candidates that results from switching votes may mean that the otherwise winning candidate will face a stronger opponent in the final round of the runoff (whether an instant runoff or a separate runoff). In traditional separate runoff election systems this possibility may open the door to strategic manipulation, in which voters vote for a weak opponent of their true favorite in the first round, and then switch to their true favorite in the second round. Fortunately, with IRV the appeal of such manipulative strategy is practically eliminated, because a voter is not able to change her first choice between rounds, and thus the risk of the strategy backfiring is much greater with IRV than in a traditional runoff. For a fuller discussion of the monotonicity issue see

13. Dopp: "Delivers other unreasonable outcomes..."

Unreasonable outcomes are less likely with IRV than with any other single-seat voting method in use today. Every single voting method ever proposed can deliver "unreasonable outcomes" in some scenarios, but real-world experience has shown IRV to be one of the best methods. The overwhelming number of election method experts agree that IRV is fairer and more democratic than plurality voting even if some might prefer other theoretical voting methods. The American Political Science Association (the national association of political science professors) has incorporated IRV into their own constitution for electing their own national president. Robert’s Rules of Order recommends IRV over plurality voting.

As to the specific examples…Irv can indeed have more ties, because there may be numerous rounds in the tally. However, such ties are for last place and elimination, and have little possibility of mattering. Most jurisdictions with IRV use a rule that eliminates in a single batch all candidates at the bottom with no chance of winning, so that none of these potential ties among write-ins or fringe candidates need to be settled.

As to the vanishingly rare but mathematically possible pair-wise "lose to everybody except one" possibility, Dopp fails to mention that plurality elections frequently elect the candidate who in pair-wise comparisons would lose to every other candidate (the "Condorcet loser") which can never happen with IRV.

Real world experience for over 80 years in Australia proves that IRV does not in fact favor extremist candidates over centrists. Certainly not has much as plurality elections can. This is because under IRV a candidate needs to not only have strong core support, but also appeal to the supporters of other candidates for second choices. If anything, candidates in the political center have the benefit that it is easier for them to win second rankings from the supporters of candidates on either side of them politically, than for candidates at the margins.

14. Dopp: "Not all ballots are treated equally…"

This charge reveals a lack of understanding of how IRV works. All ballots are treated equally. Every one has one and only one vote in each round of counting. Just as in a traditional runoff, your ballot counts first for your favorite candidate and continues to count for that candidate as long as he or she has a chance to win.

Your rankings should be considered as backup choices. Your ballot will only count for one of your lesser preferences if your favorite candidate has been eliminated. Every ballot counts as one vote for your highest ranked candidate who is still in the running in every round of counting.

Note that courts in Michigan and Minnesota have upheld IRV for this very reason and Robert’s Rules of Order recommends it over plurality voting. For some key quotes from the Michigan court decision upholding IRV's equal treatment of ballots, please see note <1> below, or for the full court decision see

15. Dopp: "Costly…"

The two main expenses associated with the transition to IRV are voting equipment upgrades and voter education. Both of these are one-time costs that will be quickly balanced out by the savings coming from eliminating a runoff election in each election cycle. In San Francisco, for example, the city and county saved approximately $3 million by not holding a separate runoff election in 2005, easily covering the mostly one-time costs spent in 2003-2004 to implement the system.

In North Carolina, counties spent $3.5 million for the Superintendent of Public Instruction runoff in 2004, at election with statewide turnout of only 3%. In 2007, IRV elections in Cary (NC) avoided the need for a runoff in one of the city council districts that would have cost taxpayers $28,000.

An effective voter education program can also be done for relatively little money by learning from what types of education worked well in other jurisdictions and what types did not – with the biggest factors being a good ballot design, clear voter instructions and effective pollworker training in that order. In a report to the Vermont General Assembly, the Vermont Secretary of State estimated that, based on how well IRV was implemented in Vermont’s largest city of Burlington in 2006, voter education for statewide IRV in Vermont would cost less than $0.25 per registered voter. In a city of more than 100,000 people, Cary spent less than $10,000 on all IRV implementation and voter education (saving the $24,000 cost of a separate runoff election in District B) – with highly favorable reactions from voters.

16. Dopp: "Increases the potential for undetectable vote fraud and erroneous vote counts…"

Actually, just the opposite is true, so long as paper ballots (such as optical scan) are used. The reason that any attempts at fraud are easier to detect with IRV is that there is a redundant electronic record (called a ballot image) of each ballot that can be matched one-to-one with the corresponding paper ballot. Cities such as San Francisco (CA) and Burlington (VT) release these ballot files so that any voter can do their own count. Without such redundant ballot records (which are not typical with vote-for-one elections) there is no way to know for certain if the paper ballots have been altered prior to a recount.

17. Dopp: "Violates many election fairness principles…."

This charge reveals either a general lack of understanding, or intentional miss-representation. Every single voting method ever devised must violate some "fairness principles" as some of these criteria are mutually exclusive. Dopp's example in appendix B of "Arrow's fairness condition" (the Pareto Improvement Criterion) completely misunderstands the criterion, and gives an example that has no relevance to it (and contrary to her implication, IRV complies with this criterion). IRV works essentially the same as a traditional runoff election to find a majority winner. When the field narrows to the two finalists in the final instant runoff count, the candidate with more support (ranked more favorably on more ballots) will always win. Some theoretical voting methods may satisfy some "fairness' criteria, such as monotonicity, but then violate other more important criteria such as the majority criterion, or the later-no-harm criterion.

18. Dopp: "Unstable and can be delicately sensitive to noise in the rankings…"

This point has some validity, but is of extremely minor significance. Whenever there is a close election with many candidates, regardless of the voting method, there is a chance that the ultimate winner may win due to the votes of the "most ill-informed voters." Holding a separate runoff as proposed by Dopp as a way of effectively excluding many voters (due to typically smaller turnout) is an anti-democratic approach.

No implementations of IRV in the U.S. have suggested mandating that voters rank all candidates, as is typical in Australia. It is rarely ever important that a voter "fill out a ballot ranking every candidate 10 deep." A study of the related single transferable voting method used in Cambridge (MA) found that in a race with over 20 candidates for nine seats on the city council, approximately 90% of voters saw their first or second choice elected. Dopp's final suggestion of "restricting the ranking depth" so as to ease the cognitive burden on voters and eliminate "noise" of low rankings has been implemented in many U.S. applications, though most experts agree it is preferable to give voters the option to rank as many candidates as they wish.


<1> The rank order ballot used in instant runoff voting (and other voting systems) is known by political scientists as the "single transferable vote" or STV. This balloting procedure has been consistently upheld in United States courts as constitutional and upholding the "one person, one vote" principle. As an example, here is what the Michigan Court ruled in upholding the use of instant runoff voting in an Ann Arbor, Michigan Mayoral race in a 1975 challenge:

"Under the "M.P.V. System" [IRV], however, no one person or voter has more than one effective vote for one office. No voter's vote can be counted more than once for the same candidate. In the final analysis, no voter is given greater weight in his or her vote over the vote of another voter, although to understand this does require a conceptual understanding of how the effect of a "M.P.V. System" is like that of a run-off election. The form of majority preferential voting employed in the City of Ann Arbor's election of its Mayor does not violate the one-man, one-vote mandate nor does it deprive anyone of equal protection rights under the Michigan or United States Constitutions."

page 11, Stephenson v Ann Arbor Board of City Canvassers File No. 75-10166 AW Michigan Circuit Court for the County of Jackson

The Judge also observed on page 7, "Each voter has the same right at the time he casts his or her ballot. Each voter has his or her ballot counted once in any count that determines whether one candidate has a majority of the votes. . . . Far better to have the People's will expressed more adequately in this fashion, than to wonder what would have been the results of a run-off election not provided for."