Reply to an IRV Advocate

The following is a reply by Russ Paielli to claims made by a prominent advocate of Instant Runoff Voting (IRV) and proportional representation (PR). He refers to the Condorcet election method as "pairwise" voting. His claims are in quotes.

"As a whole, while single-winner election system reform is certainly important, it is much less important than making the move to Proportional Representation."

Consider the structure of the US federal government, which consists of three branches: executive, legislative, and judicial. The executive branch obviously has a single winner. And that winner, the President, nominates the members of the Supreme Court (although Congress can reject those nominations, it has no power to put its own preferred judges on the court). Two of the three branches of the federal government are therefore essentially controlled by single-winner elections. Of the two houses of Congress, furthermore, PR may only be practical for the House of Representatives. That leaves only half of one of the three branches that is not controlled by single-winner elections. Consider also that single-winner reform would affect even the House of Representatives. Even without PR, therefore, the House will still be more representative than it is under the current system, diminishing somewhat the need for PR. PR is certainly a worthwhile goal, but it is less important than single-winner electoral reform.

"The leadership of the PR movement is not terribly interested in spending a lot of time debating Pairwise vs IRV."

If IRV were clearly superior or even comparable to the Condorcet pairwise method, the statement above might be understandable, but it is not. In fact, many election system experts believe that IRV is clearly inferior to Condorcet. It is unfortunate that the leadership of the PR movement seems to have prematurely closed its mind on this crucially important issue.

"We've pretty much decided to push IRV, because it is equal to Pairwise on theoretical grounds, and superior on pragmatic grounds."

The Condorcet pairwise method beats IRV on several important theoretical criteria, but I don't know of any theoretical criterion for which IRV beats Condorcet, nor can I think of any pragmatic advantage of IRV. As explained above, IRV is very good at preventing minor parties from disrupting the two-party system, but only if the minor parties never become truly competitive. As soon as they become competitive, the entire political system could become chaotic under IRV.

IRV has a very significant practical disadvantage when it comes to counting the votes. IRV requires every ranked ballot to be available at a central location, which can be a logistical and security nightmare in a large election. Consider the US presidential election as an example (and assume that the electoral college has been abolished, as it should have been long ago). Suppose that 10 candidates are running and 100 million ranked ballots are cast. With IRV, all 100 million ranked ballots must be sent to a national center and stored there (or the counts must be sent one round at a time, which is complicated). That is, you cannot even start counting until you have the data from every single separate ranked ballot available, because it is impossible to know in advance which candidates will be eliminated and how the votes will transfer. The transfer and storage of that much data is certainly possible with modern technology, but it is much more complex and prone to security problems.

Under Condorcet, on the other hand, the votes of a race with 10 candidates can be reduced to a 10x10 pairwise matrix. Each precinct can determine its pairwise matrix and send it off to a regional center. Each regional center can then determine its 10x10 matrix (by adding all the precinct matrices) and ship it off to the county. The process continues until the overall national 10x10 matrix is determined at the national center. The winner is then determined according to the rules of the Condorcet system. The total amount of data transfer is drastically less than for IRV, and verification and security are drastically simpler.

"Pairwise has a bit of a problem with handling circular ties."

The ambiguity of a circular tie is a true ambiguity in the preferences of the electorate. The fact that Condorcet accurately recognizes such an ambiguity is not a deficiency, but rather a strength. Rather than ignore it and choose the winner capriciously, Condorcet allows the ambiguity to be resolved rationally and fairly.

"Note that IRV winners will tend to be those with enthusiastic support, while Pairwise winners will tend to be compromise candidacies."

That's one way to put it. Another way to put it is that IRV can allow a relatively small group of zealots to prevail over the consensus of the majority. The enthusiastic support of a minority of voters should not be enough to elect a candidate unless some degree of the enthusiasm spreads to a majority, and if it does it will be reflected perfectly accurately in Condorcet voting.

"Pairwise has a potentially fatal flaw: Since a voter's second choice could easily help defeat their favorite candidate, campaigns might very well ask their voters to bullet vote. This is reality - it killed the Bucklin system."

It is misleading to say that "a voter's second choice could easily help defeat their favorite candidate." It can only happen in the case of a cyclic ambiguity in which the majority of the other voters rank that voter's second choice over his first choice. But if a voter votes on the assumption that such a situation will occur, he is more likely to hurt than help his own cause. In any case, the Bucklin system is completely irrelevant.

"IRV has a huge advantage over pairwise from the point of view of PR activists. When we write up ballot language for STV [single transferable vote], we can use exactly the same language for IRV."

This claim is both misleading and revealing. It is revealing because it indicates that PR advocates may biased toward IRV for reasons that have nothing to do with a desire for single-winner reform. It is misleading because the Condorcet single-winner method could be adapted to the multi-winner case just as easily as IRV can, by simply applying it repeatedly, starting over after each winner is "eliminated" from the counting (or, even better, by using Tideman' method, which is discussed elsewhere at this site). The result may not be exactly proportional representation, but neither is the result when IRV is adapted to PR in the form of STV, as suggested.

"IRV is tested and proven in public and non-governmental elections. Pairwise, to my knowledge, has never been used in a public election, and is very rare in non-governmental elections."

IRV has certainly been tested, but has it passed or failed the test? Australia, where IRV has been used since around 1920, still has a predominantly two-party system, and parties rarely field more than a single candidate in a general election for fear of splitting the vote. And the fact that Condorcet has never been used in public elections is hardly a reason to preclude it from consideration. Where would we be today if such "reasoning" had prevailed against voting itself when first proposed?

"We need to remember that voters typically have a strong preference for one candidate. They are not dispassionate machines simply ranking preferences. Pairwise advocates tend to forget that if I rank the candidates, B, A, C, D, that it might very well mean that if I had one dollar to spend to help each candidate that I might give 90 cents to B, 10 cents to A, and zero cents to C and D."

First of all, the premise is not necessarily true. Campaign workers and other fierce partisans usually have a strong preference for one particular candidate, but I suspect that most voters are not so devoted to a single candidate--unless they believe that a particular candidate is their "only hope," which is usually the case in our current plurality system.

Secondly, the mathematics of ranked balloting are such that it hardly matters how much a voter prefers one candidate over another. Only the order matters, as it should. After all, if only two candidates were running, would it matter how much a particular voter preferred one over the other? Suppose unsure voters could give 6/10 of their vote to one candidate and 4/10 the other. That would be equivalent to reducing the weight of their vote to only 1/5 of a vote. How many voters would voluntarily reduce the weight of their vote by a factor of 5, regardless of how unsure they were? Not many, I suspect.

"I strongly suspect that in the real world, the Condorcet winner and the IRV winner would be the same person about 90% of the time. So realistically, is the difference really worth much? I doubt it."

Even if it is true, this conjecture misses a key point. The actual results of elections are only the most visible effect of an election method--the "tip of the iceberg." Less visible but just as important is the overall effect of an election method on the entire political system, because that determines what choices the voters have in elections. I strongly suspect that, in the real world, an IRV system would perpetuate the two-party duopoly indefinitely, whereas Condorcet could actually promote a genuine multi-party system.

"Note that anyone can come up with scenarios which favors their position, in any debate."

That's why we have objective mathematical criteria, many of which favor Condorcet, and none of which favor IRV.

"I can not cite the source, but it has been shown mathematically that in a public election it is virtually impossible for a voter to vote strategically in any useful manner when using IRV."

This claim is both wrong and misleading. First of all, unless or until a third party seriously threatens the two major parties, it is very easy to vote strategically in an IRV system. In fact, it is identical to voting strategically in the current plurality system. If a third party ever does threaten the two major parties, then it will indeed become very difficult to vote strategically under IRV. However, this difficulty is hardly an advantage. It is, in fact, a huge disadvantage. Voters would be in a terrible predicament in which they desperately need to vote defensively to oppose their worst political adversaries, yet cannot figure out how to do so. The entire political system could be thrown into chaos, and parties considered unacceptable by a large majority could get elected.

"Pairwise advocates often throw up all kinds of IRV horror stories and suggestions of strategic voting. But in reality, none of this has ever happened, despite IRV's extensive use in Australia and Ireland, much less non-governmental elections."

The problem with Australia is that it still has a predominantly two-party system even though IRV has been in use since around 1920. Australian voters still routinely vote strategically for the "lesser of two evils." So what good has IRV done? In Ireland, on the other hand, it is used for a largely ceremonial office.

The horror stories begin if and when a third party seriously challenges the two major parties, which has not yet really happened under IRV. As explained, voters would then be in a dilemma in which they desperately need to vote strategically but are unable to figure out how to do so.

"We need to avoid the mistake of thinking that voters think like we do. The typical voter is anything but a mathematician and election system analyst."

If voters are unable to figure out the best vote in support of their principles and wishes, they will be informed by experts. The fact that sincere voting is likely to backfire under IRV will not stay secret for long.

"One potential ugly result with pairwise could occur. A virtual unknown rogue candidate could insert themselves into what is effectively a two-person race, position himself as the middle candidate, and win."

If the voters vote for an "unknown" candidate simply to defeat a candidate they're sure they dislike, then they are taking a calculated risk. That may or may not be wise, but it is surely the right of voters to do so--or do IRV advocates think that election methods should prevent voters from making unwise choices? If so, perhaps we should eliminate voting altogether and just let the wise advocates of IRV select our leaders!