IRV, sometimes referred to as ranked voting ( though actually IRV is only one possible form of ranked voting) has been used in elections around the world. India, Sri Lanka and Malta use IRV for presidential elections and IRV is used widely both in Australia and in Papua New Guinea. London and San Francisco have used it for mayoral races and it is used for a variety of races both in popular elections, by political parties and various other organizations around the world. It seems likely that the great interest in IRV is motivated by the understanding that it is a way to dodge the spoiler effect and also by the hope that using IRV will promote the participation of a greater number of political parties.
With IRV, voters go to the polls only once to cast a ballot that enumerates, in order of preference, the various candidates. To determine the winner of an IRV election the ballots are counted and recounted, with each successive count removing just one losing candidate from consideration (so in an election with N candidates, the ballots must be counted N-1 times). The loser in each count, determined using the FCV approach, is the candidate who is the first choice by the fewest voters is eliminated from the race.
Whenever a candidate has been eliminated through some earlier count, votes cast for the other candidates are promoted to fill the missing slot before the next re-count. So, for example, in counting a voter's ballot whose first two choices have been eliminated earlier, that voter's third-ranked candidate is promoted so as to be treated as the first-choice vote in the next re-count. At each stage of counting it is only the (remaining) top choice that is considered.
An IRV ballot is apt to seem ridiculously complicated to a voter accustomed only to FCV and there are reports that, due to the need for educating voters, the adoption of ranked voting can be expensive. There are indications as well (http://instantrunoff.blogspot.com/2009/06/truth-about-instant-runoff-voting-it.html) that adopting IRV may reduce voter participation. And the evidence is weak that IRV actually results in new parties or more candidates. Still, IRV has great appeal because it allows each voter to describe fully his or her preferences among the candidates for office.
But how easy on the voter is IRV? In a contest with even a half-dozen candidates running for an office, how many voters can make a clear decision about how to rank them all? How easy is it for a voter to construct such an ordered list and how ambivalent are voters about how to position some of the candidates?
It seems that when there are many candidates to choose from, a typical voter is apt to have strong feelings, positive or negative, about only a few of the candidates, some candidates seem about the same and the voter not have much of an opinion about some others; a given voter may never have even heard of some of the candidates. Forcing such a voter to construct an ordered list could lead to understandable frustration. Some positions in the will probably be placed there through the mental equivalent of a coin toss. The completed ballots and the elections results, might seem in some respects arbitrary or even misleading.
In fairness, IRV ballots usually do not require the voter to produce a complete list, but just to rank some candidates in order of preference. But there has to be a first entry on the list and a voter may feel that none of the candidates deserves to be placed at the top of the list. Lastly, one has to wonder how many voters will be unclear about how exactly their list, abbreviated or not, effects the election results? IRV is a complicated system and the concept of the counting is difficult to grasp.
What should the IRV voter to do who does not particularly like any of the candidates but is very strongly opposed to one of them? Negative feelings are common and often felt every bit as strongly by the voter as the positive ones. IRV is supposed to allow voters to fully express their opinions, but it doesn't seem to provide a way for a voter to express the very common feeling of dislike for a particular candidate. So what might an IRV voter try to express heartfelt opposition to a candidate? The voter might construct a list of candidates with the disliked one at the very bottom of the list, but surely that would have to be a complete list. With IRV, the voter can be assured that last candidate in the list of N will never receive help from the voter because is totally ignored in the counting because there are only N-1 iterative counts of the ballots.
IRV has no way to distinguish between a candidate that a voter actively dislikes and another about whom the voter is simply indifferent. The same is true with FCV, but the big virtue of FCV is its simplicity and FCV makes no pretense of allowing the voter to fully express his or her views.
In our next article we will show how IRV can be altered just slightly to make it balanced (any vote for a candidate can be balanced by a vote against that candidate). This will repair many of its problems and because it is balanced. this new system will surely make elections much less hostile to minority parties. It will not make the ballot shorter or the counting simpler; but because it allows the voters' opinions to be more directly expressed, it is likely to make voting less stressful.
Problems with IRV turn out to be a recurring topic in future articles in this series. It is re-visited in a future article, Isn't IRV a Great System for Voting and again in an article, Self-Expression vs. Effectiveness in Voting.