From 1294 to 1621, forty-one "conclaves" (gatherings of an average of about forty cardinals to select the Pope) incorporated a voting methods called approval voting into their procedures. Among a large number of candidates, participants could cast equally weighted votes for more than one candidate. Repeated balloting would take place, with participants able to drop out of the race after a round of voting. Winners had to secure support from at least two-thirds of those casting ballots.
Repeated balloting captures what makes instant runoff voting work: the number of candidates declines and voters settle on other choices when their top choice drop out. But starting from scratch each time allows opportunities to change one's vote and zero in on consensus candidates. It takes time, but when you're all in the same room and have that time, it has real value.
The Church eventually abolished the approval voting element of the process in after the 1621 conclave. One reason apparently was some gaming of the system. In 1559, for example, a cardinal almost won when an ally confidentially met one-on-one with many participants and asked them to cast a "token" approval vote for his friend to avoid him being shut out -- instead the cardinal won votes on 17 of 32 ballots, close to what it took to win. But the biggest reason for the repeal apparently was that casting only one vote was seen as easier -- and guaranteed cardinals in each round voted on an equal basis. Repeated rounds of voting alone was enough to establish the greater consensus sought by Church leaders.
Approval voting continues to have a best mixed experience in real-life elections. This year it was repealed for Dartmouth Alumnni trustee elections by a vote of 82% to 18% of more than 10,000 alumni and in 2002 was repealed by its other most significant user, the Institute of Electrical and Electronics Engineer (the world's largest professional society), because, according to a December 2004 letter in the IEEE's newspaper The Institute, "80% of IEEE members were not using approval voting... also. their comments indicated that most disapproved of it; some even refused to vote because of it."
I in fact recommend approval voting in some situations -- for instance, when a hiring committee is trying to narrow down big fields of choices to a more manageable number. But as soon as the election becomes highly competitive and it is clear to some people that voting for more than one person counts against your top choice, it becomes highly problematic.
The biggest political problem for approval voting's advocates is trying to explain a very plausible result: a candidate with the first choice support of a majority of 51% or more might lose if supporters of that candidate mistakenly also give an approval vote to a lesser choice. Elected officials and charter commissions typically aren't keen on the idea of a clear majority choice losing. The onus is on its sometimes-zealous backers to show that such realities do not undercut the system's political potential in the United States.
The Colomer-McLean article shows that approval voting certainly is not a new idea. It had early uses as far back as the 13th century, but today instant runoff voting is used by far more professional societies and associations, including many of our nation's largest and most respectable ones in contested elections. Instant runoff voting is also used by more than 50 colleges and universities for student elections, while I don't know of any using approval voting. And of course instant runoff voting is used by millions of voters in governmental elections, while approval voting is not used for any such elections.
We plan to regularly report on elections held by other single winner voting methods -- and of course keep suggesting that when electing a legislature, policymakers should look first at forms of proportional voting in multi-seat districts like choice voting instead of any winner-take-all system.