By Alan T. Sherman '87, Warren D. Smith '84, and Richard T. Carback III. This is a local copy of material that appeared in the September/October 2008 online edition of Technology Review.
Spoilers and Vote Splitting
As we write at the beginning of July, Libertarian Party candidate Bob Barr is greatly feared by Republicans, and gleefully anticipated by Democrats, as a potential spoiler for John McCain in the general election. In "A Candidate Runs to a G.O.P. Chorus of 'Don't,'" an article that appeared on the front page of the New York Times on June 28, Barr summed up Republican complaints this way: "They all said, 'Look, we understand why you're doing this. We agree with why you're doing it. But please don't do it.'" As Barack Obama's campaign manager put it, "If Barr were to get 2 percent in most states, our belief is he'll get 4 percent here [in Georgia], most of it coming out of McCain's hide."
And the spoiler phenomenon may already have affected the primaries. Before quitting the race Mitt Romney claimed he would have defeated McCain for the GOP nomination if Mike Huckabee had not been in the race. In the New Hampshire primary, the presence of John Edwards helped Hillary Clinton defeat Obama. Had Edwards campaigned a little longer and harder, Clinton might have won the Democratic nomination.
Similar dynamics have affected the primaries before. In 1964, Republican Barry Goldwater lost the biggest landslide in U.S. history, to Democrat Lyndon B. Johnson. A GOP vote split had handed the Republican nomination to Goldwater instead of Pennsylvania governor William Scranton. A study by William R. Keech concluded that Scranton would have defeated every Republican rival head to head--defeating Goldwater, specifically, 60 to 34. Our analysis of Gallup polls indicates that Scranton would have fared better against Johnson.
With range voting, vote splitting is extremely unlikely since giving a high score to one candidate never prevents giving a high score to another as well.
In Operation Chaos, Republican radio ranter Rush Limbaugh organized what he called "dittoheads" to infiltrate the Indiana and North Carolina Democratic primaries and vote for Clinton, who he believed would represent a smaller threat to McCain in the subsequent election. Limbaugh and the Obama campaign both claimed that was the reason Clinton won Indiana. This example illustrates a problem with the U.S. system of primaries followed by a general election: voters can have an incentive to vote for those they detest instead of those they prefer. The same thing can happen with runoff voting.
Although range voting in the primaries would not solve the problem of Republican misvoting in a Democratic primary, range voting reduces incentives for other forms of dishonest voting. In particular it is never worthwhile to score your favorite below the top, and in a three-way race it is never worthwhile to rank A>B if you honestly think B≥A.
"Top-Two" Domination and Removal of Voter Choice
Over time, both plurality voting and instant-runoff voting (IRV) lead to two-party domination, leaving voters with fewer choices.
The final two nominees in the current presidential race, McCain and Obama, both support legal immunity for telecom companies alleged to have wiretapped, recorded, and stored millions of phone calls and e-mails in America as part of a massive warrantless wiretapping government program. (See the Electronic Frontier Foundation lawsuit for details.) McCain and Obama also repeatedly voted to allocate money for the Iraq War. Polls indicate that both stances are opposed by large majorities of Americans. But with plurality voting those Americans will have no effective way to express their sentiments by voting for an anti-immunity or antiwar candidate. Instead, they can only choose the candidate they see as the lesser of two evils. This has happened before: in 2004 the majority of the U.S. public was antiwar but had no antiwar candidate to support in the general election.
In the 2007 Australian elections, which used IRV for house seats, no third-party members won a house seat, though rank-order votes indicate that nine Greens apparently would have defeated every rival head to head.
The reason for this pathology of plurality voting is simple: votes for anybody besides one of the top two are probably going to be wasted. Knowing that, voters won't risk voting for anyone other than those two candidates. By contrast, range voting encourages diversity because it reduces the penalties for expressing true feelings.
Early-Cash Dominance and Related Insanity
The "must vote for one of the top two" mentality makes it extremely important for a candidate to appear to be one of the top two. That's accomplished by acquiring and conspicuously spending a tremendous amount of cash up front. Early in the Republican primary campaign, Mitt Romney spent a million dollars to win an unofficial straw poll in Davenport, IA. In a logical system, anybody wasting so much money on such an irrelevant event would be judged an incompetent decision maker, unfit to be president. But by appearing to be a front-runner, Romney induced voters to vote for him and donors to donate to his campaign--at least for a while.
Romney's maneuvers didn't end up determining the Republican nomination, but two of the most qualified and experienced Democratic candidates, Joseph Biden and Christopher Dodd, dropped out of the race after a poor showing in only one (comparatively small) state, Iowa. In a logical system, it would have been insane to drop out so early. But since they were not among the top two, they had no hope of donations or votes.
As a result of these problems, which give such undue influence to early donors, most of the country had no opportunity to express its opinion about most of the Democratic nominees. Voters were able to choose only among the perceived front-runners. This may have denied the Democratic Party its objectively best nominee, and the United States its objectively best president.
The root of these problems is that if plurality voters think X is best but has no chance to win, then they won't vote for X, and he or she really won't win. But if range voters think X is the best candidate, X will win even if everybody thinks he or she has little chance. That's why range voting makes it much less important to look like a front-runner--and greatly reduces the influence of money.
For more information, visit the Center for Range Voting.
Warren D. Smith '84 cofounded the Center for Range Voting (http://rangevoting.org) in 2005. Alan T. Sherman, PhD '87, is an associate professor of computer science and a member of the National Center for the Study of Elections at the University of Maryland, Baltimore County. Richard T. Carback III is a PhD candidate in computer science at the University of Maryland, Baltimore County.
Copyright Technology Review 2008.