It’s heartening to know, as important elections approach, that ours may be the worst of all the voting systems in common use. That’s the takeaway from Gaming the Vote: Why Elections Aren’t Fair (and What We Can Do About It), one of a dozen books penned by William Poundstone, a professional skeptic who studied physics at the Massachusetts Institute of Technology before he began pumping out nonfiction in 1982.
Poundstone is particularly fascinated with how scientific ideas—mathematics in this case—play out in everyday life. His 2005 book Fortune’s Formula: The Untold Story of the Scientific Betting System That Beat the Casinos and Wall Street was hailed by the New York Times as perhaps “the world’s first history book, gambling primer, mathematics text, economics manual, personal finance guide and joke book in a single volume.”
He became interested in voting theory after reading about Arrow’s Impossibility Theorem, the work of economist Kenneth Arrow, now a professor emeritus at Stanford University. Arrow identified what he perceived as a fundamental flaw in our democracy: Put simply, he said, devising a perfectly fair voting system is mathematically impossible.
Gaming the Vote entertainingly probes the combative history of voting over the past few centuries. It covers unusual territory, such as Alice’s Adventures in Wonderland author Charles Dodgson’s (a.k.a. Lewis Carroll’s) obsession with voting, the legacy of sleazy campaign tactics spawned by GOP political consultant Lee Atwater, and how the idiosyncrasies of our election system left Louisiana voters to choose between a notoriously corrupt liberal and a former Ku Klux Klan leader for governor. (The crook won.) I caught up with Poundstone for a primer on voting methods.
Is there a way around Arrow’s Impossibility Theorem?
For decades, there was almost a kind of despair among voting theorists of getting any better system than we had. What’s interesting, though, is that the impossibility theorem doesn’t apply to systems where you score the candidates rather than rank them. With scoring, you’re essentially filling out a report card—if you think there are two candidates who deserve four stars you can give them both four stars—whereas with ranking you have to artificially give one a number one and one a number two. That turns out to be crucial.
And yet plurality voting—where a person can vote for only one candidate for a particular office—is the most common system in use. What’s wrong with it?
Whenever you have two candidates whose support overlaps, that’s bad for both of those candidates, the obvious example being Nader and Gore in 2000. So a candidate can be a spoiler and cause the second most popular candidate to win. This is something that’s been appreciated at least going back to the 18th century, and people have tried to devise different ways of dealing with it, but for a very long time this was one of those unsolvable problems.
Do you consider plurality voting undemocratic?
Anything that tries to allow people to make a decision collectively would qualify as democratic. I think, rather, it’s bad technology. About 11 percent of the time we have a spoiler and get the second most popular candidate, so we really want to do better.
How does plurality voting skew outcomes in the primaries?
It’s even worse in the primaries where you might have three or more strong candidates. If someone likes Huckabee and Romney, they can only vote for one, so they’re basically getting penalized.
What are some other instances in which a spoiler has thrown the presidential election?
The famous one was the 1912 election. You had William Howard Taft and Teddy Roosevelt each trying to get the Republican nomination, and of course at that time Roosevelt was maybe the most popular ex-president ever. Taft was an incumbent and had an obvious lead, but there was bad blood between them. When Roosevelt didn’t get the Republican nomination, he ran on his own as the Bull Moose candidate, and although the two collectively got more than 50 percent of the vote, they split the Republican vote and instead you had the Democrat Woodrow Wilson winning.
How have political consultants exacerbated the spoiler problem?
Just in the past few years, they’ve started promoting the campaigns of spoilers they think are going to benefit their candidate. The Republicans were paying for Nader signature drives in 2004, but since then it’s become thoroughly bipartisan and you’ve had Democrats paying for radio ads for Libertarian or anti-immigrant spoilers and vice versa, so it has kind of become the new technique.
They’ve learned that the spoiler effect can be profitable, because it’s actually more cost effective in many cases to divert some of your candidate’s money to a spoiler candidate than buy additional ads for your candidate.
If spoilers create such uncertainty, why don’t the major parties support a method that eliminates the problem?
I think they’re just so used to the current system. The people in politics have a certain skill set that’s geared to this. They’ve even discovered how in some cases they can make use of it, so there isn’t quite as much motivation for [change], unless there would be an overwhelming popular outcry that this is something we want, and there hasn’t really been that.
What does the Constitution say about how we elect people?
It’s kind of funny. When we first wrote the Constitution there’d been a lot of thought in revolutionary France about what’s the best way to vote, and they basically discovered the spoiler effect and a lot of these problems Arrow was addressing. And because of that, the founders really didn’t guarantee anyone the right to vote for president or Congress or anything in the Constitution. The democracy we have now is kind of a retrofit.
Which alternative methods have been tried in the United States?
Instant runoff [ranked-choice] voting has been used in San Francisco, Minneapolis, and they’re phasing it in in the state of North Carolina. That’s where you rank the candidates and if your first choice is someone who is not one of the front-runners, your vote is basically transferred to the more preferable of those two front-runners. Everyone gets to honestly say who they really like, but your vote also counts where it really needs to count—in the crucial matchup.
Briefly summarize the pros and cons of the various voting methods you cover in your book. Let’s start with our current system.
Plurality voting is the simplest system possible because each person casts one vote for one candidate, so it’s very easy to count votes and so forth. The con, the one thing on which all the experts basically agree, is that plurality is the least fair of all the systems.
How about the Borda system?
The first modern system invented to try and better plurality voting was the Borda count, invented in 18th-century France by a guy named Jean-Charles de Borda. Unfortunately, it’s very easily manipulated. It’s used in sports a lot—for determining the Heisman Trophy, the most valuable player—where they basically take a poll or a ballot of sportswriters to find the best players. Unfortunately, if you have a player you like and rank him high, you might want to rank his rival at the very bottom of your list to penalize that rival. In fact, there’ve been scandals in sports where they do that. Borda basically said, “My system is intended for honest men,” and people weren’t entirely honest.
Condorcet was Borda’s great rival. He had a system where you rank the candidates, but the idea is that whichever candidate can beat all the other candidates in two-way races should be the winner. The main problem is sometimes you don’t have a candidate who beats everyone else. You can actually have a very weird situation where candidate A beats B, B beats C, and C beats A. It is also possible to manipulate.
There’s also “approval” voting.
It’s very simple. It basically uses the same ballot we have now except, if you want, you can vote for more than one candidate, and whichever one gets the most votes tends to win. I’m not sure there’s any real known disadvantage. Approval is actually used in the U.N. for voting on the secretary general, and it was used in Renaissance Venice for 500 years and apparently it worked pretty well there.
Ranked-choice, or instant-runoff voting?
The voters rank the candidates based on how much they like them. It works quite well as long as you have what we might call a typical American election, where there’s a Republican and a Democrat and you’re sure one of those two is going to win. The problem is when you have three or more strong candidates as you would in the race for a party’s nomination, then it’s subject to some of the same vote-splitting effects as the plurality vote.
You seem to favor this system called range voting.
Range voting is the newest in the sense of people being aware of it and promoting it: If you’re rating a video on YouTube you give it one to five stars, and they take that information and show you the average score of all the people who bother to rate it. We use it with a report card. The valedictorian of a school is the winner of a range vote by the teachers for each of their classes. In the Olympics, they hold up those cards to rate someone’s performance—that’s another example. People are pretty familiar with the idea. Nobody has given a convincing argument that there’s anything seriously wrong with it—the one thing you sometimes hear is it’s complicated, but that’s about it.
In 2000, mathematician Warren Smith published a study where he ran simulations to determine which of the common voting methods gave the most satisfactory, or least regrettable, outcome for the greatest number of voters. He found that range voting was the most fair. That one paper has convinced a lot of people that we ought to be taking range voting pretty seriously.
How did the other methods rank?
The second best was approval voting, which is the short-form version of range voting. Instead of rating someone on one to five stars, or one to ten, you basically have two ratings—thumbs down and thumbs up—and it’s almost as good. Next is the Borda count, although this particular simulation doesn’t factor in that there’s an incentive to manipulate the vote, so I wouldn’t rate that as too great an endorsement—Condorcet voting is a little better. Then you get into instant runoff and then plurality voting, which is the worst of all these systems.
What if we had adopted range voting in 2000 or 2004?
It’s pretty clear Gore would have won Florida and New Hampshire, so Gore would have been the president. Bush’s victory over Kerry in most of the states was less than the Nader effect, so you still would have had a Bush victory.
How might it have affected the congressional races?
Libertarian spoiler Stan Jones was basically responsible for Democrat Jon Tester winning the Senate seat in Montana. That was the 51st Senate seat, so you can say this Libertarian spoiler was responsible for giving Democrats control of the Senate in 2006. If an extraterrestrial came to Earth and asked me to explain our political system, it would be tough to justify both our president and control of the Senate. They’re each kind of the opposite party that they should be.
Why should we trust a voting method most notably used by the website Hot or Not?
Hot or Not uses range voting but they invented it entirely independently, which actually isn’t so unusual. Founder James Hong explained that the reason they adopted this is they wanted it to be as simple as possible to look at these two pictures and decide which of them was hot or sexy or whatever. One of the ideas was that you would be asked to say which one you prefer, but often that was difficult if you saw two people who looked about equally hot. Another possibility is you’d look at one picture and decide hot or not, and that, too, could be very difficult—if someone was in the middle, people wouldn’t know what to do. But with the range-voting thing you just move this cursor wherever it feels right; if someone’s about a seven, you just move your cursor over to seven and click. And they did studies and found the range-voting idea was actually the fastest and easiest of the various solutions.
If range voting is superior, why hasn’t it been used in a real election?
People didn’t really have the idea this is something we should be trying in elections until 2000 and Smith’s study. So it’ll be interesting to see how it plays out.
Of the alternatives, instant-runoff voting has the most momentum, and yet it’s flawed. Who’s pushing it, and why not range voting instead?
The main organization is FairVote, founded by an activist named Rob Richie in 1992. They’re involved in a lot of voting reforms—a direct popular vote for president, security of electronic voting, proportional representation—but one of the things they’ve done very successfully is instant runoff voting. When they started, no one was talking about range voting, so they kind of got locked into this, I think. Richie doesn’t seem to believe range voting would be practical—I’m not terribly impressed with his arguments against it. But he has done a lot of good just in getting people aware there are other ways to vote. And he has been successful in getting IRV, which is definitely an improvement on what we’ve got now. I would like to see a few communities try range voting and see how that plays out, and get more of an idea of how both of these work in modern campaigns.
Have any politicians stepped up and said we should try this?
Not range voting. They certainly have for instant runoff voting; Barack Obama and John McCain have both endorsed IRV, and Howard Dean as well.
Which method most benefits the small parties?
Anything that addresses the spoiler effect. The Greens and Libertarians would probably get many more first-place votes or high scores than they do now. I think that would have the effect of legitimizing them and making it a little easier for them to raise money. Once you legitimize them, there would start to be elections in some areas of the country where they would be able to win a few races.
Are some of these systems a no-go simply because they will confuse voters, or because they’d be too difficult to implement or sell to the public?
Either you have a ranked ballot or a report card, scoring-type ballot, and I think both of those are really pretty easy. If they get complicated, it’s in how you count the ballots. Some are more complicated than others, but that’s for the vote counters—the public doesn’t have to worry too much about that.
But couldn’t a more-complicated method create all sorts of havoc after an election? We have enough problems with our current system, where it’s supposed to be easy to count.
There is a concern about IRV in that, unlike all of the other systems, you have to transmit all the votes to a central repository—you can’t count them at the precinct level the way the other systems can. Some people are concerned about that. My own feeling is that it’s more a question of what is fairest and what you can actually sell to the public.
We’re still in a situation where the people in politics really don’t understand the various systems—it would almost have to be that the public would start having an outcry saying we want this, that, or the other thing. And that’s where FairVote has made a very useful contribution to the conversation, because they’ve managed to get people in some cities quite interested in the spoiler effect and what we can do to change it.
Do you think America will ever scrap plurality for something better?
In a generation or two I think it is very possible.
You’d think 2000 would have been the wellspring of such a movement.
I spoke with [New York University political scientist] Steve Brams, who is one of the co-inventors of approval voting, and that’s basically what he said. He figured if 2000 didn’t get people interested in changing our voting system, what would? And he just kind of shrugged.