THE FATAL FLAW IN FRANCE'S - AND AMERICA'S -VOTING SYSTEM, AND HOW AN "INSTANT RUNOFF" SYSTEM MIGHT REMEDY IT

By AKHIL REED AMAR AND VIKRAM DAVID AMAR
Friday, May. 03, 2002

America is not the only great democracy with serious flaws in its presidential election system. Consider, for example, this disturbing fact about the French system: As French citizens go to the polls this weekend to elect their President, they will not be allowed to vote for the candidate whom more than a third of them-and perhaps even a majority-would prefer over the two candidates who are actually on the ballot.

Why Lionel Jospin, Despite Strong Popular Support, Is Not On the Ballot

The two men currently on the French ballot are President Jacques Chirac, the conservative incumbent running for re-election, and the ultra-rightist Jean-Marie Le Pen. The odd man out is Socialist Prime Minister Lionel Jospin.

Why, then, is Jospin's name not on the ballot? Because in a qualifying-heat election in late April, Jospin came in third in a field of sixteen candidates, a nose behind second-place finisher Le Pen. Chirac came in first in this initial round, with just under 20% of the votes cast, Le Pen got 17%, and Jospin drew 16%. Thirteen minor candidates split the remaining 47% of the vote.

But many of the voters within that 47% strongly preferred Jospin over Le Pen and Chirac; and only a few of those 47% preferred Le Pen. Thus, if the qualifying-heat ballot had been designed to elicit and honor these voter preferences, Chirac and Jospin would have entered the finals as the two most popular and credible candidates, with Le Pen a distant third.

How STV/Instant Runoff Systems Work

A system designed to honor the preferences of those who vote for minor candidates in precisely this way is called "single transferable voting" (STV) or "instant runoff," and we have mentioned it in passing in a previous column. Here's how it works: When a voter casts her ballot, she is free not merely to identify her first choice, but also to rank order the remaining candidates.

For example, if five candidates are running, a voter could vote for one as her first choice and then proceed to identify other candidates as her second, third, and fourth choices respectively. If our voter is indifferent between her low-ranking choices, she might simply rank her top two or three choices and stop there.

A simple illustration shows the difference between STV and cruder plurality-rule systems. Suppose that 4 candidates are running in a qualifying-heat election whose purpose is to identify the top two candidates, who will then square off in a final-round election held later. In this first heat, in which 100 persons vote, candidate A gets 35 votes, followed by B with 30 votes, C with 29 votes, and D far behind with 6 votes.

If we simply picked the top two initial vote-getters for the final round-a la France-A and B would be the two. But suppose voters were allowed to list their second choices, and all 6 who voted for D ranked C as their second choice. In this STV system, because D got the lowest number of first-place votes, he drops out, and his 6 votes are reassigned to the person ranked second on these 6 marked ballots-in our hypothetical, candidate C.

Now, A and C are the two top vote-getters, with 35 votes apiece, compared to only 30 for B. If only two candidates advance to the final round, the two should be A and C, not A and B, because C is actually more popular than B once we factor in the second choices of those who in effect "wasted" their initial vote on minor candidate D.

How STV Would Have Helped to Register Voters' Preferences for Jospin Over LePen

Indeed, STV could even eliminate the need for a separate final-round election between the top two vote-getters. (This is why STV is sometimes called an "instant runoff" system.) Just as D's initial 6 votes were reallocated because D came in 4th out of 4, so too candidate B's 30 votes could then be reallocated: he came out 3rd out of the three who remained after D's removal. (A and C, remember, are the top two candidates with 35 votes apiece after D drops out.)

With A and C tied for the lead, if more of B's 30 supporters ranked A higher than C on their list, A wins; if more ranked C higher, C wins. This is similar to what would happen in a separate runoff election between A and C held later-both A and C would woo B's electoral base, whose members would hold the balance in the final round. The main differences between STV and an actual runoff are that some voters might change their minds in the time between the first heat and the runoff; the runoff might be preceded by focused debates between and additional attention on the top two finalists; and different voters might show up on runoff day.

Why the French System is Relevant to American Elections

What, you might ask, does any of this have to do with American democracy? For starters, the problems created by ignoring second-choice preferences of voters who "wasted" their initial votes on fringe candidates are hardly unique to France, or to qualifying heats. These problems can also surface in virtually every American election.

Take the 2000 Presidential election. Because so many flaws were exposed at once-misleading ballots, voter fraud, widespread disfranchisement, primitive and unequal voting machinery, long lines at some precincts, badly drafted voting codes, counting and recounting delays, absentee ballot problems, judicial overreaching, and of course the inherent flaws of the electoral college-it is easy to overlook yet another quirk of our system. Call it the Nader effect.

Two years ago, Ralph Nader won almost three million votes nationwide, out of 105 million cast. That may not seem like much, but because the election was so close overall, there is a strong case to be made that Nader - as much as the butterfly ballot or the Supreme Court - cost Al Gore the election. Why? Because many Nader voters would have voted for Gore had Nader not been in the race, and the pickup of these Nader voters would have given Gore more states.

Exit polls indicate that roughly half the Nader voters would have preferred Gore to Bush, and only about 20% would have preferred Bush to Gore, while around 30% would have preferred to sit the election out if Nader weren't around. If STV had been used to reallocate Nader's votes in 2000, Al Gore would certainly have won Florida (where Nader got almost 100,000 votes, about 50 times the Bush margin of victory there), and may also have won New Hampshire, where Nader collected over 22,000 votes and Bush won by under 7,400.

Either one of these states would have given Gore an electoral college win. And if we forget the electoral college and focus instead on the national vote count, STV would have increased Gore's slim half-million vote margin over Bush to a comfortable million and a half.

Of course, had STV been in place in November, 2000, candidates in the preceding months would likely have waged different campaigns. For example, knowing that Nader would not be siphoning off many votes from Gore, Bush might have moved closer to the center and still won-though with a slightly different platform.

The Election of Abraham Lincoln in 1860 Viewed From the Perspective of STV

It's also fascinating to imagine about how STV might have played out in the fractured four-way presidential race of 1860. The Democrats in effect splintered over slavery-related issues and ended up splitting their votes among 3 different candidates-Stephen Douglas, John Breckinridge, and John Bell. Together, these three got more than 60% of the total national popular vote, and most of this 60% probably preferred Douglas over Abraham Lincoln.

But even if all of these anti-Lincoln votes had somehow been transferred to Douglas (or to any other anti-Lincoln candidate), Lincoln would, amazingly enough, still have won in the electoral college! In effect, he drew an inside straight by winning outright majorities in enough key Northern states to assure an electoral college victory.

STV's Possible Use in Congressional and State Elections

STV is a system that could be used not just for presidential elections, but for all sorts of other races as well-state governorships, congressional seats, state assemblies, and so on.

In 1998 Jesse Ventura won the Minnesota governorship with a bare plurality in a three-way race-- 37% compared to 34% for Republican Norm Coleman and 28% for Democrat Skip Humphrey. This would not have been enough to ensure victory under STV; rather, the third-place Humphrey would have dropped out and the Humphrey ballots would have been reallocated to the second-choice candidates indicated.

When forced to choose between Ventura and Coleman, whom would Humphrey supporters have picked? Ventura would have won only if a very good chunk-almost half-of non-indifferent Humphrey voters preferred him to Coleman.

Objections to An STV/Instant Runoff System and Possible Responses

STV is not without problems. It might promote fringe candidates, who might get more initial first-choice votes if supporters knew that their votes would not be wholly wasted, but quickly reassigned to more major candidates. Some would consider this a bad thing. But at the end of the day, STV for a single office like a governorship would generally make it harder for a fringe candidate to win with a simple plurality in a three-way race, a la Ventura.

STV elicits more information from voters about their deep preference structure, and then tries to tally votes in a way that honors that structure. When Americans go to the grocery store, they understand the second-choice concept: Get Ruffles, but if they are sold out, get Pringles. If Americans can handle this level of complexity as shoppers, why not as voters?

Voters would become more adept at understanding and using STV ballots if such ballots were used more prominently in private organizations-for elections in chess clubs, bowling leagues, homeowner associations, fraternal orders, and so on. Historically some urban areas have used variants of STV for city council or school board elections- including Cambridge, Cleveland, New York City and Sacramento-and the system has also been used at the national level in places like Ireland and Australia.

If a few states were to adopt STV for governors' elections, Americans in other states could see the system in action and decide whether to emulate it. This is how similar democratic innovations such as initiative, referendum, and direct election of senators took root in America.

The Founders' original version of the electoral college also featured a somewhat complicated ballot in which electors cast two votes for President. The framers imagined that most electors across America would cast one vote for a local favorite son-thus scattering votes widely-and a second vote for a more nationalist second-choice candidate.

As we have explained in an earlier column, this system was changed by the Twelfth Amendment adopted after the election of 1800-01, but it is a reminder that the eighteenth-century Americans understood some of the problems created by simple plurality-rule systems. It is a lesson that twenty-first century Americans-and Frenchmen too-would do well to ponder.


Akhil Reed Amar and Vikram David Amar are brothers who write about law. Akhil graduated from Yale College and Yale Law School, clerked for then-judge Stephen Breyer, and teaches at Yale Law School. Vikram graduated from U.C. Berkeley and Yale Law School, clerked for Judge William Norris and Justice Harry Blackmun, and teaches at U.C. Hastings College of Law. Their "brothers in law" column appears regularly in Writ, and they are also occasional contributors to publications such as the New York Times, the Los Angeles Times, and the Washington Post. Jointly and separately, they have published over one hundred law review articles and five books.

Ads by FindLaw