Previous: The Electoral College
Next: Instant Runoff Voting

Duverger's Law and the Libertarian Party

Peter A. Taylor
April 5, 2005 (cleaned up Sept. 7, 2006)

[An earlier version of this article appeared in The New Libertarian vol. 1 issue 2, May 1, 2005. I am indebted to Jeff Daiell for his constructive criticism of my earlier "Reasons for Supporting Spoilers" essay, which covers much of the same material.]

Sociologist Maurice Duverger1,2 wrote that the relationship between plurality elections (the type normally used in the US) and the two-party system comes as close as anything does in the field of sociology to being a natural law. The mechanical effect of small parties not being able to win seats in the legislature in proportion to their numbers is compounded by the psychological effect that voters do not like to support "spoilers," seeing their support wasted when voting for a different party could have made a difference. This tendency for plurality voting to "pulverize" third parties and produce a two-party system has come to be known as "Duverger's Law."

The long name for this voting system is Single Member Plurality (SMP). Candidates run in single-seat districts, and there are no runoffs, so only a plurality is needed to win. Single seat districts force serious parties to nominate centrists, and the lack of runoffs creates an "electoral straightjacket,"3 punishing voters for supporting any but the two largest parties. In contrast, runoffs allow more than two serious parties to compete if they run centrists. In further contrast, multi-seat districts allow Proportional Representation (PR), where fringe parties can win seats. If you like centrist parties but want more than two choices, you would probably like single-seat runoff systems such as Australia's "Instant Runoff Voting" (IRV)4. If you like parties that specialize in serving ideological minorities, you would probably like multi-seat districts with some form of PR5. Much of the opposition to IRV involves confusion between runoff elections and multi-seat districts.

Jon Henke observed in "The Marginal Futility of the Libertarian Party" that the Libertarian Party (LP) was dominated by activists who were ineffective because they refused to make necessary compromises. This is true, but even if it were not, the LP would still be ineffective because it has never really come to grips with Duverger's Law. How does a minor party deal with an election system that condemns its candidates to being spoilers? Here are some ways:

 1.  As Mr. Henke observed, the US Socialist Party6 got much of its platform enacted without ever winning significant numbers of seats. It served as a publicity device rather than as a party that was seriously trying to win seats. This approach makes sense if you think that the major parties do a good job of following public opinion, but that the public are ignorant.

 2.  Abraham Lincoln's Republican Party was able to survive the status of spoiler because of the rapid collapse of the Whig party. This scenario makes sense if you think that the public are well informed, but the major parties have become unresponsive to public opinion. The fact that we are talking about what is initially a two-party system further implies that either the major parties are roughly equally unresponsive, or that "brand loyalty" among voters is skewing the competition.

 3.  Britain's Liberal Democratic Party (formerly the Liberal Alliance) has used the threat of running spoilers against the Labour party in order to blackmail Labour into supporting electoral reform. This strategy especially makes sense if you think it is good for the country in the long term to have more than two serious parties competing in the political marketplace. The Liberal Democrats specifically favor a mild form of PR, but similar logic would apply to support for IRV.

 4.  Several Canadian political parties survive by being regional. While each province has a more or less two-party system, the country as a whole does not. It helps that Canada doesn't have presidential "coattails." Again, trying to establish a regional third party makes sense if the major parties are chronically unresponsive, at least locally.

If a minor party's supporters are geographically dispersed, it might help in becoming a major regional party to pick a region and have lots of their supporters move there. Some Libertarians are trying to do this in New Hampshire as part of The Free State Project. This appears to be prohibitively expensive. It is also dangerous because the participants are self-selected for zealotry.

Another idea that might help a minor party with geographically dispersed support to become a major regional party is vote trading. For example, Green voters in a swing state like Florida will make a gentlemen's agreement over the internet to vote Democratic in exchange for Democratic voters in safe states like Texas or Massachusetts agreeing to vote Green.

 5.  Another way a minor party can acheive its objectives without winning elections is suggested by the 1968 candidacy of George Wallace7. The Democrats and Republicans appeared to be competing closely for the median voter on the left-right spectrum, but there were (ahem) "cultural issues" in Alabama that did not fit on the left-right axis. The major parties had drifted away from the median voter on this "cultural" axis, and Wallace was able to "help" them find the median voter again. This scenario makes sense if you think the major parties are trying to be responsive, but don't know where the average voter stands. It is similar to the Lincoln scenario in that the major parties have drifted roughly equal distances away from the center, but here the major parties reacted quickly, and the minor party's appeal was frankly rather limited.

Figure 1
1968 Presidential Election
Sociological "Issues" 
median voter

 6.  Minor party activists often hope to use the threat of running spoilers in order to blackmail the more similar of the two major parties into adopting a position closer to that of the minor party. Normally, this is exactly the wrong thing to do. A competant politician such as Bill Clinton will respond to this with a "Sister Souljah moment," denouncing the fringe influence and moving away from it to court the voters in the center of what is left of the electorate after the minor party's supporters have effectively removed themselves from it. An incompetant politician such as Al Gore will move away from the center, towards the fringe, and thereby lose the election. Some Greens argued at the time that a more Green-friendly position would help mobilize the Democrats' base, making up in base turnout what they lost in centrist votes. But with plurality voting, Gore needed two votes from Nader supporters to make up for every one voter he lost to Bush, and Nader's influence also helped mobilize the Republican base.

|                       ^ | ^                       |
|               Democrats | Republicans             |

        <----- effective part of the spectrum ----->
| ^     |                 ^ | ^                     |
|Greens |         Democrats | Republicans           |

What made the Lincoln and Wallace scenarios unusual was that the nearest major party moved toward the center if it moved closer to the minor party. When this is true, the minor party doesn't need to engage in blackmail, and otherwise blackmail is counterproductive. The Liberal Democrats did something similar by temporarily focussing on an issue that doesn't fit on the left-right spectrum, but mostly they survive by being regionally concentrated.

 7.  A common response to the spoiler problem is denial: "It doesn't matter if we are spoilers because the major parties are equally bad, and besides, we draw votes from both of them." Typically this means that the minor party activist is pretending to be in the position of Lincoln or Wallace, when in fact, he is in the position of Sister Souljah.

 8.  Another form of denial is to claim that the spoiler problem is a self-fulfilling prophecy, and if you ignore it, it will go away. Unfortunately, minor party activists can't make the swing voters ignore it.

 9.  Maurice Duverger wrote that there were two kinds of political activists, one whose motto might be "Fight to win" and another whose motto might be "Fight to feel good about losing." Duverger's Law may not be a problem if your objective is self-expression rather than influence.

 10.  A minor party could focus on races that use runoffs, such as Mayor of Houston.

 11.  Major parties in Papua New Guinea8 manipulate plurality voting by encouraging spoilers to run against opposition parties. Perhaps the best thing Libertarians can do is to support the Green Party.

 12.  Give up and join one of the major parties. This makes sense either because you hope to be able to subvert the major party's propaganda machine or because you think you can move the party a substantial distance without hurting it's electability, as in the Wallace scenario.

Some of these strategies are obviously counter-productive. Others might make sense, depending on what the problem is that the Libertarian Party is intended to solve. So what exactly is the problem? Is the problem that the major parties are unresponsive to the median voter or is it that the median voter is unresponsive to libertarian ideas? Is the libertarian movement trying to educate people, or bust up a cartel?

I believe we have both problems, but that the purpose of the LP is primarily to bust up a cartel. As Maurice Duverger wrote, all political parties are inherently at least somewhat oligarchic. When only two viable parties can exist in stable equilibrium, there are likely to be many areas in which they are more responsive to special interests or to entrenched ideological groups than they are to centrist voters. Furthermore, these two parties set the public agendas and control powerful propaganda organs.

The best long term solution I see to this problem is to increase the number of viable political parties. I would like to see five of them. But the only way to do this is to change the election system. This goal should appeal to almost anyone, throughout the political landscape, who is dissatisfied with the current two-party system. Duverger's Law is a rare example of a political problem that has an actual technical solution, and it is too serious a problem for minor parties to ignore.


1. Maurice Duverger, Political Parties, their organization and activity in the modern state, Methuen, London, ISBN 0416683207

2. For an on-line article by Duverger, see: Maurice Duverger, "Factors in a Two-Party and Multiparty System," in Party Politics and Pressure Groups (New York: Thomas Y. Crowell, 1972), pp. 23-32.

3. The "electoral straightjacket" quote is from Arend Lijphart, Democracies: Patterns of Majoritarian & Consensus Government in 21 Countries, 1984.

4. The March, 2004 issue of Scientific American also had a nice article on single-seat election systems, advocating Condorcet's Method.

5. For arguments in favor of Proportional Representation, see The Center for Voting and Democracy (CVD).

6. See Milton and Rose Friedman, Free to Choose, Avon Books, 1981, ISBN 0380525488.

7. Richard McKenzie and Gordon Tullock discuss the 1968 Presidential campaign of George Wallace in chapter 10 of The New World of Economics, 3rd edition, Irwin, Inc., 1981, ISBN 0256024944.

8. The Papua New Guinea example is from The International IDEA Handbook of Electoral System Design, published by the Institute for Democracy and Electoral Assistance.

For more information relating to election systems, see the references at the bottom of my Instant Runoff Voting advocacy piece. I have also posted an election systems workshop curriculum (four 1.5 hr. sessions).

Previous: The Electoral College
Next: Instant Runoff Voting