BFI NewsMay 04, 2016

Disrupting the Tyranny of the Majority

Scholars mull theory and applications of a quadratic voting system


One man, one vote is the very bedrock of democratic systems. For centuries it is has been relied upon to deliver the fairest outcomes in a manner acceptable to everyone involved. But there is perhaps a more efficient, and even a more equitable, way of voting. In April, the Becker Friedman Institute hosted “Quadratic Voting and the Public Good,” a two-day conference that explored a new method of voting that could be used in both corporate and government settings.

“Tyranny of the majority is a rampant issue in all sorts of contexts,” noted conference organizer Glen Weyl. “Allowing minorities who are being severely hurt to express this can prevent manipulation of the majority rule.”

Weyl conceived quadratic vote buying as a fairer way to deal with issues of eminent domain, where the government can force individuals to sell their property as part of a public project. But after talking with Eric Posner they realized its potential was much larger. “It is a voting procedure that that can be applied to any kind of group decision-making, whether political or commercial or something else,” said Posner, who co-organized the conference with Weyl. “My view is that quadratic voting is the most important idea from economics in at least a decade, probably much longer.”

“My view is that quadratic voting is the most important idea from economics in at least a decade, probably much longer.”

In corporate decision making, shareholders get one vote per share and the majority rules. With quadratic voting, shareholders would have the opportunity to purchase votes—as many as they want—regardless of the number of shares they own.

However, buying votes becomes increasingly expensive it becomes because shareholders pay the square of the number of votes they wish to cast. Thus, while one vote will only cost a dollar, 10 votes will cost $100 and 100 votes will cost $10,000. The thinking is that shareholders will only invest large sums to vote on issues they really care about, and smaller or no sums in issues in which they have no interest. Plus, after the vote, the amount of money gathered for vote buying is divided equitably among all the shareholders, so even those who lost in the vote get something.

In a political setting, QV would be modified and money would not be used. Instead, each voter might be given 100 votes that can be used in 10 different elections.  Each voter would be to divide their votes up any way they wish among those elections depending on their interest in that particular vote, and would thus be able to put more power behind votes on issues they really care about.

Quadratic voting naturally brings with it many questions that require further analysis from a variety of experts. Weyl and Posner invited historians, political scientists, philosophers, law scholars and computer scientists to discuss where and how quadratic voting can or should be used.

Benjamin Laurence and Itai Sher presented their thoughts on the ethical issues that QV raises. From a utilitarian perspective, they acknowledged that if it is practiced without the use of vote purchasing, it may actually be superior to majority voting. However, they had trouble measuring efficiency or welfare in such a situation. They concluded by arguing that “in the presence of inequalities of wealth, any vote-buying mechanism, including quadratic voting,” will have a difficult time meeting any legitimacy requirement.

Of course, as a University of Chicago conference, those presenting often learned as much from their audiences as did the listeners. Jonathan Masur received strong feedback on his paper, “Quadratic Patent Policy,” in which he argued that the patent system is in crisis because the innovation economy is too complex to be regulated or controlled by existing institutions.

Masur instead suggested the use of quadratic voting as a method to solve intellectual policy and other legal issues. For example, in determining whether a specific type of DNA can be patented at all, the Patent Office could administer a vote.  Those interested in participating could buy as many votes as they like to support one side of the question and after a period of time, say a month, the votes would be tallied, the Patent Office could announce whether the DNA could be patented and distribute the money from the voting buying to all those who had voted.

Overall, the conference provided an opportunity “to engage in a common dialogue centered around a coherent set of issues,” Masur observed. “The event could only strengthen and improve quadratic voting as a decision-making procedure.”

Beatrice Cherrier and Jean-Baptiste Fleury provided background for discussion by presenting their paper, “Economists’ Interest in Collective Decision After World War II: A History.” Their research shows that while decision making piqued the interest of economists during the second World War, work in the field was sidelined almost completely from the 1970s forward.

Some in the audience argued that this happened because economists realized that the practical tools for collective decision making failed to transform policy making, and that policy makers were not convinced by economists’ analyses. Still others argued that that collective decision making had simply moved into other academic areas, such as business and law schools. “Whatever the causes, reluctant interest in collective decision mechanisms is what made this conference unusual. The back and forth between ideal types of the mechanism and concrete applications is not often seen at conferences and here it was very lively,” Cherrier said.

“Whatever the causes, reluctant interest in collective decision mechanisms is what made this conference unusual. The back and forth between ideal types of the mechanism and concrete applications is not often seen at conferences and here it was very lively,” Cherrier said.

Posner and Nicholas Stephanopoulos presented their research on the electoral system in the United States, arguing that it fails to allow people to register the intensity of their preferences for political outcomes when they vote. They suggest that modified quadratic voting, in which votes are not purchased, would allow election law to achieve efficient election and policy outcomes.

In a related vein, Jingjing Zhang presented research demonstrating the inability of elections to achieve socially optimal outcomes. Bidding mechanisms such as quadratic voting yield higher overall welfare and better outcomes for all voters. Many would argue that in such bidding situations, those who care deeply will try to buy the election. But Zhang argued that buying an election is typically very costly and requires redistributions that benefit all who vote.

Just as the organizers had hoped, excited discussion took place between sessions on improving results, future collaborations, ideas for practical implementation and new problems to solve. The conference papers—which also include work on security, voter turnout and efficiency—will be published in Public Choice, an economics journal that 40 years earlier examined the major intellectual predecessor of QV, the Vickrey-Clarke-Groves mechanism. But with so many obvious applications and so many theoretical and practical challenges surrounding it, the work on quadratic voting has only just begun.

— Robin I. Mordfin