Storable votes is a simple voting mechanism that endows voters with a stock of bonus votes and allows them to choose how many votes to cast over a series of binary decisions proposed over time. Each decision is taken according to the majority of votes cast. The mechanism induces voters to express the intensity of their preferences, and although not generally optimal has three main virtues. First, storable votes typically increase expected welfare relative to simple majority voting. Second, storable votes make it possible for a systematic minority to win occasionally, at small or no loss in aggregate efficiency. Finally, the mechanism has been found to perform well in laboratory experiments. This proposal discusses three new research projects on storable votes. First, the sensitivity of storable votes to possible manipulation of the agenda is not known. Could a voter "force" a welfare-reducing decision on the rest of the committee by proposing it when the others have exhausted their bonus votes? The answer is not clear because the agenda formation procedure becomes part of the game, altering all strategies. Second, how do storable votes compare to other voting mechanisms designed to elicit intensity of preferences - mechanisms like alternating dictatorship, or vote-buying, or the asymptotically efficient Jackson and Sonnenschein mechanism? Both theoretical and experimental research on these questions is proposed. (These parts of the proposal are joint with Thomas Palfrey). Third, a particularly simple, static, implementation of storable votes in large populations is to grant voters one extra, vote when voting over a bundle of simultaneous referenda. Theoretical results have been derived in the past. Here we propose to evaluate them through an internet-based experiment. (This part is joint with Andrew Gelman and Duncan Watts). Finally, the proposal discusses an unrelated but fascinating historical example of votebuying: the 1519 election of Charles V to emperor of the Holy Roman Empire. In addition to the extraordinary availability of data on the exact contracts that Charles and its opponent, Francois I, the king of France, proposed to the seven electors, the episode is remarkable for the sophistication of the agreements: the electors voted sequentially according to a codified order and conditional contracts were written where obligations depended on the behavior of the other electors. The episode shares some features of the storable votes model, in particular the sequential auction structure. (This part of the proposal is joint with Francois Velde).
Broader Impact. The interest in voting rules is at the moment very strong in policy circles, in fact stronger than in academia. It has been kindled by a number of very visible events: the 2000 US presidential election, the negative referenda on the Constitution in the European Union in the Spring of 2005, the halting discussion of reform of the Security Council of the United Nations. A book on storable votes is under contract with Oxford University Press, and large parts of it should be of general interest, but a more public discussion requires some answers to the issues raised in this proposal. Because the projects combine very different methodologies (economic theory, experimental work, numerical exercises, online experiments, and historical research) they are also ideal teaching tools, able to employ students with different talents and at different stages of the their career.
