No CrossRef data available.
Published online by Cambridge University Press: 20 July 2006
We study the problem of designing a distributed voting schemefor electing a candidate that maximizes the preferences of a setof agents. We assume the preference of agent i for candidatej is a real number xi,j , and we do not makeany assumptions on the mechanism generating these preferences.We show simple randomized voting schemes guaranteeing theelection of a candidate whose expected total preference isnearly the highest among all candidates. The algorithmswe consider are designed so that each agent has to discloseonly a few bits of information from his preference table.Finally, in the important special case in which each agent isforced to vote for at most one candidate we show that ourvoting scheme is essentially optimal.