A different look at secure distributed computation

Abstract

We discuss various aspects of secure distributed computation and look at weakening both the goals of such computation and the assumed capabilities of adversaries. We present a new protocol for a conditional form of probabilistic coordination and present a model of secure distributed computation in which friendly and hostile nodes are represented in competing interwoven networks of nodes. It is suggested that reasoning about goals, risks, tradeoffs, etc. for this model be done in a game theoretic framework.