Distributed Computing with Payout

Abstract:
The recent successes of a number of nonprofit computing projects distributed over the Internet has generated intense interest in the potential commercial applications of distributed computing. In a commercial setting, where participants might be paid for their contributions, it is crucial to define a security framework to address the threat of cheating and offer guarantees that the computation has been correctly executed. This paper defines and analyzes such a security framework predicated on the assumption that participants are motivated by financial gain. We propose a scheme which deters participants from claiming credit for work they have not done, and puts a high cost on attempts to disrupt the computation. We achieve these two goals by integrating an algorithm to assign computations to participants, an algorithm to verify their work, and an algorithm to pay participants.