On distributed verification

This paper describes the invited talk given at the 8th International Conference on Distributed Computing and Networking (ICDCN 2006), at the Indian Institute of Technology Guwahati, India. This talk was intended to give a partial survey and to motivate further studies of distributed verification. To serve the purpose of motivating, we allow ourselves to speculate freely on the potential impact of such research. In the context of sequential computing, it is common to assume that the task of verifying a property of an object may be much easier than computing it (consider, for example, solving an NP-Complete problem versus verifying a witness). Extrapolating from the impact the separation of these two notions (computing and verifying) had in the context of sequential computing, the separation may prove to have a profound impact on the field of distributed computing as well. In addition, in the context of distributed computing, the motivation for the separation seems even stronger than in the centralized sequential case. In this paper we explain some motivations for specific definitions, survey some very related notions and their motivations in the literature, survey some examples for problems and solutions, and mention some additional general results such as general algorithmic methods and general lower bounds. Since this paper is mostly intended to "give a taste" rather than be a comprehensive survey, we apologize to authors of additional related papers that we did not mention or detailed.