Suppose that the employees of a firm wish to determine their median salary without revealing to one another their individual salary. This would have been easy if they can totally trust some party to gather the information, conduct the correct computation, send its results to all of them, and reveal nothing else to anybody.
In general, the objective of secure multi-party protocols is to maintain a desired functionality (e.g., compute the median value) also when some of the parties deviate adversarially from their prescribed behavior. The desired functionality can be captured as an ideal model in which the parties have access to a trusted party, who can obtain all inputs, conduct the prescribed computation, and deliver the prescribed outputs but nothing else.
Results obtained by WIS scientists and their collaborators, in 1987 and 1988, indicate a variety of communication and computation models in which secure multi-party protocols is possible in general. These include the standard model of communication coupled with standard intractability assumptions (like the intractability of factoring large integers), and a model of communication in which private communication channels exist between each pair of parties.
Subsequently, WIS scientists made numerous contributions to the study of secure multi-party protocols.