Parallel Computing: Technological Changes and Organizational Redesign

Taking into account the very limited ability of individuals to handle and process information (for given information-processing technology), some economists argue that the sequential regime of data processing has been replaced by parallel methods (in economic jargon: hiererchical or decentralized computations (see, e.g., Radner (1993)) in order to respond to the growing problem of handling an increasing flow of information (Chandler, 1966, or Bolton and Dewatripont, 1994). Not surprisingly, therefore, organizational changes in data-processing systems (i.e., reduction of parallel computational structures) observed in the last decade (see, e.g., Business Week, 1989) are usually explained by the improvement in information-processing technology (Schein, 1989, Kennedy, 1994). However, research to date has not provided an adequate economic explanation of the conditions under which improvements in information-processing technology reduce the degree of parallelization of computational processes (i.e., the size of decentralized computational structures). Thus, the subject of the present paper is an examination of the conditions that have to be satisfied for the organizational restructuring of the structures used for parallel computing in response to changes in information- precessing technology.