Performance of U.S. Scientific Research Cyberinfrastructure: Structural and Relational Factors for Usage

Abstract

This paper identifies structural and relational factors for the performance of cyberinfrastructure as a class of cyber-physical systems, filling the gap in our knowledge about social and technical factors, as well as their coupling, for performance. This study draws from the literature on virtual organizations, network governance, and the coupling of socio-technical systems to develop a conceptual framework for performance. The empirical investigation focuses on the NSF-funded cyberinfrastructure program in the United States, arguably one of the largest and most comprehensive cyberinfrastructure programs in the country. The researchers created a unique data set with 11,143 projects over a 15-year period by combining three cyberinfrastructure program data sources. We utilized regression and random forest techniques to study the utilization and amount utilized. The results indicate the importance of structural factors - including resource allocation, field of science, and type of universities/colleges - as well as relational factors - including transaction types, resource involved, and time of coupling. The results also underscore the relevance of multi-level and actor-network perspectives in advancing theories and practice of performance for virtual organizations and network governance.