This research is to undertake a preliminary examination of the growing use of NSF sponsored supercomputers and computers that may substitute for some or all of their functions. The goal is to develop a research plan leading to a better understanding of the determinants of demand for large-scale computing services and better pricing structures to make the most efficient use of the services and their alternatives. The researchers will review the extant literature on demand and pricing of computer services, develop a conceptual model of demand for supercomputing services, will develop pricing models appropriate to the use of supercomputing services, meet with knowledgeable individuals, and develop a full research plan. NSF has an interest in the pricing and demand for computers for several reasons. First, NSF supports university-based supercomputer centers. There soon may be demands for more computer time than these centers can supply and some price or non-price rationing scheme will be needed. Second, supercomputers may involve transferring large volumes of input or output between remote users and a central computer, so network congestion and usage charges may be significant. Third, NSF grants often include an element for computer usage; so NSF has an added interest in efficient and equitable pricing policies for all types of computer services.
