EAGER: Computer Progress and Economic Prosperity

Besides contributing to U.S. productivity, information technology (IT) improvements have been an important source of competitive advantage for U.S. firms. This research project explores alternative sources of information technology improvements that could drive U.S. economic prosperity in the future, and identifies research and policy interventions that would be needed to generate the improvements.

The work combines techniques and content from two fields – computer science and empirical economics. It applies new expertise, and incorporates novel disciplinary and interdisciplinary perspectives. Many of these analyses will also be the first of these types to be done, fulfilling the EAGER goal of being exploratory.

The project identifies opportunities for computer productivity improvement as the contributions from Moore’s Law ends, focusing in three specific areas: hardware, algorithms, and software.

- Recent roadmaps developed by the International Roadmap for Devices and Systems (IRDS) and the Heterogeneous Integration Roadmap (HIR) propose domain specialization as the alternative driving force for hardware improvement. However, specialization also fractures the semiconductor market, undermining the economics of chip production. The hardware component of this project will examine how much specialization would be economical.

- The President’s Council of Advisors on Science and Technology has claimed, based on case studies, that algorithms are more important than hardware for computer improvements. The PI has recently created a first-of-its-kind census of algorithm progress, and shows that such rapid improvements only happened in a limited number of areas. The algorithm component of this project would map how algorithms are being used, to understand how much algorithm progress is benefiting users, who is benefiting, and how much this is changing over time.

- Machine learning (ML) has been producing substantial benefits. However, much of this improvement has come from deep learning, an area where economics could impact future gains. Continued progress will require either better software performance engineering or new, more-efficient machine learning techniques. The software/ML component of this project will explore the potential for these approaches.

Overall, this EAGER project will fill an important gap in understanding of the technical, economic, and policy implications of various approaches to computing and IT productivity improvement, and provide a better understanding of the connections between scientific advances, industrial policies, and economic progress.

This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.