We outline a set of steps that could lead to new quantitative analysis and understanding of science policy based on scientifically grounded conceptual framework and large-scale computational analysis of scientific activity. Getting the right conceptual and empirical framework matters, lest resources and people get squandered because incentives are wrong. Getting an empirical framework based on something other than anecdotes matters, to avoid substantive misunderstandings about the process of science. Seizing the opportunity presented by the explosion in digital information about research products and processes, will require both substantial effort to acquire, integrate, curate, and evolve large quantities of information from many sources, and much innovation in both science policy research and computational methods.
