U.S. federal research funding is generally justified by promises of public benefits, but the specific natures and distribution of such benefits often remain vague and ambiguous. Furthermore, the metrics by which outcomes are reported often do not necessarily or strongly imply the achievement of public benefits. These ambiguities and discontinuities make it difficult to assess the public outcomes of federal research programs. This study maps the terms in which the purposes and the outcomes of Advanced Research Projects Agency-Energy -a relatively (...) young and innovatively structured federal research agency-have been discussed and reported. I find that ARPA-E’s creation and funding have been justified with reference to a broad repertoire of economic, environmental, and national security values, but that the agency has been evaluated only through intermediate scientific and economic metrics and study of internal organizational structure. I suggest that these means of assessment elide ARPA-E’s lack of the financial scale, long time horizon, built-in customer, and radical vision which have been historically important to high-impact federal innovation, and I recommend the tracking of metrics more directly related to ARPA-E’s public value purposes as the agency grows older. This discussion illustrates the inadequacy of currently widespread metrics and conventional wisdom for the design and assessment of societally relevant research. (shrink)