How foreign policy recommendations are put together: A computational model with empirical applications

This paper presents and applies empirically a computational model of the way in which bona fide high level foreign policy recommendations by U.S. policy makers are assembled. We begin by pointing out that policy making can be seen as the connection of certain strings of words to other strings. We then discuss how these connections constitute certain types of foreign policy making phenomena as such. To theorize about such connections, one first needs to specify essential features of these phenomena, and we do so for one phenomenon: bona fide recommendations. We next turn to a discussion of the theory that links together the categories by which these features are represented. That theory explains how certain strings of words are assembled into new proximate goals, missions, and tools. The theory can be modeled computationally using the programming language Scheme, and we next present that model. We conclude by presenting a run of the model, showing the close fit between actual and generated strings.