Despite the suspicion that often arises between the methodological camps, it is my belief that most researchers in social science want to move in the same direction. We all have a desire to understand how economic and political actors (ranging from individual voters to crowds to nation states) make decisions. To the extent we understand social science, any of us would be happy to have policy relevance. As noted by Friedman (1953) and Granger (1999), the research community has many shared beliefs – we all, more or less, agree on what “the good” is in terms of outcomes. What we disagree about is methodology, because it is not obvious which approach provides the most leverage in understanding the world. Put another way, while we might all agree on the merits of preventing nuclear warfare, the more salient question is how to discriminate good models and good advice from bad.

Part of the problem is that the social sciences do not have a shared understanding of what constitutes a reasonable question. Too often, asking a good question is confused with providing a good answer. One might want to understand the causes of a particular event (say, World War II), but, without a great deal of care, a few cases do not constitute a coherent problem suitable for research. As I have argued throughout, the main problem that confronts us in research design is the curse of dimensionality. It cannot be ignored, whatever one's methodological orientation.