Political scientists are not meteorologists. Social science is not meant to predict the future. At the same time, prediction as a method plays a critical role in testing our assumptions about the way politics works. Today's discussion over social media of Israel's coalition breakdown between myself, Allison Good, Michael Koplow, and Brent Sasley is a good example of the positive role prediction can play in social science and policy analysis alike.
This exercise was (I hope) useful for the policymaking community. While prediction is not systematic enough for social science, it does have value for the policymaking community. While political scientists seek comprehensive explanations of political phenomena, policymakers seek to advance a state's interests in the long-term by making a series of decisions about the state's international posture in the short-term. At some point, decision makers need to make choices based on a best guess about the future. Prediction is imperfect but in policymaking an imperfect guess is usually better than no guess at all.
For academics, prediction is more problematic. In an article on e-IR today, fellow political science grad student Dillon Tatum explains the paradox of prediction in social science. He argues that human agency and the opacity of states and institutions makes it impossible to predict the future. Given that the point of social science is to systematically investigate a phenomenon, the discipline considers prediction into the future somewhat reckless. Since we cannot systematically investigate variables which have not yet come into existence and choices that actors have not yet made, making predictions is not systematic. This is because we can't account for variables that don't yet exist.
However, there is a difference between social science being predictive and social science using prediction as a tool. Political science in particular would benefit by recognizing this distinction. While we should not try to predict the future as an ends, predictions can be a useful means to test the validity of our assumptions. Prediction lies at the core of the scientific method, specifically the hypothesis. A hypothesis is a guess about the outcome of a yet-to-be conducted experiment. It is a prediction based on underlying assumptions. In our Israel coalition debate, this blogger, Allison Good, and Michael Koplow all hypothesized that given variables would lead to given outcomes (the fall or persistence of Israel's coalition). When all of us were to some extent wrong, re-examination of our explicit hypotheses helped our analytical community to better understand the variables at work in the Israeli political system.
Recognizing this distinction would also help us realize when our assumptions were wrong. For example, over the past 10-15 years political science essentially has predicted that the Middle East would remain authoritarian for the long-term. The prediction was not explicit in the literature but it directly followed from arguments made by Ross, Fish, Heydemann, Lynch, Bellin, Gause, Brown, and others. The Arab Spring has forced a wonderful and highly productive debate on whether these predictions were accurate. These same authors are making plans to evaluate all kinds of new variables and conduct research with implications for work on authoritarianism, democratization, rebellion and revolution, social movements, and political economy.
Yet the opportunity cost of writing a predictive article and waiting months for it to be peer-reviewed, revised and resubmitted, and published only to be proven wrong is extremely high in the discipline. Making wrong predictions, even if you explain why you were wrong, is not likely to get you tenure in a political science department in the United States. Given that peer-review is one of the core characteristics of academic inquiry, the speed of journal publication is unlikely to change.
However, social media offer political scientists another avenue to disseminate ideas. The speed of social media, when used correctly, can allow political scientists to run predictions which solidify arguments they make in academic journals. They allow such predictions - and the assumptions which underlie them - to be explicit and better documented. Social science cannot lower the opportunity cost of prediction in peer-reviewed journals and book manuscripts. However, it can shift its incentive structures and become more tolerant of using blogs and social media to make predictions as a precursor to more systematic academic inquiry. It can recognize that blogging is not a waste of time that could be better spent on writing articles, but rather an important part of the research process. Finally, it can recognize that such predictive processes allow social science to demonstrate its own worth to those who fund it and build bridges between policymakers and academics.