What is the relationship between videogames and violence? Popular Science, reviewing a new psychology report, noted the obvious: it's inderminate. There are two camps of researchers, neither of which can collect enough conclusive evidence to provide strong confirmation for their hypotheses. It's also unsurprising. Human behavior is complex and messy, and questions of measurement, causation, and inference are difficult to sort out. In a bid for greater "policy relevance," researchers recklessly extrapolated beyond their data or failed to note conflicts and limitations. The media only reported on sensational research. And not all of the research was produced under a sufficiently objective rubric.
But this problem is by no means unique to video games. This month's International Organization published two studies on the benefits (or lack theorof) of nuclear superiority with diametrically opposite conclusions. As Daniel Nexon argues, this poses an analytical problem for the policymaker. Here we have two articles that were judged to be of sufficient quality to be published in a top-flight journal, with completely different conclusions. Nuclear coercion, like the sources of violence, is a fundamentally messy and multicausal subject.
Even if we can come to a general agreement as to which confluence of factors is important, as Daniel Byman and Matthew Waxman did for the use of airpower in Kosovo, we still face a fundamental problem of how to make policy tradeoffs. And this, of course, still presumes that we agree about the proper weighting of the factors. Whether airpower, the threat of ground forces, or any of the other defeat mechanisms Byman and Waxman specify should be regarded as necessary or sufficient conditions is a question that is unlikely to be resolved without a substantial degree of deep historical investigation. Kosovo happened close to fourteen years ago, and authors still argue about the causes of World War I. And I'm not going to even bring up the Surge because it will likely be impossible to have an intelligent conversation about it for a very long time. Likewise, Joshua Foust has often written sharp pieces about the data problems that hold back robust study of targeted killings in Pakistan.
So we're back to the beginning: videogames. Public policy is always mostly normative choice, but indecisive science tends to highlight policy's subjective foundations. So in regards to videogames and violence, the following questions come to mind. What level of (un)certainty are policymakers willing to accept in making decisions to infringe on the freedom and choice of others? What level of potential harm do policymakers believe would justify such a decision? Should the infringement be cautious or maximal? Finally, would the government respond by heavily restricting the product known to contribute to the behavior, act primarily on other environmental variables that interact to produce the undesired behavior, or both? These are all choices that can be informed by science but not dictated by it. And in this situation the degree to which the choice can even be "informed" is fairly contentious.
As this blog more or less explicitly and implicitly suggests, the same issues involved in the videogame dispute are also true of national security policy and their interaction with political science. That's why national security professionals tend to like Clausewitz so much. On War provides a general outline of the general thing called War. As with any work of gestalt theory that does not try to directly predict certain outcomes but describes a system as a whole, On War tends to be last longer than science that advances through conjectures and refutations.
We should be looking to foreground what criteria we use when thinking about the messy problem rather than necessarily believing that research can always tip the scales one way or another. Of course you don't need a lifetime of political experience to know that not much "foregrounding" goes on in domestic politics. Hence the overlap between the audience willing to take this post seriously and those it would most help is bound to be fairly low.