Mutually Assured Disconnection

A revolutionary new way of war, the ultimate tool of asymmetric conflict, a militarised white elephant, or a new weapon of mass destruction. Cyber war has been described by academics and commentators as any one of the above in recent years. Evidence for these claims have included notable attacks such as the Stuxnet worm, the Estonian and Georgian conflicts, calls for cyber jihad and the increased interdependence of critical national infrastructure. Thomas Rid has stated that Cyber-War will not take place; that technological limitations and definitional misconstructions are at fault for fear of such a paradigm shift in warfare being so over-hyped. Brandon Valeriano and Ryan Maness argue that states have developed norms that restrict the use of such attacks between each other. I argue that these norms exist, but that their development has taken place not from a position of mutual understanding and cooperation, but from one of mutual and balanced fear. While the first salvos of cyberwarfare may have been fired, an important observation has been largely missed. We have seen no evidence of the Cyber-Armageddon that technological capability assessments combined with service interdependence warn us to anticipate. This paper investigates this conundrum. Analysis of instances in which actors have employed cyber-attacks, the trends of spending and development on these technologies, their potential capabilities, and historical trends of warfare when faced with potentially game changing new technologies leads to pointing to the existence of a phenomenon that I label Mutually Assured Disconnection. And it is this that explains why we have seen nothing that can be described as a Cyber Hiroshima. Those actors capable of developing and deploying such attacks fear to do so because those whom they would attack have potentially similar capabilities. The technological nature of these attacks means risking symmetric warfare and mutual self-destruction through deliberate action or the unintended consequences of attributional error.