MULTILAYER NETWORK MODELING FOR INTEGRATED DETERRENCE AND ARMS CONTROL

The international security and conflict environment has fundamentally changed and now includes two nuclear near-peer competitors, growing nuclear threats from states like North Korea and Iran, and concerns of nuclear latency from both allies and adversaries. Additionally, competition in and proliferation of advanced technologies like hypersonic missiles and artificial intelligence mean that tomorrow’s battlefields may exist in entirely new domains and will almost certainly reach across and connect domains in novel ways. Previous theories of deterrence (largely based on game theory) are illequipped to provide understanding of this new multiplayer, multidomain problem set. The emerging concept of integrated deterrence relies on the explicit and implicit interactions of domains, actors, geography, and alliances to successfully deter across a wide spectrum of adversary actions along escalatory pathways. However, because integrated deterrence is a relatively recent concept in U.S. strategic literature, it lacks a defined method of operationalization. This project’s goal is to determine whether multilayer network modeling could serve as a useful method for depicting and quantifying the relationships between different aspects of potential conflict domains. Integrated deterrence can be described as a range of efforts from disparate domains (e.g., nuclear, cyber, space, conventional [maritime, land, air]) instruments of national power (military, diplomatic [arms control, nonproliferation measures], economic etc.), and actors (whole of U.S. government, allies, etc.) that are utilized in a coordinated and integrated approach across geographic theaters and across the spectrum of conflict to deter adversaries. The complexity of the relationships between these different types of efforts points to the value of a multilayer network model, where instruments of national power are modeled as individual network graph layers that include efforts from all domains. The unique capability of multilayer network models is the ability to explicitly illustrate points of interconnection between network layers and evaluate the impact of how changes in these connections impact overall performance—using percolation measures to describe impacts from changes in perceived alliance strength, for example. From this model, we hope to provide a rigorous and structured approach for evaluating—and comparing the relative risks among—different escalatory pathways.