Observations on the role of influence in the difficulty of social network control

Previous work introducing the idea of distribution-based network control determined that some seemingly similar networks can have significantly different levels of controllability. This work investigates these differences in controllability in more detail and finds that one of the driving factors behind controllability may be the influence dynamics within the network. These results suggest that existing structural heuristics for control set selection that do not take into account influence, such as the FAR heuristic used here and in previous network control works, can produce substandard control sets that result in poor controller performance. The development of an algorithm for control set selection based on network influence measurements is identified as an important direction of future work.