What airports needs to be shutdown in the next epidemic? From which cities does fake new most rapidly spread? Modern society is filled with networks where people, animals, business are represented as nodes and their connections are mapped as edges. Examples include road networks, social networks but also gene-interaction networks, food webs and so on. Understanding how these so-called complex systems operate is inherently difficult. One traditional approach is to relate the structrual features of a network to the causal importance of the node. In many fields of science there is a tendency to implicitly assume that well connectedness of a node, i.e. iets centrality, is proportional to the causal importance of that node. For example by knowing the structure of a road network, one can also infer the traffic that occurs on a particular road by looking at the number of lanes. This form function relation occurs through the arts and the sciences where the structure and function are correlated.
In this study, we answered the question “What node is most important?”. That is, we tested the status quo to what extent structrual features can predict causal importance. By measuring so-called information flows and using interventional strategies the driver nodes of a system were tracked. It was shown that the structural features are not predictive to determine the most causal node. In order to determine the most important node, the dynamics that exist on the network must be taken into account. We proposed a novel metric that can reliable determine the causal driver node. The results from this study emphasize that for complex systems the relation between form and function is highly non-trivial and warrants a second look when using network science.
