P-graph Causality Maps

Abstract

A new class of bipartite causality maps is developed in this work to model influences in socio-technological systems. Unlike earlier forms of causality maps such as fuzzy cognitive maps, this novel approach uses two kinds of elements represented as different types of graph nodes. Based on the P-graph framework, “objects” and their “mechanisms” (or actions) are represented by O-type and M-type nodes, respectively. Objects consist of both tangible (e.g., technologies) and intangible (e.g., policies) elements, while mechanisms are the roles that objects play within the system. Objects in the network influence each other through these mechanisms as indicated by arcs. Two P-graph algorithms are used for this application. First, the maximal structure generation (MSG) is used to generate a maximal causality map based on specifications of local links of individual objects; the causality map assembly is automatic and does not require human understanding of the global problem topology. Next, the solution structure generation (SSG) algorithm is used to algorithmically generate all structurally feasible causality maps for the problem. The alternative networks can be evaluated to gauge practically viable candidate policy and technology combinations. Criticality of different objects can be quantified based on the frequency of occurrence in the enumerated causality maps. The methodology is illustrated using a case study on a policy framework for negative emissions technologies (NETs) for CO2 removal. The potential of this technique to determine how to sustain “virtuous networks” or to deactivate “vicious networks” is also discussed.