Abstract for: Screening potential leverage points in complex systems using an analytical framework based on Causal Loop Diagrams

Leverage points are usually identified and tested via stock-and-flow models. Nevertheless, large Causal Loop Diagrams (CLDs) are increasingly used to understand complex systems and suggest “where to intervene” without simulation. However, there is no transparent way to identify potential leverage points from CLDs: current practice is narrative, heuristic, or based on opaque graph tools. This work asks how to provide a structural screening of candidate levers directly from large CLDs. An analytical framework is introduced that treats large CLDs as directed graphs and combines in-eigenvector centrality and influence–dependence scores to structurally screen candidate leverage points. Interpretation focuses on the smallest set of variables covering most eigenvector “mass”. A second, contextual screening step maps these candidates onto feedback loops and archetypes, using simple loop metrics, and is implemented transparently in Python for both unweighted and stakeholder-weighted CLDs. Ongoing tests on a large CLD from the REGENYSYS project illustrate how this structural screening can support narrative interpretation of complex systems and generate focused hypotheses and design options for subsequent simulation work, without claiming that qualitative analysis alone is sufficient to prove the existence of leverage points. Only for refining english grammar