Influence of spreader dispersion on rumour propagation in complex networks

Abstract In social systems, information and misinformation can enter the underlying interaction network through multiple independent sources. Its cumulative reach depends not only on the influence of the sources but also on how they are collectively positioned in the network. Here, we study how the spatial dispersion of multiple initial spreaders influences rumour propagation on complex networks. This setting differs from epidemic spreading because rumour spreaders may become inactive after contacting already informed nodes. Hence, the placement of the initial spreaders affects not only initial coverage but also the timing of deactivation. We consider classical three-state rumour dynamics with interaction-driven stifling and compare dispersion-based initial spreader selection strategies, constructed using farthest-first traversal and a hybrid local-connectivity refinement, with non-dispersion-based strategies. Numerical simulations on Erd"{o}s-R\'{e}nyi, Barab\'{a}si-Albert and Watts-Strogatz networks, together with six real-world network datasets, show that dispersed initial spreaders generally increase both the final rumour size and the peak fraction of active spreaders. This occurs because dispersion reduces overlap among spreading regions and delays deactivation caused by contacts with already informed nodes. The hybrid strategy, which combines topological separation with local connectivity, typically outperforms degree-based seeding over a broad range of spreading and stifling probabilities. Our results show that in multi-source rumour spreading, a combination of broad separation between the initial spreaders and strong local connectivity is crucial for maximizing rumour reach and intensity.