How to quench a rumor in a decentralized network
C. Bordenave (Univ. of Toulouse)
We will present a simple model of rumor propagation in a decentralized network where nodes may be (S)usceptible (unaware of the rumor), (I)nfected (aware of the rumor and spreading it) or (R)ecovered (aware of the rumor and trying to quench it). The key difference with a classical SIR model being that a node cannot be recovered before one of the nodes who has told him the rumor is himself recovered. For a simple network model, we will show that the birth-and-assassination process (Aldous & Krebs 90) is a scaling limit of our rumor propagation dynamics. We will state new results on this process and exhibit an heavy-tail phenomena. Finally, we will apply these results to the rumor quenching process.