Misinformation Spread in Gossip Networks: The Influence of Transmission Mutations

The rapid dissemination of real-time updates in interconnected networks often encounters the challenge of misinformation spreading alongside accurate information. This paper examines the interplay between timeliness and accuracy of updates in fully-connected gossip networks, where probabilistic mutations during transmission can convert truth into misinformation. We consider a network of <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">n</i> user nodes that receives updates from a source and employs an age-based gossip protocol for faster dissemination of version updates to all nodes. When a node forwards its packet to another node, the packet information gets mutated with probability <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">p</i> during transmission, creating misinformation. The receiver node does not know whether an incoming packet contains correct information or misinformation. The receiver runs a gossip protocol that looks only at the version age of the incoming packet and accepts it if it is fresher than the packet in its possession. For the case when the incoming packet has the same version age as the receiver’s own packet, we consider two system models: In the first model, we assume that truth prevails over misinformation, and therefore, when a receiver encounters both accurate information and misinformation corresponding to the same version, the accurate information gets chosen for storage at the node. In the second model, we assume the opposite scenario, where misinformation prevails over truth. For both models, we study the expected fraction of nodes with correct information in the network and the version age at the nodes using the stochastic hybrid systems (SHS) method. We observe that when truth prevails over misinformation, very high or very low gossiping rates help curb misinformation, and misinformation spread is higher with moderate gossiping rates. However, when misinformation prevails, misinformation rises with increased inter-node gossiping. We support our theoretical findings with simulation results which shed further light on the behavior of the above studied quantities.