the simplest model of social learning (Morris DeGroot, 1974): each agent repeatedly replaces her opinion with a weighted average of her neighbours' opinions. the trust weights form a row-stochastic matrix, so by perron-frobenius-theorem the process converges to a consensus — and each agent's influence on that consensus is her entry in the stationary distribution
social influence is eigenvector centrality in disguise: the same mathematics that ranks particles by focus ranks people by trust