This document discusses computationally viable handling of beliefs in arguments for persuasion. It introduces persuasion problems where an agent tries to persuade another by exchanging arguments. It proposes using a belief distribution to represent the opponent's belief in arguments, and updating the distribution as new arguments are added. It describes splitting the distribution using a metagraph to group related arguments and reduce computational requirements. Experiments show the approach scales to larger argument graphs by updating individual flocks of arguments rather than the full distribution.