We show how important phenomena in social networks like coordination, trust and the communication of unsubstantiated information (gossip) can be modelled and understood using epistemic networks or epinets: directed graphs comprising networked agents and the key facts, statements or other kinds of propositional beliefs relevant to their actions. We use epinets to sharpen the explanatory and reach of social network analysis to situations problematic to both network-structural approaches and epistemic game theory.