Conversations among competitors

I develop a model of bilateral conversations in which players may honestly exchange ideas with their competitors. The key to incentive compatibility is a strong form of complementarity in the information structure: a player can only generate a useful new insight on a given topic if he has access to his counterpart's previous thoughts on the topic. I then embed this model into a linear social network in which player A first can have a conversation with player B, then player B can have a conversation with player C, and so on. I show that relatively underdeveloped ideas can travel long distances over the network and thus be shared by many agents. More valuable ideas, by contrast, tend to remain localized among small groups of agents.