Treasure Hunt *

We seed a large real-world social network with binary information and analyze subsequent social learning. A unique feature of our field experiment is that we measure both the pre-existing social networks and the actual conversation network. Our experiment allows us to test how rational agents behave when processing information that originates within their social network. We find that information decays quickly with social distance and that agents mainly incorporate information within social distance 2. Conversations trough common friends do not increase the weight that a subject places on signals from direct friends but linearly increases the weight on signals from indirect friends. This suggests that agents are able to avoid double-counting information from indirect friends. We propose a simple “streams model” of social learning that is consistent with the evidence from our experiment. JEL Classification: C91, C93, D83