Abstract: 

We develop a model of directed Bayesian search over a multi-dimensional landscape of available ideas, consisting of two fields of knowledge as well as their combinations. Success of ideas is represented by the sample paths of a Brownian staple, an extension of the Brownian motion framework of Callander (2011) to higher dimensions. We characterize prediction and optimal search by a sequence of short-lived researchers. Prediction is complex: predicting the outcome of any novel idea generically requires considering the entire set of previously explored ideas. Derivative research plays a pivotal role in mitigating such complexity. We demonstrate that optimal frontier search is gradual, advancing at most one field at a time by combining a familiar idea from one field with a novel idea from another. These search dynamics align with observed patterns in patent innovation, scientific citations, and drug discovery.

For further information please contact: erika.somma@unibocconi.it