(R&R at the American Economic Review; with Mohammad Akbarpour and Amin Saberi) Identifying the optimal set of individuals to first receive information in a social network is a widely-studied problem in settings such as the diffusion of information, microfinance programs, and new technologies. We show that, for some frequently studied diffusion processes, randomly seeding S + X individuals can prompt a larger cascade than optimally targeting the best S individuals, for a small X. Given these findings, practitioners interested in communicating a message to a large number of people may wish to compare the cost of network-based targeting to that of slightly expanding initial outreach.

(With Matthew O. Jackson and David McAdams) We examine how well agents learn when information reaches them through chains of noisy person-to-person relay. If noise only takes the form of random mutations and transmission failures, then there is a sharp threshold such that a receiver learns fully if she has access to more chains than the threshold and nothing with fewer. Moreover, simple information processing rules can perform as well as fully Bayesian learning. However, if some agents deliberately distort message content, learning may be impossible with any number of chains, even if the fraction of such biased individuals is small.

I study how limited information and ex-post evaluation by third parties with the benefit of hindsight affect how regulators approve innovations. In the face of ambiguity over innovation characteristics, such a regulator limits or delays product approval, even when she is not waiting for new information to arrive. When evidence is costly for firms to generate but can be selectively reported, the regulator delegates information acquisition to the firm with the objective of minimizing max-regret. This model can explain observed patterns of correlation between firm costs and benefits of approval, why regulators drag their feet on approval decisions even in the face of strong favorable evidence, and support for regulatory sandboxes even when they do not hasten learning.