Title: A Quest for Knowledge

Abstract: Which discoveries improve decision making? Which question does a researcher pursue? How likely will a researcher discover an answer? We propose a model in which the answers to research questions are correlated. Revealing an answer endogenously improves conjectures on unanswered questions. A decision maker uses conjectures to address problems. We derive the benefits of a discovery on decision making. A researcher maximizes the benefits of discovery net the cost of research. We characterize the researcher's optimal choice of research question and research effort. Benefits and cost of research crucially depend on the structure of pre-existing knowledge. We find that discoveries are rare but advance knowledge beyond the frontier if existing knowledge is dense. Otherwise, discoveries are more likely and improve conjectures inside the frontier.