Evidence suggests that innovation benefits from low-powered incentives, yet innovative organizations use widely varying incentive structures. I offer an explanation based on a characteristic feature of creative work: dead ends. To solve a problem, an agent works on successive ideas, each of which may succeed with some probability. At each instant, the agent chooses whether to exert effort. The agent may also abandon his idea, incurring delay to come up with a new one. High rewards for success can slow innovation because the agent is reluctant to incur the cost of delay, spending too much time on unpromising ideas. I apply this framework to study intellectual property rights (IPR) and optimal contracts for innovation. Dead ends provide a new explanation for the inverse U relationship between IPR and innovation, suggesting that "low-hanging fruit" suffers most from strong IPR. In a principal-agent setting with moral hazard, we get front-loading because high continuation values increase the cost of current incentives. Contract structure depends on whether the principal is more or less patient than the agent. Impatient principals impose deadlines, while patient principals grant tenure, using delay rather than the threat of termination to reduce incentive costs.