We examine the incentives for experimentation in the context of innovation and market competition. A monopolist chooses whether to sell early-stage product or perform costly scale-up R&D. Early market participation facilitates learning about demand but invites knowledge spillovers and competitors, while R&D acts as a barrier to entry. The firm's optimal policy can exhibit both under- and over-experimentation vis-a-vis the socially optimal policy. Patents can control the pace of innovation and restore efficient experimentation. When the surplus from R&D is large, rewarding early-stage innovation encourages market experimentation and limits wasteful R&D investment. We offer a theory of two-tier patent policy involving "petty" patents.