This paper, authored by Joseph Farrell and Garth Saloner, explores the impact of an installed base of durable goods or training on the likelihood and desirability of innovation. The authors analyze the private and social incentives for adopting a new technology that is incompatible with the existing installed base. They identify three main sources of benefits from compatibility: interchangeability of complementary products, ease of communication, and cost savings. The presence of these benefits can lead to "excess inertia," where users are reluctant to switch to a new standard due to network externalities. The paper presents two models to examine this phenomenon. The first model assumes that the new network is built up through the adoption of new users, while the second model assumes that the new network is built through the switching of existing users. In both models, the authors discuss the conditions under which adoption or non-adoption can be efficient, and the conditions under which there is excess inertia or excess momentum. The authors also explore the anticompetitive behavior of firms, including product preannouncements and predatory pricing. Product preannouncements can prevent a new technology from gaining momentum, while predatory pricing can deter entry by temporarily reducing prices. The paper concludes with a discussion of possible extensions and future research directions.This paper, authored by Joseph Farrell and Garth Saloner, explores the impact of an installed base of durable goods or training on the likelihood and desirability of innovation. The authors analyze the private and social incentives for adopting a new technology that is incompatible with the existing installed base. They identify three main sources of benefits from compatibility: interchangeability of complementary products, ease of communication, and cost savings. The presence of these benefits can lead to "excess inertia," where users are reluctant to switch to a new standard due to network externalities. The paper presents two models to examine this phenomenon. The first model assumes that the new network is built up through the adoption of new users, while the second model assumes that the new network is built through the switching of existing users. In both models, the authors discuss the conditions under which adoption or non-adoption can be efficient, and the conditions under which there is excess inertia or excess momentum. The authors also explore the anticompetitive behavior of firms, including product preannouncements and predatory pricing. Product preannouncements can prevent a new technology from gaining momentum, while predatory pricing can deter entry by temporarily reducing prices. The paper concludes with a discussion of possible extensions and future research directions.