Human activities increasingly take place in online environments, providing novel opportunities for relating individual behaviors to population-level outcomes. In this paper, we introduce a simple generative model for the collective behavior of millions of social networking site users who are deciding between different software applications. Our model incorporates two distinct mechanisms: one is associated with recent decisions of users, and the other reflects the cumulative popularity of each application. Importantly, although various combinations of the two mechanisms yield long-time behavior that is consistent with data, the only models that reproduce the observed temporal dynamics are those that strongly emphasize the recent popularity of applications over their cumulative popularity. This demonstrates-even when using purely observational data without experimental design-that temporal data-driven modeling can effectively distinguish between competing microscopic mechanisms, allowing us to uncover previously unidentified aspects of collective online behavior.branching processes | complex systems T he recent availability of datasets that capture the behavior of individuals participating in online social systems has helped drive the emerging field of computational social science (1), as large-scale empirical datasets enable the development of detailed computational models of individual and collective behavior (2-4). Choices of which movies to watch, which mobile applications ("apps") to download, or which messages to retweet are influenced by the opinions of our friends, neighbors, and colleagues (5). Given the difficulty in distinguishing between potential explanations of observed behavior at the individual level (6), it is useful to examine population-level models and attempt to reproduce empirically observed popularity distributions using the simplest possible assumptions about individual behavior. Such generative models have arisen in a wide range of disciplinesincluding economics (7,8), evolutionary biology (9, 10), and physics (11). When studying generative models, the microscopic dynamics are known exactly, so it is possible to explore the population-level mechanisms that emerge in a controlled manner. This contrasts with studies driven by empirical data, in which confounding effects can always be present (6). The value of explanations based on mechanisms has long been appreciated in sociology (12-14), and they have recently received increased attention due to the availability of extensive data from online social networks (15-18).One well-studied rule for choosing between multiple options is cumulative advantage (also known as preferential attachment), in which popular options are more likely to be selected than unpopular ones. This leads to a "rich-get-richer" agglomeration of popularity (7,9,(19)(20)(21)(22). Bentley et al. (5,23,24) proposed an alternative model, in which members of a population randomly copy the choices made by other members in the recent past. As a result, products whose popularity l...