Tullock's (1971) Paradox of Revolution uses an Olsonian logic to conclude that revolutions should not happen in large societies. Cantoni et al.'s (2019) Hong Kong Experiment shows that, in sharp contrast to the literature that models protest as a coordination problem, actions can be strategic substitutes. We develop a model to address these standing puzzles, and investigate its empirical implications. We show that when the movement's goal is modest, free-riding concerns dominate the citizens' interactions, making their actions strategic substitutes. By contrast, when the movement's goal is to topple the regime, coordination concerns dominate, and actions become strategic complements. Moreover, with natural other-regarding preferences, some citizens participate in costly revolt even in large societies. A new empirical implication of the model is that as a regime grows stronger in the sense that a larger fraction of citizens is needed to overthrow it, the likelihood of regime change may rise.