A revolt or protest succeeds only if sufficient people participate. We study how potential revolutionaries' ability to coordinate is affected by what they learn from those around themselves. We begin by exploring the trade-off s from such learning: some supporters of the revolution will meet other supporters and become more convinced that there is strong support for the revolution, while others will meet partisans of the status quo and become discouraged. We show that which effect wins is determined by the prior beliefs of the agents and the strength of the correlation of preferences across the population. With this understanding in hand, we then explore how homophily (situations in which people are more likely to meet others who have similar preferences) affects the possibility of a revolution. Homophily undercuts learning (at a quantifiable rate), but also has some other interesting effects. Under homophily, potential revolutionaries learn less about the support of the revolution from meeting another supporter of the revolution given that meetings are biased; but still learn nontrivially when they do meet a supporter of the status quo, and so learning becomes more asymmetric. We explore these effects on revolutions in detail. Beyond this, we also use our simple model to explore the role of protests and counter-protests before a revolution, not only in revealing support for and against a revolution, but also in signalling relative sizes of extremist factions. We also discuss why holding mass demonstrations before a revolution may provide better signals of peoples willingness to actively participate than other less costly forms of communication (e.g., via social media). We close with some discussion of government actions, including propaganda, redistribution, and violent repression.