Why do some conflicts escalate to global war while others stay contained? I provide a formal model of conflict that considers an international system of many countries located on an economic network. Countries may attack an adversary of their choosing and, if a war breaks out, the remaining countries choose whether to join the fight. I demonstrate how the distribution of power and globalization can interact to affect the expected size of conflicts in a number of ways, motivating quantitative analysis. Connecting the theoretical model to data on international disputes, I estimate deep structural parameters governing country preferences, including the underlying economic network. Lastly, I perform counterfactual analysis on World War I combatants facing a decision to join the war between Austria-Hungary and Serbia, finding that changing power and economic relations of the 20th century make the Allies more likely to fight but the Central Powers less likely.