In recent decades, a number of authors have relied on the multiple realizability argument to reject methodological individualism. In this article, I argue that this strategy results in serious difficulties and makes it impossible to identify social entities and phenomena.