What is the relationship between theory and observation in the social sciences? This article advances a sociological reformulation of the “theory-ladenness” of data. Drawing on insights from the history and sociology of science, I argue that knowledge infrastructures, not just individual experts’ perceptions, are theory-laden. This conceptual framework helps to explain why scholars missed the rise of top incomes in the 1980s and 1990s. In the 2000s, newly-analyzed tax data revealed that top incomes had begun a dramatic upward climb in the early 1980s, summarized as the rise of “the 1%.” This article explains why it took two decades for this increase to become salient. I show how mid-20th century economic theory shaped the assembly of two knowledge infrastructures for identifying stylized facts about income inequality, both of which were incapable of tracking top incomes. Macroeconomists focused on labor’s share of national income, but ignored the distribution of income between individuals. Labor economists drew on newly- available survey data to explain wage disparities. By relying on surveys, these scholars filtered top incomes out of view: surveys top-coded high incomes and thus were incapable of detecting the rise of the 1%. Studies of top incomes that relied on income tax data fell by the wayside, creating the conditions under which experts, policymakers, and the public alike could be surprised by the rise of the 1%. Knowledge infrastructures are laden with past theories; in turn, these infrastructures shape the production of new theories and new ways of understanding contemporary social problems.