In a 1957 paper, Robert Solow exploited the mathematical properties of the aggregate production function to isolate the role of disembodied “technical change” in economic growth. Solow’s method allowed to disentangle the role of technical change from that of production factors, with the residual serving as a measure of total factor productivity growth. His method and results were met equally with praise and criticism. The interrogations around the residual gave rise to an abundant literature from the late 1950s which made it possible to improve the technique of calculation and refine the results. The Solow residual inspired a surge of interest (and criticism) in the 1980s when used by Finn Kydland and Edward Prescott to justify empirically the concept of technology shocks. In this paper, I argue that the resulting debates were not essentially different from those which took place—within the National Bureau of Economic Research already—in the 1950s and 1960s. Though, they have been accompanied by a change in the “epistemic status of shocks” (Duarte and Hoover 2009, 228) in economics, which redesigned the Solow residual from a source of secular growth to be quantified to the initial impulse of short-term economic fluctuations. From then on, they entailed a choice among competing models of business cycle. I allege that the Solow residual, highly malleable and easily decomposable from both a theoretical and an empirical point of view, turned to be a clear-cut—although porous over time—demarcation line within the Freshwater/Saltwater spectrum (according to Hall’s 1976 metaphor) between economists believed the business cycle was mostly driven by supply-side versus demand-side factors, and a formidable weapon in the battle that divided them.