This paper investigates the nature of the IV method for tackling endogeneity. By tracing the rise and fall of the method in macroeconometrics and its subsequent revival in microeconometrics, it pins the method down to an implicit model respecification device—breaking the circular causality of simultaneous relations by redefining it as an asymmetric one conditioning on a non-optimal conditional expectation of the assumed endogenous explanatory variable, thus rejecting that variable as a valid conditional variable. The revealed nature explains why the IV route is popular for models where endogeneity is superfluous whereas measurement errors are of the key concern.