Real-world experiments that test new technologies can affect policy and practice by introducing new objects of intervention through tinkering; the ad hoc work of realigning relations in the face of frictions, surprises, and disturbances that occur when introducing a technology. In a pilot study on aggression detection, tinkering moved aggression in and out of the human body. In the end, the pilot study constituted aggression as a set of acoustic-physical variables representing the aroused human body, alongside other signals of aggression. How aggression as an object intervention was established by tinkering is relevant because it involved inclusions and exclusions by the authorities who identified aggression, the methods they applied, and mandate for intervention. A focus on relations that are tinkered with in a real-world experiment permits critical engagement with this format. Although the format of experimenting outside of the laboratory is credited with producing knowledge about a technology’s ‘actual’ performance, actors and events at the pilot study location were made only selectively relevant. Analyses of real-world experiments should therefore explain how experiments selectively make the world relevant, giving only particular objects of intervention a truth status.