The visions surrounding "self-driving" or "autonomous" vehicles are an exemplary instance of a sociotechnical imaginary: visions of a future technology that has yet to be developed or is in the process of development. One of the central justifications for the development of autonomous vehicles is the claim that they will reduce automobility related death and injury. Central to this narrative is the assumption that more than 90% of road crashes are the result of "driver error." This paper describes the process by which this statistic has been constructed within road safety research and subsequently accepted as a received fact. It is one of the principal semiotic components of the autonomous vehicle sociotechnical imaginary: if human drivers are responsible for similar to 90% of road crashes, autonomous vehicles should in principle be able to reduce road death and injury rates by a similar percentage. In this paper, it is argued that death and injury are not an aggregate of events that can be distributed across the three central variables of traditional road safety research: the driver, the vehicle, and the environment. The autonomous vehicle sociotechnical imaginary has embraced the central assumption of road safety research, that road violence is not an intrinsic property of automobility but is contingent because largely due to driver error. On the basis of this assumption it has been possible to configure autonomous vehicles as the solution to road violence. Although sociotechnical imaginaries are typically oriented towards the future, it is the significance of the autonomous vehicle sociotechnical imaginary in the present that is the focus of this paper. Autonomous vehicles are not the radically transformational technology their proponents claim but simply the most recent of a succession of automobility sociotechnical imaginaries. They are not transformational because their promotion ensures the continued reproduction of more of the same: namely, more automobility.
