Is it possible to estimate some 'safety' metric to assess the effectiveness of Intelligent Transportation Systems? In particular, we are interested in using Inter-Vehicle Communication (IVC) beaconing for increasing drivers' safety at intersections. In the last couple of years, the vehicular networking community reported in several studies that simple network metrics are not sufficient to evaluate safety enhancing protocols and applications. We present a classification scheme that allows the quantification of such improvements by determining how many potential crashes happen or can be avoided by a specific IVC approach. Using a modified road traffic simulator that allowed selected vehicles to disregard traffic rules, we investigated the impact of safety messaging between cars approaching an intersection. We show that in suburban environments simple beaconing is not as effective as anticipated. Yet, simple one-hop relaying, e.g., by vehicles parked close to an intersection, can improve drivers' safety substantially. Since the key purpose of IVC is safety, the paper closes the loop in the evaluation of the effectiveness of vehicular networks as defined today.