When ‘push’ does not come to ‘shove’: Revisiting ‘faster is slower’ in collective egress of human crowds

We revisit the assumption stating that greater levels of rush in pedestrians’ collective egress through narrow bottlenecks impedes the discharge process and makes it slower, commonly known as the ‘faster-is-slower’ phenomenon. The question is of great practical significance because it ultimately can translate into whether crowds of evacuees should be dissuaded from rushing at bottlenecks in order to minimise their evacuation time. Yet, there is a large mixture of evidence on this phenomenon in the existing literature. Here, we re-examine this assumption based on empirical tests with an aim to identify explanations for these discrepancies. Our experiments were conducted with a crowd of 114 individuals, under varying doorway widths (ranging from 60 cm to 120 cm) and under three different levels of (non-aggressive) rush/competitiveness. Under our most competitive condition, crowd density behind the exit frequently exceeded 8 ped/m2 and even reached 9 ped/m2, as may be observed in a real case of egress under severe time constraint. This elevated level of crowd pressure and competitiveness, however, never translated in slower egress even for the narrowest exit. Based on every relevant measure of movement efficiency and regardless of the door width, faster was invariably faster. Discharge rates were larger, time headways between successive exits were smaller and evacuation times were shorter when individuals pushed more intensely (compared to more orderly types of conduct). We also observed that pedestrians exited in bursts and that the burst sizes were bigger under the greater levels of rush. Overall, all measurements indicated that for moderately large crowds, as long as the competitiveness does not amount to dangerous physical pressure, and as long as individuals do not display ‘explicit’ or ‘aggressive’ forms of pushing (i.e. as long as ‘push’ does not come to ‘shove’), rushing per se does not prolong the discharge process, rather it shortens the collective discharge. A contrast between these observations and previous experiments in earlier studies indicates that the presence or absence of ‘explicit’ or ‘aggressive’ shoving in the crowd could possibly be a major determinant of faster being slower or faster. This study suggests that the ‘faster-is-slower’ term, although very common in the literature, might be an overly simplistic terminology that does not offer adequate neuance for describing a rather complex phenomenon. To determine whether faster is slower or faster, one may need to break the question down into more details and view it through context-specific influential factors such as ‘the nature of pushing’, ‘the size of the crowd behind the bottleneck’, or 'the physical characteristics of the door'. We particularly suggest that another possible factor in determining when faster is slower or faster could potentially be the ‘crowd size’, a dimension that could be systematically investigated by future studies.