A recent study in Communications Physics is among the first to systematically analyze whole city networks of spider-webbing roads and sidewalks. Examining 10 global cities, including Boston, Paris, and New York, researchers created an algorithm that could help planners choose certain sidewalks to expand for walking or biking, while minimizing disruption to car traffic. “There is a lot of interest in redesigning cities toward pedestrians,” says computer scientist lead author Daniel Rhoads, a graduate student at the Open University of Catalonia in Barcelona. “What’s missing is a methodology to propose these changes.”
In early 2020, when the COVID-19 pandemic was in its infancy, Rhoads and his supervisor, computer scientist Javier Borge-Holthoefer, the study’s senior author, wondered if many urban sidewalks were even wide enough to allow the 1.5-meter (5-foot) social distancing that some governments called for. Rhoads and Borge-Holthoefer also wanted to know whether unsafe walkways could be improved—through strategic widening that entailed closing adjacent roads to cars and opening them to pedestrians.
To find out, the two researchers and several collaborators first collected existing maps of roads and sidewalks, painstakingly collecting the sidewalk data from open data portals in the 10 cities. (Data regarding city footpaths is rarely publicly available and often difficult to track down.)
The researchers then systematically removed the narrowest sidewalks, where distance between pedestrians is lowest, cutting any walkway from the network that was less than 5 feet wide. What remained was a picture of COVID-safe sidewalks—and, unfortunately for most cities, that network turned out to be discontinuous and fragmented.
Next, the researchers, starting from scratch, again removed the narrowest sidewalks based on their width and distance between walkers. But this time, before they removed a sidewalk from the network, they assessed if the road next to it could be closed to cars and converted to a pedestrian walkway. “For each segment of the network you have a decision: Should I give up the road or the sidewalk?” explains Borge-Holthoefer. “Someone will lose.” The researchers set their algorithm to do the least harm, meaning they closed roads when the continuous connectivity of pedestrian transit in the entire city would be relatively more disrupted by the loss of a single sidewalk than would the driver transit network due to the loss of the road. Surprisingly, opening enough streets to pedestrians to keep the sidewalk networks functional only reduced the connectivity of roads by 5 to 10% across the cities, translating to slightly fewer paths from point A to point B, and hence minor disruptions for drivers, Borge-Holthoefer says.
“Even having a pedestrian network you can ask these questions of is a rare thing,” says postdoctoral computer scientist Nicholas Bolten at the University of Washington in Seattle. Typically, researchers might have a list of street names and some sense of which have sidewalks; but comprehensive data tracking the width of footpaths, how they’re connected in a web, and which have curbs or other details required for transit planning is rare, he says. This study and others, including recent work Bolten himself coauthored, are building those citywide data networks needed for planning.
This latest paper “privileges the pedestrian,” says city planner Michael Batty at University College London, who was not involved in this recent study. “That’s quite innovative,” he says, in part because so few papers have combined pedestrian and vehicular road transit networks in a single analysis. Yet, only by joining those two networks together can researchers analyze the tradeoffs and conflicts between them, he adds.
In London, where Batty lives, the pandemic has precipitated a number of changes to the urban landscape, including narrowing some two-way streets to one-ways and widening adjacent sidewalks by laying down new pavement. The new algorithms proposed in this paper would enable London and cities like it to take a systematic rather than ad hoc approach when deciding which streets to close or open to give pedestrians more space, Batty says. “We need to be sure the city continues to function,” even with such changes, he notes. This paper offers a method to develop a much more integrated view of the entire street system, he says. “They’ve almost suggested it could be a design tool in this way.”
One possible limitation, however, is the scale of the analysis. This study analyzed cars and pedestrians on citywide scales. Of course, pedestrians move over much shorter distances than cars. To better achieve social distancing aims, it might be possible to use this algorithm to look at different scales, such as city centers and local neighborhoods, rather than entire cities, Batty notes.
Future studies might include updating the algorithm by layering in more information about traffic flows—for instance during rush hour at the heart of a town versus on its periphery—which could point to more efficient choices about redesigning cities and redistributing space.
Other recent papers recommended by Journal Club panelists: