Laura Bliss is a staff writer at CityLab, covering transportation, infrastructure, and the environment. She also authors MapLab, a biweekly newsletter about maps that reveal and shape urban spaces (subscribe here). Her work has appeared in the New York Times, The Atlantic, Los Angeles, GOOD, L.A. Review of Books, and beyond.
Here’s the latest in the smart national effort to cut crash fatalities.
Two years ago, New York City partnered with data scientists at Datakind and Microsoft to build the “holy grail” of road safety engineers: a software platform capable of predicting and quantifying the outcome of any given engineering intervention. Would a bulb-out at 7th and 39th reduce crashes? By 10 percent, 27 percent, 62 percent? How about a pedestrian crosswalk signal?
The model turned out to be overly ambitious for the small number of crash-prevention features currently on New York streets, and the amount of data the city has on them.
But officials did come away with a tool that should prove useful as the city pursues its goal of zeroing out all crashes: a traffic “exposure” model, which uses AI to estimate the volume of cars coursing any road, at any time. Spots with high crash rates but low traffic volumes, for instance, could hint at a flaw in the street’s engineering and design; conversely, places with high volumes and low crash rates might teach officials something about what makes a safe street.
Now every city has a chance to harness data to build safety into roadways. Inspired by New York’s work, Open Data Nation, a Washington, D.C., company that wields public data repositories to build analytical tools for businesses and governments, announced Wednesday that it intends to work with any interested city to lay foundations for a similar model—but one that’s designed to predict where crashes happen.
In partnership with Microsoft, the data scientists hope to advise cities that have largely shied away from “data-driven” anything to amass the right kind of information to eventually build a tool. They’ll also select three additional cities to build analytical frameworks that flag dangerous intersections and recommend crash-reducing interventions.
Carey Anne Nadeau, the CEO and founder of Open Data Nation, says it’s hard to anticipate exactly how the model will be built; it’ll depend on the information these cities already collect and their particular traffic-safety needs. It’s also possible some of her company’s goals won’t pan out—for example, there may not be enough data in any town to say that narrower lanes are a sure-fire crash-reducer.
“The important step is to take what we learned from New York and see whether and how it applies elsewhere,” Nadeau says. “We want to bring it to other mid-sized cities, test the concept, and learn about what barriers there are to doing this nationwide.”
If successful, it won’t be the first time Open Data Nation has brought a local open-data initiative to scale. In 2015, it worked with health-code violation data from the city of Chicago to predict restaurants serving up unsanitary meals, and turned that success into a tool that four other cities are now using.
A crash-prediction tool that any city could use ups the ante. According to the National Highway Traffic Safety Administration, more than 40,000 people died in car crashes in 2016, up 14 percent since 2014, the largest increase in half a century. With now more than 30 U.S. cities signed on to the “Vision Zero” platform—a policy pledge to winnow crash casualties to zero—any effort to help is likely welcome.