Crosstown LA uses public data to create hyper-local journalism
Big Data For News Publishers | 16 July 2024
At United Robots, our focus is squarely on local media. Much has been written about its demise — and there are doubtless huge challenges — but with innovation, local journalism really has an opportunity to thrive.
As I see it, local innovation should be about ideas that make local publishers even more local. It’s about creating differentiation where no national or international media can compete — on the local stories, the connections with people in their actual communities. A perfect starting place is public data — data which can tell so many hyperlocal stories, if you just know where and how to look.
I’d like to share the story of Crosstown LA, a ground-breaking case of a publisher that turns public data into valuable, hyper-local journalism. It was presented as part of the Innovate Local programme, which United Robots supports, at the recent World News Media Congress in Copenhagen.
The Crosstown case was presented by Professor Gabriel Kahn at the USC Annenberg School of Journalism, which is the publisher and editor of this local journalism venture in Los Angeles. Kahn joined forces with a Ph.D. student in engineering and set about creating a tool whereby journalists can generate hyper-local stories from publicly available, but previously inaccessible, data.
This resulted in 114 weekly neighbourhood newsletters on everything from crime to public health, housing, and traffic. One stat proves the value of a hyper-local focus: The open rate is north of 90%.
The starting point: scalable stories from public data
Kahn’s basic idea was to use data and automation to bring down the costs of covering local communities and increase engagement. His base is Los Angeles, but the same is true of many U.S. cities: They publish troves of data that track quality-of-life issues including public safety, complaint calls from residents, building permits, parking tickets, and traffic accidents.
What if it were possible to make this public, but mostly inaccessible, data available to journalists? This would give them the chance to surface stories on a hyper-local level and distribute them to readers on an equally granular basis.
In other words, they could produce hyper-local relevant content in a financially scalable way.
The constituent parts: data, dashboards, and a newsletter platform
The data
In Los Angeles, the authorities publish 15 different datasets on their Web site on a daily, weekly, or monthly basis. This is data about crime, arrests, housing, construction, traffic, citizen complaints, potholes, garbage, graffiti, home values, parking, business licenses, restaurant inspections, and more.
A very useful aspect of this data is that each row of it has a latitude/longitude notation. In other words, it is as granular as it can be in terms of geography. Crosstown uses those coordinates to tell stories at the neighbourhood level — and there are 114 of them just in Los Angeles.
The data dashboard
For journalists — in the first iteration that was just Gabriel Kahn and another journalism professor — to access and use the data, the USC engineering student built a data dashboard.
“The data is very difficult to access” Kahn said. “Basically, it’s publicly available but not publicly accessible. You need a Ph.D. in data science in order to make it accessible. And once you get into this data, you realise that it tells the same stories that local news has been telling forever. If you can unlock it, you have a thousand stories that you can tell at any one time and you can tell them at the local level.
“The dashboard turns every journalist in the newsroom into a data journalist, because what it allows us to do is to ask questions of the data in a very simple way. We can take millions and millions of rows of data and find a pattern in them.”
The newsletter platform
What remained was the issue of getting the right stories to the right readers. This is what Kahn refers to as “the last-mile problem.”
The solution was to build a platform allowing Kahn and his colleague to write newsletters and populate them with hyper-local data. In other words, they write one newsletter but automatically deliver 114 unique ones, one for each neighbourhood. The newsletter platform also includes a library of custom data visualisations which journalists can use.
The result: “We contextualise news and we hyper-localise news”
Each dataset can generate a number of different stories depending on how you slice and dice it. Journalists can ask any question of the data, thereby uncovering lots of stories without having to do any sort of big mathematical operation. The answers appear right away.
“We often find that other news organisations in Los Angeles are following us because we have this data before they do. We know how to read it,” Kahn said.
Crosstown’s work brings context to the city’s daily news. Thanks to being able to access and query the underlying data, it’s possible to challenge public perception of an issue, such as an increase in crime, by looking at whether isolated events in the news are, in fact, part of a pattern or not.
There is a lot of talk in the U.S. news industry about news deserts in rural parts of the country. The fact is, they actually exist in the middle of Los Angeles. Thanks to the data dashboard, the Crosstown team can now examine how what is happening in the city impacts equity, Kahn said: “We can essentially pull apart the city, and we can really act as advocates for those who are underserved and underrepresented.”
The Crosstown platform is currently in use in newsrooms in three U.S. cities: Los Angeles, Chicago, and Raleigh, North Carolina. The plan is to scale up the solutions to other newsrooms and cities.