There’s a nice piece by Tyler Dukes, managing editor at Duke’s Reporting Lab, that muses about new ways newsroom content could used – if only we collected it regularly, and in a structured way. (Full disclosure: I’m in the piece, so I’m a little biased towards liking it. But it’s good despite my presence.)
Tyler cites as an example the difficulty of proving a negative – how hard it is, despite reams of police reporting in New York, to say with any certainty that a particular day was the first in memory to be murder-free.
It’s a question that manages to be both fascinating and infuriating at the same time, because locked away inside every news archive is a wealth of information that could be used to enrich daily reporting. But far from teasing this institutional data (think names, dates, addresses, convictions) out in any structured way, most news organizations can barely manage a passable internal search feature.
A site like Homicide Watch could probably answer the question, because it collects and presents information a particular way, as various people have noted, including here. Regular crime reporters, on the other hand, don’t – mostly because they’re focused, not unreasonably, on their day job, which is writing stories for that day’s paper. But the lack of rigor in what’s collected and what’s kept means that reams of valuable information is thrown out every day. Or, in some cases, collected in too-sporadic a fashion to be useful.
If we had that information at hand – whether on homicides, or car accidents, or school board minutes, or whatever – we could, at least in theory, reuse that information in new and interesting ways.
The trick is to be able to do that, while not giving up daily stories that inform people on a daily basis – the central thesis of structured journalism. And that means fundamentally deciding what events and data points you want to collect regularly, and making sure that happens, even if there’s no story to be done or if the story doesn’t require those facts. Then making reporters and editors fill out – or automate the filling out of – databases once the story is done. And, of course, having a CMS that can collect that information and store it in a structured way, and APIs that let you tap it easily.
This isn’t cost-free; it takes time, even if only minutes, to make sure data is regularly entered into a database. It may mean covering things – such as homicides, or school board meetings – where there’s no likely story, because leaving them uncovered would mean an incomplete (and hence less valuable) database.
But if you get it right – and this is, of course, highly dependent on what data or events you choose to cover comprehensively – there’s real potential to build new value and new apps on top of it. Tyler points to Settle It!, an app intended to help settle fact-based political arguments (true – there may not be a lot of those…) that is powered by the data collected for the award-winning Politifact.
Too many news apps are designed in response to the question, “How can we get all the stuff we publish into an app?” Instead, we should be asking, “What does our audience need in an app, and how can we provide it?”
And in this case, they settled on a fact-checker rather than all the other possible uses they could have come up with for the Politifact data; and then, because Politifact isn’t made up of narrative – it is actually, but highly-structured narrative – it makes it easy to turn out a different “news product” based on the original data.
All of the great features we built in this app were possible because PolitiFact’s Truth-O-Meter fact-checks have always been stored as data, not just as text archives.
While most news organizations still produce “articles,” PolitiFact creates highly structured data.
Any news organization would be wise to start building a foundation of content-as-data and APIs to support all kinds of mobile apps and other content uses for the future.
Amen to that.