Thursday, May 26, 2016

You ignore the work on structured stories at your own risk

If you are a journalist (or PR writer), you should read David Caswell's latest about his research into structured stories and follow the links, especially to the structured stories database.

Here's a definition from one of Caswell's earlier atricles
Structured Stories is a form of structured journalism, an approach in which reporting is entered directly into a database and then extracted as needed to create digital news products. Early examples of structured journalism, such as PolitiFact and Homicide Watch D.C., are limited to fixed news items in narrowly defined subject areas. Structured Stories, however, attempts to encode any journalistic news — from any subject area — into structured events and narratives.

Articles are not obsolete in a structured journalism approach but instead are organized within much larger journalistic structures that provide context, coherence and flexibility. These narrative structures are then used to make news stories intelligible to computers and, therefore, available for a variety of digital applications.
This is a not-so-nascent-anymore corner of journalism thought that says, essentially, that the beautifully crafted narrative you just wrote is nothing but data. That fire story you wrote? The address is data, so is the amount of damage, the cause, the type of house, etc. And it's data insurers and others might be willing to pay for.

Adrian Holovaty famously proposed this disaggregation of journalistic stories in 2006. Matt Waite extended it seven years later.
Caswell has now been testing the idea on an operational scale in New York, Los Angeles and Missouri (Where he's a Reynolds Fellow).
You risk not paying attention to this at your own peril. 

Let me put it this way, how many of you scoffed not that long ago at the concept of computers "writing" journalism stories? How's that worked out?  (A search on "Automated Insights" will fill you in a bit more, such as the AP's wide use of the software, if you haven't been paying attention. Here's some more on AP.)

Now, if you go to Caswell's Structured Stories site and look at some of the work, to those steeped in "storytelling," the examples don't look like much. Bullet points, cards, etc. Certainly not an eye-pleasing "story" (and, let's be honest, we misuse that term a lot anyhow; much of what we do is factual exposition, not story).

But what you are seeing is Holovaty's vision beginning to be turned into reality. And here's why it's important:
  • When you break stories into data, you can repackage that data in many ways and resell it, meaning more streams of income in an era when that's guaranteed to get executives' attention (after all, what other business do you know that leaves more than half of its raw material on the shop floor?).
  • This inevitably means changes in workflows, training and, perhaps, the romantic notion of the storytelling journalist. From another of Caswell's articles: Working with structured information enables the journalist to become like an air traffic controller for news: coordinating, routing, verifying and organizing news as well as identifying gaps in knowledge and filling them by assigning journalistic resources to conduct original reporting. This level of coordination is an impossible, even meaningless, task in a media environment based on text articles, but in a structured media environment it becomes easy and valuable.
  • Caswell says he's shown in real operational situations that structured journalism can be done. 
  • Finally, if you scoff that "people will never read this stuff," I want you to think about two things. First, much of this is not designed to be read by people; it is designed to be read and repurposed by machines. Second, go to the top of any one of those story databases on the Structured Stories site and click on the "told as" drop down menu. Go down to "natural language."  I have not found a natural language version yet, though I haven't gone through all the items,. But take a close look, and what does it say? "Natural language generation by Automated Insights." Uh huh.
Oh, and there's one other. In an era where consumers more and more are accustomed to blending their information and media inputs as they wish, this is the ultimate way for them to do that with journalism.
In many ways, this is another aspect of the "semantic web," essentially, the idea of turning much, if not all, of the web into data. Projects such as Open Calais have been grniding along for several years, trying to figure out how to turn all this unstructured information (in our business = story) into structured data.

So whether the prospect makes you recoil or not, it's time to start paying attention. Your future in journalism (and PR) probably depends on it.

Labels: , , , , , , ,


At 5/27/16, 2:17 AM, Blogger DigiDave said...

I have been watching Caswell's work from afar and have been really impressed by it. Agree - it's not pretty. But the concepts he is playing with are great. I saw the backend once. Very impressive.


Post a Comment

Links to this post:

Create a Link

<< Home