Features:

Subverting the Story Model

How we broke the story model and remade it to suit the new pace of news—and how you can, too


(Maxime Le Conte des Floris)

Since news arrived on the internet, members of the digital news world have warned against applying the story structures that governed print and broadcast to the web. The internet has undoubtedly sped up the news cycle, and the incremental story—taken from a world where daily updates were all that was possible—fails to keep up with the pace.

As the Trump Administration swiftly rolls back Obama-era regulations, new information leaks to the press, and Congress holds controversial hearings, combating the breakneck pace of news is more urgent than ever.

That urgency has created new opportunities for the NPR Visuals Team to guide parts of our newsroom towards alternative story models. This moment is an opportunity to evangelize the work our community has been attempting since its inception. And I hope to convince you that you have an opportunity to fundamentally change how your newsroom covers the administration in a way that will benefit your audience.

Why Now?

Let’s articulate the problem. There are too many damn Trump stories. The emergence of products like What The Fuck Just Happened Today?, the New York Times’ The Daily and FiveThirtyEight’s TrumpBeat are a clear indication that audiences are clamoring for a more distilled, high-level view of everything happening at once.

Newsrooms, even those with the largest staffs, are struggling to keep up. Every story seems like a bombshell, so we divert most of our attention and resources to the most recent story. It hardly gives reporters enough time to process.

More importantly, our audiences cannot keep up with everything happening unless they read all day. That might be an unrealistic expectation.

Currently, newsrooms self-organize around the 800-word incremental story: as events develop, newsrooms write new incremental stories at each significant moment. As these collections of incremental stories grow, you end up with a disorganized group of stories that represent snapshots in time. It makes following a story that develops over time difficult to understand.

In the current digital and political environment, three factors undermine the effectiveness of that model. First, the sheer number of stories make it hard to effectively construct and promote everything. Second, there are large stories—Russian hacking into the 2016 election, conflicts of interest in the Trump administration, and more—that develop at a slower pace. The incremental story struggles to keep all of the necessary context in place. Third, everyone is writing the same story. Short of a novel scoop, your story will look and read just like your competitor’s.

To solve these problems for our newsrooms and our audiences, we must rethink the model.

Within the firehose of Trump stories are patterns: President Trump tweets an unverifiable claim. Sean Spicer defends a policy at a press briefing. President Trump signs an executive order. The House votes to repeal an Obama-era regulation. The key to subverting the story model is to use these patterns to your advantage—to use them as organizing principles.

Other newsrooms have tried to organize around an alternative story model before. Vox’s Card Stacks are a promising, if under-utilized attempt to track developing stories. Axios’s stream-like homepage tries to make the river of stories more scannable. Both of these attempts came at the start of a new organization.

But short of starting a new news organization, your best chance at getting your newsroom to think this way is through subversion of your current processes. If you’re reading this publication, you probably have skills in your newsroom that can greatly assist this effort.

To pitch subversion to your newsroom, you will need a framing that demonstrates what makes your idea better. I have two to offer, and maybe you have some to share as well.

Turn Spectacle Into Evidence

The first uses a phrase I’m borrowing from my editor, David Eads: Turn spectacle into evidence.

This frame addresses the aspects of the Trump administration—and the American political system generally—that are spectacle. These aspects draw the most attention, but rarely offer much in the way of substantive policy.

Annotating Live Events

The first time we used this framing was for the presidential debates leading up to the election. A typical newsroom might flood the zone with stories on the debate: a general recap, an analysis from a lead political writer, a specialized recap on a subject area like the economy or foreign policy, a story about the polls following the debate. That newsroom would spread its resources across many disparate stories, hoping one of them caught on.

This time, we decided to focus our resources more narrowly and harness the spectacle of the event by building a live transcript of the event that any of our reporters could annotate. The system we built ingests a caption feed and dumps the feed into a Google Doc. Dozens of our reporters and editors are in that document, ready to annotate it. We then transform the Google Doc into HTML suitable for our live web page. (You can read a deep dive on the technical details here. )

Our process on handling these events is constantly evolving as we get better, but typically, we have anywhere from 5 to 25 reporters working in the doc to annotate a transcript or other document. A copyeditor and a couple of backreaders clean up the transcript as it comes in, and a lead editor approves all of the annotations before they go live. It took some significant development and design time to build the system, but the resulting editorial workflow is beautifully simple.

Our newsroom houses experts in many disparate subjects, any of which can arise in a wide-ranging presidential debate. We know, from obsessively watching our Chartbeat dashboards every day, that the average day has one story that outperforms all the others. Rather than have our disparate subject matter experts compete with each other to get that story, we changed the model. Now, we have a model—annotating the primary source document—that allows us to harness all of our expertise in one resource.

The liveness of the experience plays into the spectacle of the presidential debate. But through the annotations and the complete transcript, the experience of the spectacle becomes a resource with a long life. Unlike in a 800-word story, we can offer the full context of any given moment. Between the three presidential debates and one vice presidential debate, our annotated resources were npr.org’s most successful digital product of all time.

So we kept doing them. We annotated President Obama’s farewell speech, President-elect Trump’s first press conference after the election, President Trump’s inauguration speech and more. They continually draw large audiences and continue to engage them at high rates.

Annotating President Trump’s Twitter

The ongoing, daily spectacle of the Trump administration is the president’s liberal use of Twitter, and we needed a similar framing to think about our coverage. The traditional way to handle this in a newsroom is to write a quick story every time the president tweets something notable. But that’s exhausting—not only do writers feel the need to fill enough space to make the story serviceable, we also need to enlist our engagement team, editors, copy desk, photo editors, and everyone else who needs to follow a story to publication.

Instead, we built a system that would allow us to annotate the tweets and add them to one resource. That way, an annotation can be 50 words when it needs to be 50 words and 500 words when it needs to be 500. The system we built is quite simple. A cron job gets the most recent tweets from President Trump’s Twitter accounts and enters them into a database. Then, a Django app reads that database and provides an admin interface for annotating those tweets. The annotations are exported as a JSON file which is ingested by our web page.

This allows us to quickly bring in experts when we need them. When President Trump tweeted about Christians in the Middle East to defend his immigration executive order, we brought in our Middle East Editor, Larry Kaplow, to write the annotation. When President Trump is tweeting more about health care, our health care reporters write the annotations. Over a longer period of time, annotating tweets in one place allows us to pool our resources in the same way the live transcripts do.

Expandable Resources, Not Incremental Stories

The second framing that has worked for us is to think about how to convey information in a way that is an expandable resource. Of course, our tweet annotator fits this model as well. The key benefit of this framing, however, is that it works best for complex, developing stories where writing incremental stories assumes a level of background knowledge that many audience members will not have.

From our Trump Ethics Monitor feature.

We thought this way when we built the Trump Ethics Monitor. Through a week of research and organization, we were able to collect all of the promises Donald Trump has made as a candidate or president about his conflicts of interest, and whether or not there is any evidence that he has fulfilled his promises. Again, the build was simple—we organized the information in a spreadsheet and built the app out of that spreadsheet with copytext.

The week of investment allowed us to build a comprehensive resource that we can point our audience to whenever a new development regarding President Trump’s conflicts of interest arises. We can then quickly add new developments there, instead of writing an incremental story that needs to include all the necessary context to understand the story.

Why These Worked

In all of my examples, there are a couple important things to note.

First, each model inverts the system by which primary source material is included. In the traditional model, reporters pull in primary source material at their discretion to justify their reportage. In these models, annotation is brought to the primary source as needed. Because we provide the full context of the complete primary source, audiences can trust our journalism more easily.

Second, each model encourages internal collaboration. It is not one person’s job to annotate tweets or cover Trump’s ethical promises. We have created systems simple enough for anyone to contribute to these resources. At their best, these new models are even fun to use. Your newsroom may find them a breath of fresh air.

Challenges

These models have some inherent struggles. They tend to live as resources rather than one-off stories—things we can refer back to and continue to develop over time. The current models for promotion do not play kindly with this model. Facebook has an (unwritten) penalty for a page posting the same link within a certain time frame. Re-promoting the same thing over and over on your homepage will get stale for return visitors. The only traffic source that prefers more of a resource-style model is search, where more incoming and outgoing links enhance the SEO.

The ways we have chosen to build our systems require alternative publishing models. That means overhauling the entire editorial process in order for these resources to get regularly updated. We’ve been lucky enough to find willing collaborators on our politics and business desks, but it takes a concerted effort to make updating these regular.

We’ve tried to make our tools as simple as possible—Google Docs, spreadsheets, and Django admins—but the fact remains that our newsroom does not normally incorporate these into its processes. Thus, despite seeming more efficient, the new model seemed like more work than writing incremental stories at first. But if you’re making a similar shift, keep at it! The efficiencies will come with familiarity.

Finding Your Project

The projects I’ve discussed all came about in different ways at NPR. Sometimes the Visuals team pitched them, sometimes we transformed a pitch that came to us.

The connecting thread among them is that they helped us rethink and repurpose ongoing, repetitive work we were already doing. The politics team was already generating transcripts by hand for major events and trying to annotate them as quickly as possible. We built the tool that made it feasible to do collaboratively, live. An ad-hoc group at NPR was already trying to report out all of President Trump’s conflicts of interest. Politics and breaking news were constantly writing stories about Trump tweets.

What could your newsroom report better if you stopped using the incremental story model to report it?

This Is An Opportunity

The work these new models require is the work that data journalists have always been doing: aggregation, collection, summarization, contextualization. In that way, what I am suggesting is nothing new.

In 2006, Django co-creator Adrian Holovaty wrote, “Journalists should have less of a concern of what is and isn’t ‘journalism,’ and more of a concern for important, focused information that is useful to people’s lives and helps them understand the world. A newspaper ought to be that: a fair look at current, important information for a readership.”

The work we do as collectors and organizers can direct our newsrooms to provide useful information in more productive ways. That has always been true, but the skill is more necessary now than ever.

You probably won’t change the entire reporting structure of your organization (though, if you do, then you’re gonna have to tell us about it). Incremental stories will remain the primary output. You will continue to report them yourself, and that work will continue to be good.

But at this moment, newsrooms are hungry for new ways to make sense of the endless onslaught of news. Because of that hunger, our team at NPR has had success applying these skills to collaborative projects in ways we have not had success with before. I would bet your newsroom is feeling similarly anxious to try something new to cover this unprecedented administration. Happy pitching!

Organizations

Credits

  • Tyler Fisher

    Tyler Fisher is a news apps developer on the NPR Visuals Team working on audiovisual storytelling, data and tools. Previously, he was an undergraduate fellow at the Northwestern University Knight Lab.

Recently

Current page