stdout.be

A blog about programming, information architecture and journalism

in journalism, information architecture

On bundles and blobs

Summary Wrapping my head around two of the big contemporary transformations in online journalism.

News is information. (It’s many other things, too, but forget that for now.) The news industry often bundles information, because bundling is a convenient and an efficient way to disseminate information.

We bundle information over time: instead of reporting everything as we find it out live, we gather up all kinds of related information and bundle it into a story. Bundles come in different sizes: we can churn out a quick news report in half an hour, or we can save up months of work for an enterprise story. They also come in different colors: most often we bundle topically related information (a story), but sometimes we bundle by type of information (a rumors section) or time (today’s linkblog).

We also bundle — weave — information into narratives. Stories are not concatenated facts, they’re not bullet-point lists. Stories combine related information and glue it together in paragraphs. Let’s call these narratives blobs.

Now, the thing about bundles and blobs is that they’re simultaneously awesome and stupid. Bundles and blobs are like nuclear fission, really: splitting the atom powers your house and it happens to be a great way of killing people.

Bundles are convenient and efficient, but slow. We can’t have slow in a 24-hour news cycle. Blobs are the stuff of every great story, but they’re unstructured which means we can’t use computers to make them even better.

You know when we know

Whenever we bundle information over time, we’re stashing information onto a backlog when we could be releasing it into the world right away — that’s why bundles are slow. Newspapers are yesterday’s news and for stories we really care about, that just doesn’t cut it. Hence 24-hour news channels, live reporting, text alerts, tweets, sneak preview blog posts and living stories.

Slowness is one grievance, another is the distance stories create between the writer and her audience. There are bound to be many smart folk who use our news websites, and by only giving them access to the finished product, we’re refusing to let them help us tell a better story. Journalism is one great big conversation… but only if we allow it to be.

There’s a catch… Unbundling (cue dramatic orchestral soundtrack) has a price. More bits means less context and less understanding. Let me explain that.

Our brains are really good at context switching, but it’s taxing. Leaving your email client on all day and reading each mail immediately as it arrives is a productivity killer like few others. And don’t bug me on the phone while I’m working, unless you want to ruin my day.

Our brains are also really good at connecting the dots, but they’re best at it when there’s a good chunk of information floating around in the working memory. You can’t read Aristotle’s Organon one sentence a day and end up a Peripatetic.

Learning twenty different facts about a news story one at a time and spread throughout the day implies lots of context switching and a pernicious lack of that chunkiness you need to get your brains up to full power. The end result: tweets and bits, much like giant cookies, are best produced and consumed in moderation.

It’s not surprising that together with the advent of update-as-we-learn-more styles of reporting, there’s also a growing interest within media circles for long-form stories and explanatory journalism.

As we do more bit-sized reporting, we have to figure out the right mix between long-form and short-form. Nothing but tiny updates and readers will tell us we’re not doing our jobs, that they’re drowning in information but don’t feel like they’re learning anything. This goes double for unfamiliar topics: readers don’t care for and can’t understand nuggets without a good old-fashioned story to tell them what it’s all about first. No updates and no process journalism on the other hand, and people will flock to faster and more participatory platforms, where they are treated as more than just a spectator.

Structure

I mentioned before that stories, which bundle information by translating it into sentences and paragraphs, are unstructured. Stories are big blobs of text. News stories hint at all sorts of structure, mind you, mostly of the x is y type: this person said that, this review gives that pub a four-star rating, this story is about that organization, which organized this event which happened at that place right there. But all of that structure is hidden in narrative prose, which makes it hard to parse for computers, which in turn makes it hard to display in another way or to repurpose it for another goal. Publish a map alongside each crime story? Nope. Create something as simple as a Top 5 of all recent movie reviews? Nope. Grab all the latest quotes from interviews with local politicians? Definitely nope.

The solution here is doing away with that other kind of bundle we’ve just talked about: the blob. Don’t for a second think that because you’ve done away with bundles, you’ve done away with blobs too. You can slice up your story into individual sentences if you want to, but instead of one big blob you’ve now got many tiny blobs. Doesn’t help you very much. You can tweet and live report all you want, but that doesn’t give you any more structure to work with than a big fat story would.

If we want to repurpose news or create mashups and need structure to do so, we have to put every individual piece of information we care about in a database — say, a database that links all stories to the geographical coordinates of the places those stories talk about. (Coordinates instead of place names, because names are free-form, thus blobby, thus bad.)

There’s a price to pay for unblobbing the news too. Applying structure to news usually means adding structural tags in addition to categories, dates, locations and other metadata. What it usually boils down to for a journalist is this: first, we submit our story as usual; then, we fill out a form so computers can sort-of understand our story too. But “just” filling out a form for every story we work on can turn out to be more drudge work than we signed up for. We don’t want to hasten through the categorization and data entry either, because if we don’t do it right, our data will be a mess and all our structure will mean nothing because it’s unreliable.

As metadata is such a time sink, getting rid of blobs is the art of doing just enough and not too much. It’s an art most news organizations don’t yet master.

So I guess what I’m saying is…

Stories make sense because they have just the right information density and allow journalists to add some much-needed context to the news in a way bit-sized updates can’t. But they’re slow, and they shut out readers instead of making them part of the process. Tweets and blog posts and short updates fare better in that regard — as long as you don’t overdo them because then it gets confusing and annoying. How much is too much depends on the story.

Stories make sense because people learn more from engaging narrative than from fact sheets. Without those fact sheets, however, computers can’t help us explore fresh ways to tell a story. We need new ways to tell stories: readers are losing their bearings in a sea of news and we need the tools to turn that experience into something better.

Keep writing the trusty old news article. But do more than that.


1 comment

Well said, as usual, Stijn.

The fact that unbundling has become sort of a default setting for fast-paced news organizations these days is all the more reason that we should be focusing on a better contextualization. Important news tends to come in storylines -- but besides simple tagging (lame), we don't really organize these episodic events into order. And we certainly don't indicate to readers which episodes were more important than others. Coupling our quick-hit stories with big-picture analysis on regular basis would be a huge public service.