The Stories We Tell

Once upon a time there was a startup called Circa.

It was an innovative startup built on some very interesting ideas and run by some really smart journalists like David Cohn and Anthony De Rosa, and it got a lot of great press for much of its life.

And then it ran out of money, and closed.

Much ink (pixels?) has already been spilled on the untimely death of Circa a few months ago – for a sampling, look here, here, here, here and here – and this post really isn’t about the site’s closing as much as it is about a broader issue that some of us engaged in structured journalism are grappling with, and which has been blamed at least in part for Circa’s end.

Put starkly – as Jason Calacanis does at Medium:

Circa failed because it was so efficient, so obsessed with the truth, that it was boring.

Ouch. So was Circa boring?  And is much of structured journalism – at least the reconstituted, recombined content that’s one of its selling points – destined to be boring?  And will that doom the idea?

It’s certainly one of the topics that Bill Adair, Laura and Chris Amico, and David Smydra discussed when we last got together in a windowless room in Boston to trade ideas about structured journalism. As Bill noted in a guest post here:

A common theme in our discussions: the need for narrative and context in structured journalism.

Which is another way of saying that structured journalism is great at delivering information, but perhaps less so at being engaging, and enticing people into that information, the way a great piece of storytelling might.  And there’s a logic to that – although, the more I reflect on it, the more I think there’s also a false dichotomy to that argument.

But first let’s back up and recall what Circa was, and what it offered.  Fundamentally it was a mobile-only news site that took news stories, broke them into chunk-sized bits that could be easily read on a small screen, and more importantly, tracked what you already knew about a topic from your previous browsing, and just fed you the things you hadn’t yet read about. It was a simple, yet breathtakingly smart idea – a really clever implementation of structured journalism and a nice advance on the old Google Living Stories experiment.

(Probably one of the best pieces about Circa is by Mark Coddington at Nieman Labs, who writes about the broader impact of atomizing information, the advantages of doing, and some of the unintended benefits of the practice.)

Circa’s branch-based taxonomy of news stories also hints at the potential of an idea that’s a conceptual sibling of the “atomic unit” — the importance of structured data in news. Circa’s branch system wasn’t truly structured — it wasn’t a relational database, or even a spreadsheet — but it was an attempt to systematically organize all of the events Circa had ever covered as individual pieces of one holistic network of news.

…

Circa demonstrated the result when events are treated as individual, atomized units capable of being structured into larger wholes: It leads journalists not to break them down into isolated occurrences, but the opposite — to think of them as more thoroughly connected to broader ongoing stories and issues.

But back to being boring.  As Frederic Filloux wrote in a smart 2013 Monday Note:

Circa is in many ways a contemporary product. First, it neatly addresses the attention span challenge. Remember: it’s 9 seconds for a goldfish, 8 seconds for a human in 2012 — vs. 12 seconds in 2000 —  and let’s not forget that, according to Statistic Brain, 17% of web pages are viewed for less than 4 seconds. Seriously, Circa found ways to save our precious time. Second, its content is more than neutral, it’s sanitized, deodorized. It’s a perfect fit for a generation of readers for whom facts are free and abundant, opinions are suspect and long form stories a relic of the past...

That last point – the bit about being neutral and sanitized – would become a criticism for many just a year or so later, even for those who agree that Circa’s atomized information system is a much better way of getting news.  Josh Benton makes both points here:

The traditional story form — start at the beginning, read to the end — is an inefficient way to transmit information. People who know a lot about a story get bored by obligatory background; people who don’t know a lot about a story don’t get enough context. If you atomize a story into its constituent “chunks,” you can do interesting things with them — like add them to or subtract them from a larger “story,” or send granular updates to stories people have expressed interest in.

 But then he goes on to add:

Chopping up a story into bits risks draining all human voice from it. Think about how, say, The Economist and BuzzFeed would write up their takes on a given story. They’d be quite different, obviously, but they’d also be identifiably theirs. Circa stories are bland and sapped of personality — a CMS strategy confused with an editorial one.

It’s makes sense that the process of atomization – and recombination – does take voice away.  It has to.  And even more so if and when you get machines to write the stories, which I’ve argued is both the only way to scale personalized/on-demand story creation as well as take full advantage of a database of atomized facts. Even stunning, rich data visualizations of the kind we created for Connected China aren’t immune to the notion that readers don’t just want to explore; they also want to be driven through a narrative that’s been carefully assembled for them.

So it’s true: A handcrafted story, written for a particular audience to be read at a particular moment in time, is likely to be superior to – in fact, is probably vastly superior to – a machine-assembled collection of facts and context. So how does or should structured journalism compete?

The same way machine-generated sports stories compete: By finding an unoccupied field to play in. The likes of Narrative Science and Automated Insights don’t try to outdo human sports writers covering the Super Bowl; they generate stories about college and high school games that no one covered before.  Are their stories worse than ones written by a human?  Let’s stipulate yes.  Are they better than no story at all? Absolutely yes.

So a Homicide Watch page tracking the murder and investigation of a single person in the capital is probably less engrossing than a fully-reported feature about that crime.  But that fully-reported feature probably doesn’t exist, which makes that Homicide Watch page not just the only – and hence better – alternative to nothing, but it also makes it a real public service for a hitherto deeply underserved community.

(You don’t need machines or structured journalism to serve underserved communities, of course – witness Buzzfeed’s commitment to covering LGBT issues around the world, filling a long-neglected gap in mainstream media coverage.)

Similarly, Connected China could probably have used more directed visualizations to help guide users through a really deep and detailed database, but it certainly provided far more information about an opaque power structure than was available in one place anywhere else. That Homicide Watch page isn’t going to replace great crime reporting; but that crime reporting isn’t ever going to scale to cover the range of murders that Homicide Watch tracks.  And you shouldn’t need to – and you don’t need to – choose between the two. You can and should have both.

Good crime reporting can feed a Homicide Watch database, and supplanted by some additional data gathering, a good Homicide Watch database can provide lots of information that’s of interest to small audiences.  Everyone wins, at least in theory. And in any case, as we move towards needing to create multiple versions of the same story – one for the print publication, one for online, one for mobile, one for a small screen when the reader is standing in a line at the bank, an audio alert for when the reader is running with headsets on, and so on, how else are we going to scale up unless we have quasi-atomized stories that we can recombine and reconstitute?

So was Circa boring?  Perhaps the bigger issue was that Circa was fighting on turf that was already staked out by lots of other people who were writing rich narratives on the same subject; and that it’s less the format that’s at issue than the competitive field.  Perhaps if they had picked a niche area of coverage, they may have gone further. Or perhaps, as with some other structured journalism projects, it was simply too far ahead of its time.

Disruptive technologies, as Clay Christensen tells us, typically gain ground by providing poorer quality products to underserved markets – and are scoffed at precisely for not being as good as the market leaders – but then improve and iterate, and eventually start encroaching on the incumbents’ turf. Ken Doctor put Circa’s issues in perspective very well in a late 2014 post:

The presentation metaphors offered by Circa and AJ+ are immediately attractive, but they may be less well-suited to fuller meals of news. Another way to say that is that the Circa has deconstructed news but it still figuring how to reconstruct it. … If the now much-maligned inverted pyramid — the foundation of AP-like “new top” writing, ironically thrust on the news industry of the time by an earlier tech upheaval, the arrival of the telegraph — is being replaced here, we might call it a diverted pyramid.

…

Facts, and wee bits of context work, but how will, or can, such sites encompass narrative text storytelling — which will remain a fixture of the future.

As far as structure journalism is concerned, automated creation of narrative storytelling is still in the future.  But there’s the real possibility of a here-and-now of a marriage between humans and machines to write both great stories and scale up the provision of targeted information to people who aren’t well-served with only great stories. Both important public services.

And then who knows what else the future might bring?  Perhaps there’s a happily-ever-after in here somewhere.