Kraftwerk, Manchester International Festival 2009

Can certain elements of the news writing process be given over to an automated processes? In blunter terms, are newsroom jobs being done by trained humans that could be performed by robots?

It’s not as stupid as it sounds, not like something out of Terminator (as one tweeting reporter suggested) nor the apocalyptic Death of Journalism(tm) some may describe it as. Online publishers of all kinds – from local newspapers, to lifestyle brands and sports sites – are looking for efficiencies and extra traffic, so if there was a way to increase traffic, the amount of relevant content you publish and the quality of your site, you’d consider it wouldn’t you?

Illinois-based startup Narrative Science (NS) – profiled here in the New York Times over the weekend – has just the automated story writing system you might be looking for. It raised $6 million (£3.79 million) in January (via AllthingsD) and has 20 customers so far, including local US newspaper chains and B2B brands. CEO Stuart Frankel is a former DoubleClick executive.

Narrative Science takes key facts and adds in connective sentences so a story makes sense. This college football report (American football, not football) was generated using NS’s algorithms. It’s not Pulitzer Prize winning material, but it’s factually accurate, informative and functionally does the job of creating a short match report. NS clients are automatically turning local sports results and company financial earnings into short stories, posted online within 60 seconds of the full time whistle or stock market release; as Frankel puts it to the NYT: “Mostly, we’re doing things that are not being done otherwise.”

The ability to add context is what might have some journalists really worried however. As the NYT reports:

“The Narrative Science software can make inferences based on the historical data it collects and the sequence and outcomes of past games. To generate story “angles,” explains Mr. Hammond of Narrative Science, the software learns concepts for articles like “individual effort,” “team effort,” “come from behind,” “back and forth,” “season high,” “player’s streak” and “rankings for team.”

Turning data into language

Back in 2006 my then-colleague Martin Stabe mentioned NS and its automated newscasts, when the company was still a project run out of the Intelligent Information Laboratory at Northwestern University (which retains a stake in the business). So this isn’t anything particularly new, nor that revolutionary.

But with the advance of data-based publishing, the time of automated reports may be increasingly upon us. So if you have a bunch of data, an audience who want you to translate it from numbers into language, but no journalists. B2B publisher Hanley Wood uses NS on to provide short, automated reports on 350 regional markets, for $10 a 500-word article. There are obvious potential problems there in terms of quality and accuracy, but how many journalists would it take to do this the old fashioned way?

And as Facbook founder Mark Zuckerberg is prone to tell his employees via motivational posters (via @EdwardRoussel): Done is better than perfect.

Should we be worried by this, or be the first to welcome our new robot overlords? And seriously, if things like this offer an efficient way to add value to your publishing mix by turning untapped data streams into human language, surely that’s worth investigating?

I’ll leave you with this vision of the future courtesy of the Simpsons. Maybe “don’t praise the machine” will become a catchphrase in a newsroom near you?

The Simpsons DJ – 3000 from Chronite on Vimeo.

Main picture is via Any Miah, via a Creative Commons licence.