Robotic reporters

The threat to the newspaper industry might be at an end. Unfortunately for some in the media the solution seems to be the replacement of journalists with algorithms which create news stories.

The algorithms have been developed by a US company, Narrative Science, which produces computer-written news stories. The development was first talked about in Wired last year ( ) but it now appears the product is becoming more successful. In a post, Ray Kurzweil (a remarkable author, inventor, innovator and futurist) has updated how it is going. ( .

At first the fact that both articles were published in April (rather close to April 1) made the blog wonder a bit but Narrative Science exists and its offer can be found at   (

So far Narrative Science’s clients seem to be restricted to niche publishers although Forbes has run some articles. Current usage includes producing reports on Little League Baseball games from detailed data submitted by parents through an iPhone app with, according to Kurzweil, some 400,000 reports produced last year and a total of more than 1.5 million expected this year. And, in case you think Little League baseball is minor stuff the reality is that it meets a real local need for news about things that matter to locals. When the blog worked in local newspapers the managing editor, the legendary George Gilmour, hammered home to us the need to include as many local names and activities in the pages of suburban and regional newspapers. His mantra: “A local newspaper with 100 names in it is a good paper, 500 is better and 1000 is a great paper.” Detailed sports reporting of local clubs was a key element in this strategy. The enduring appeal of such local media is probably why Warren Buffet is buying local and regional newspapers (and radio stations) at a time when many bigger papers are closing or slashing staff. It also dovetails with recent research by the Pew Research Centre on declining community interest in US news media which the blog will address in a while in the context of its implications for Australia. (

What the algorithms are expected to be good at is mining “vast troves of data to produce ultra-cheap, totally readable accounts of events, trends, and developments that no journalist is currently writing,” Kurweil says.

It’s not all bad news for journalists. Narrative Science Chief Technical Officer and co-founder, Kristian Hammond, says the company set up templates using experienced writers and they were also needed in the initial stages as the program ‘learnt’ more about producing articles, although their involvement does seem a bit like asking the condemned to help construct the gallows. So far the algorithms have mastered sport and finance and other areas are being studied. The other areas include company communications suggesting that PR people might face some threats too.

The major drawback of the stories for Australian media is that the output tends, although competently and plainly written, to be factual. News media these days is more about comment and speculation – often with a specific political agenda – so one can assume some of that will stay. Indeed, we may end up with a world in which hard news is produced by the computer algorithms and the rest of the media output is that of the Bolts et al. But for proprietors, it all seems like upside – fewer staff, lower costs, more stories, new approaches – and perhaps more readers of things they are interested in rather than what editors or their proprietors think they are, or should be, concerned about. NOTE: Once again the blog is grateful to its polymath friend, John Spitzer, for news on Ray Kurzweil.