16 May 2012

Robot Journalism

What started as an experimental lab at Northwestern University with journalists and technologists working together is now a fully-fledged business that turns data into stories of a type which will not be winning many Pulitzers, but which certainly pass the Turing test of making one unsure whether they were written by a person or machine. The lovable "stats monkey", which came from the same series of research experiments, does the same for sports stories, without the attendant vet bills, bananas and spelling errors associated with employing a real monkey.
Although this algorithmic approach to compiling stories is by no means new – the lab which spawned Narrative Science was conducting and publishing work a number of years ago – the ultimate ramifications of what the approach symbolises seem to be taking a long time to sink into most newsgathering organisations.

To be honest, I’m surprised that robot journalism is as new as it is since most news stories seem to be written by retarded fourth-graders who have just realized what a sentence is.  Seriously, look at any newswire story, and they all seem to be the same.  Look at the news story I quoted in this blog post.  All the paragraphs were a mere sentence or two, and there was no cohesion to story.  It was just a bunch of random facts strung together with a couple of quotes interspersed throughout.

Worse yet, all news stories are just like this.  They all start with a one-sentence paragraph provides the basic gist of the story, followed by a couple paragraphs of random facts, followed by a couple paragraphs of quotes and analysis, and topped off with what is supposed to be either a clever or an insightful concluding paragraph.  Of course machines can replicate this. The proverbial monkeys at their proverbial keyboards could do this.  Come to think of it, they’d probably be better at it.

No comments:

Post a Comment