The Case for Software Eating Journalism

by Rob Poulter


A lot of what I think about seems to be sparked from podcasts lately. I listen to a lot of them as I drive between towns, while I walk or run, as I shop. A lot of them are about technology, with a few generally interesting ones thrown in (lots of the NPR podcasts are great - it'd be nice if we did more of this in Australia). Recently I've had a bit of an overlap between two podcasts that I listen to: Vector (about technology) and Planet Money (which I guess is economics, but really seems to cover just about everything - I'm sure the economists out there will say that this is just as it should be).

The topic is that of journalism. The Planet Money episode is one of the series that they did (which I've written about before) on automation, specifically the topic of automated journalism. One of the gimmicks of that show was having the "robot" journalist (a system called WordSmith)  compete with an experienced human journalist to write a report on the earnings report for the dining chain Denny's.

The consensus from the hosts and listeners from the results were that the human did a better job in terms of flavour, but both stories were essentially the same when it came to the facts. 

Where this fits in with Vector is the episode (81: Celestial Revisionism - sorry about the unofficial link, the iMore page for the series doesn't seem to be up to date) which discussed the situation around the MPAA and others conspiring with Attorneys General against Google over copyright interests. Now regardless of how you feel on the whole music/movie piracy issue, business interests conspiring with the AG behind the scenes (as opposed to lobbying out in the open) is a pretty bad situation. The fact that further to this they also planted coverage in the Wall Street Journal sympathetic to their cause is just icing on the cake.

How these two situations go together is that I think there may be a pretty good argument for, as the title says, having software eat journalism. Stop having bias being introduced into reports by individuals. Software does a pretty good job of taking raw facts and writing a clear story, so just start doing it. Many of the facts are generated by people, but we have a large enough catchment area that these can often be automatically verified or debunked, or at least attach a decent confidence rating to things which are less than certain. 

One of the potential issues with this idea of automated journalism is that of trust. Many of us don't trust machines to do things for us, despite having them do many things invisibly for us every day. Again, there was an episode of Planet Money which looked at driverless cars, and some of the historical problems of automating things. Their examples were elevators and planes, and the mental perception of some of these types of automation that people had to overcome before they used them. I think that many people would object to a piece of software writing a news story because they would have trouble trusting it. The more worrisome problem in my opinion is both a blessing and a curse: it would be harder to tell from casual inspection whether the software had been tampered with (or always biased) to have a bias to its reports (or to be flat out incorrect), but at the same time it should be provable upon detailed inspection as to whether or not it is fair. For human journalists it can be difficult to prove whether or not their reporting has a bias.

Impartiality aside there is a large amount of bias in pure news coverage (as opposed to opinion - which is fine when disclosed as such). I think it would be an interesting exercise to see how "robot journalism" (with apologies to Jason Snell and John Siracusa's Robot or Not? podcast) could potentially improve the way in which we consume fact-based journalism. Would it make individuals less partisan? Would it make us less interested in the news because it was less sensational? Would it not make a difference at all since we would make up our own narrative anyway? I look at "news" and "current affairs" and is just makes me sad since the facts are often buried in spin of some sort. I think I'd like news journalism to be eaten by the machine.