Skip to content
Dec 18 / King Kaufman

Nieman Lab’s predictions for journalism 2015, including some created by a bot

The Nieman Lab is running a series called “Predictions for Journalism 2015.” The Lab asked “some of the smartest people we know” to write about what they think will happen in the coming year, and their answers will keep appearing through the end of the week.

As I write this, three dozen have been published. And some of those include more than one prediction. Here’s Sarah Marshall of the Wall Street Journal with 10 predictions. This is an ambitious project.

I haven’t read many, but my favorite so far is this one by Matt Waite, who wanted to say something about bots writing stories.

To save Nieman Lab a lot of (fictional) money, and to prove a point, I am not going to write (more) predictions for next year. It’s too hard and too expensive. So I’m going to let my computer do it.

To do this, I took all of last year’s predictions—all 38,053 words of them—and fed them to a Highly Technological Natural Machine Language Learning Algorithm Bot (nerds: a simple Markov chain generator). Using that corpus and that algorithm, I generated 200 Original Statements About the Future.

Then, using a Highly Advanced Editorial Workflow System I call editing, I cut those 200 statements down to a set of predictions that I will now claim as my own.

Among them: “We’ll probably see some more super-rich people jumping into the product.” And “We will pass from the connections. It makes sense to fact-check a twerking video. Thankfully there’s no thundersnow.”

As Waite points out, his predictions couldn’t do any worse than his predictions did a year ago, and it only took him a few seconds. That’s progress.