A repository of materials for a proposed class on automated story bots.
Switch branches/tags
Nothing to show
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Type Name Latest commit message Commit time
Failed to load latest commit information.


JOUR491 - An Introduction to Storybots

On the morning of March 14, Ken Schwencke rolled out of bed, logged into the LA Times content management system and published a story about an earthquake that happened three minutes prior.

This is interesting only in the particulars: Schwencke didn't write a word of the story - no human did. The story was written by a bot. Specifically, a bot programmed by Schwencke to write a story about earthquakes minutes after they happen, so the LA Times can have something on its website about it first, and faster than any human can do the job. All Schwencke had to do was check the story to see if the bot did its job and hit publish.

Welcome to the dawning days of storybots: programs that write stories so humans don't have to. Bots now write simple stories about softball games, corporate earnings reports, the weather and, yes, earthquakes. Some of them are extraordinarily simple -- basic programming knowledge is all that's required. Others are quite sophisticated -- they rely on more complex tools to arrive at a simple story.

Journalists (including your professor) have written other storybots that:

  • Summarize crime statistics for a neighborhood
  • Update readers on real estate trends for a neighborhood over the last month, six months and year.
  • Write a four paragraph story evaluating crime trends for every city in the US.
  • Summarize little league games
  • Automatically write U.S. corporate earnings stories for the AP
  • More to come as I find them

One company said their bots are on track to write a billion stories this next year. That's billion with a B. That's a lot of stories.

Even your phone does this. Have an iPhone? Swipe down for the Notification Center. Depending on how you have it configured, you'll get a couple of sentences telling you what the weather will be like today, how long it will take you to commute somewhere, and what your next appointment is on your calendar.

But where is all this bots-writing-stories stuff going? And how hard is it, really?

This course is going to take your journalism education and combine it with some programming knowledge to produce storybots. We'll discuss the ethics, the practicality, the limits, the nature of what it means to be human and we'll make a few storybots of our own.

To Do

  1. Evaluate some textbooks. Specifically Matt Jocker's book and the Python NLTK book.
  2. Evaluate language choice - R or Python? Gut says Python, but R is worthy challenger. Need to make a call.
  3. Develop the two story bot assignments as midterm and final.
  4. Develop class meetings that get to the points where they can complete the midterm and final.
  5. Find course readings across a range of disciplines.
  6. Develop lectures.
  7. Develop progressive assignments that build toward midterms and finals.