News Diffs: A New Project To Track News Revisions

After 36 hours of hacking at Knight Mozilla MIT hackathon this past weekend, a new tool for tracking news article revisions was born from the handiwork of Eric PriceJennifer 8. Lee and Greg Price: NewsDiffs.

As a starting point, NewsDiffs tracks revisions for CNN and The New York Times stories, with more news organizations to come later. From the site’s homepage:

Sometimes the changes are minor — small edits in language or correction of spelling mistakes. Other times, the stories change and evolve rapidly, as a result of breaking news. Occasionally, the lede and substance of an article change as more reporting comes in on a fast breaking situation. Sometimes those changes provoke criticism.

NewsDiffs solves a fundamental problem with the minute-by-minute news cycle — changes are happening constantly and subtly with no form of documentation for those changes. A reader could stumble upon a homepage during breaking news, then come back a few minutes later to a different-looking page that reflects the most recent developments on that particular story. There’s no way to easily detect what exactly has changed since you last visited the site.  Because of the concept of “writing through” articles that live on one URL (as many news organizations do), those developments get completely lost over time.

As a homepage producer at The Seattle Times, I see this problem daily. Stories develop rapidly and, wanting to ensure that readers have the most up-to-date information, we update prior versions of a story and homepage centerpieces to reflect the most recent information. But there’s a story to be told in the revisions leading up to the most recent version of an article. Small details that were important at one point in time may disappear to be replaced by more important details. Wording might be changed to be more politically correct after a quick rush to get information out there. Sometimes errors or typos can get quickly changed, and those changes don’t always end up with a story-level correction to alert readers that something incorrect was once out there.

From the examples already posted to the site, you can see changes in news judgment and wording are reflected in later updates to particular stories — for example, a New York Times story that changed the phrase “single-party rule” to “single-party dominance.”

The display for comparing revisions is similar to what you’d see in a graphical interface for version control systems, like Git or Subversion (in fact, each revision is stored in a git repository). You can compare two different versions of an article side-by-side and see at a glance what was added or removed in each iteration.

For those interested in publicizing your own revision history (or tracking someone else’s), the code is available on GitHub. There are also a few WordPress plugins that do something similar for WordPress blogs, though they’re not nearly as extensive as NewsDiffs and you can, of course, only use a WordPress plugin on a site you own.