I guess from a technical perspective I can admire what the folks at Newsbreak have tried to build when it comes to automating news aggregation, but from a journalistic perspective, their inability to distinguish between two cities with the same name located in different states is a real problem.
Take this sponsored Newsbreak Facebook post that is prominently appearing when you search for the name of our newspaper right now:
It declares with journalistic certainty that the Richmond, Indiana Walmart store is going to close later this month. And Facebook “helpfully” surrounds it with posts and imagery from our newspaper, which at least for local searchers on Facebook, lends some credibility to the Newsbreak info by proximity alone.
The store closing declaration has been picked up by a number of local folks on Facebook, and there’s now growing speculation about why Walmart is leaving town.
The problem is that the Richmond, Indiana store isn’t closing. (I confirmed this with a Walmart media relations representative.) A Walmart store in Richmond, Virginia is on the list of upcoming closures recently announced by the company, but Newsbreak’s algorithm apparently can’t tell the difference, and decided this was big Richmond, Indiana news.
This isn’t an isolated incident. The Newsbreak Richmond, IN feed is full of stories that mention the word Richmond but are from other parts of the world. Who could have possibly anticipated that there’s a Richmond Farmer’s Market in California?
I’m all in favor of building smart news aggregation tools to help people find information — I built and run one myself! — but doing it in a way that leads to the rapid spread of incorrect information by design, and actually setting up automated purchasing of ads to ensure that happens even faster, is just plain dumb.
Worse, it feeds a narrative that news publications are inherently full of errors, damaging the credibility of the work that newsrooms powered by human reporters do every day to try to get things right.
Newsbreak (and anyone else using similar algorithms) should stop publishing until it fixes this.