“When Google News was launched over a year ago there was mild nervousness amongst journalists. Was this a new kind of news site that might beat them at their own game? …We checked to see how well Google News, compiled “without human intervention”, picked the top two stories chosen by leading US news sites.
…
Google News initially picking the Top 2 stories of the moment around 59% of the time, improving to around 63%. CNN and Yahoo! News hover throughout at around 77%, a remarkably consistent performance.
Over the year Newsknife has probably studied Google News more closely than most people. We are a little surprised at the results. We thought Google News “hit rate” at choosing the popular top two stories might have increased more over the year, especially as Google are still calling Google News a BETA (not final) version. This suggests they’re fine tuning it.
Actually, we’re surprised Google News is still a BETA version after over a year….”
Newsknife has some ideas about how Google News can increase its ‘hit rate’ through automated processes, but tells human editors they don’t have to go back to the job market just yet. I agree with their observation that Google’s strength is pulling together a massive cluster of links to news pieces covering a single story, but that it fails a version of the Turing Test by being unable to appreciate nuance in news coverage. The article gives an example:
For example, if a terrorist bombing has just occurred the first wave of news should give details of the bombing. The second wave might be reaction from around the world. At what point should reactions be introduced? And some reactions are more important than others.
In Newsknife’s opinion this is a challenge for Google News.
