I had a rare chance to hear the creator of Google News, Krishna Bharat, this weekend explain how the service's algorithm picks the topics and stories on its home page. While he didn't give away all his secrets, the Google principal scientist did outline why certain articles get higher placement than others.
Generally, he said, articles are ranked based on originality, freshness, quality, expertise of source and whether a lot of other sources around the Web are pointing to a particular article. Of course, what's of interest to those of us hoping to get our stories into Google News (as opposed to folks like Associated Press who are trying to keep their content out) is how the algorithm makes those determinations.
Most of that remains part of Google's secret sauce, but Bharat did share the criteria for rating a source's reputation:
* Volume and originality of content produced consistently about a topic. For instance, based on what it posts on its site, he noted, espn.com would likely be seen as an authoritative source about sports but not, say, about business.
However, a site that simply posts a lot of content straight from other sources without generating original articles of its own would, at least in theory, have a lower reputation than one that produces a lot of its own articles -- good news indeed for his audience, which was mostly content-creating writer and editors like me.
* Links around the Web. Do a lot of other sites link back to that source? That's not as straightforward as it sounds, though, since both Bharat and other speakers at the Online News Association's search engine optimization panel intimated that quality as well as quantity of links matter.
For example, another panelist after the session noted that lots of links in syndicated content across the Web might not be as valuable as other types of links, if a search engine's algorithm determines that there's a business relationship between two sites.
* What users do in response to links to that source on Google News. Are some sources' links clicked on more often than others?
Google News search results (as opposed to what shows up on the hone page) are based on several factors, he said: relevance to the user's query, "editorial interest" (i.e. how many editors consider certain stories containing the term more important than others), page placement (have editors put a story on a site's home page or just buried it somewhere deep inside the site?), freshness and -- not sure this is a word, sorry -- localness to the user. On the one hand, Google News likes to expose more than just U.S.-based sources on its home page; on the other hand, he said, localized country Google News pages tend to favor local sources in results.
One of the most interesting points raised during Q&A is whether Google News promotes herd journalism at the expense of innovative investigative pieces. The questioner didn't quite put it that way, but complained that Google News touts the popular at the expense of innovative investigative pieces -- especially since competitors sometimes won't link to or otherwise acknowledge great work on a story that they themselves missed.
Bharat acknowledged the issue, and said the Google News Spotlight section is a way to try to deal with that. There are a few problems with that, though.
For example, if you've already customized your Google News page, you may not even know there is a Spotlight section. And, given that the top Spotlight stories when I checked this morning were 1) kidnapping victim Elizabeth Smart's testimony and 2) a Michael Jackson autopsy report piece, I'm not sure Spotlight is necessarily uncovering unique enterprising journalism (although the number 3 story was an interesting New York Times piece about heart-valve research). I guess we'll have to hope that lots of other sites link back to great, unique, enterprising articles -- even if they're not about a celebrity or highly publicized crime.
Follow me on Twitter: @sharon000