Feb. 8, 2012 at 3:52 PM ET
This is a blog post about the sexy social media technology Twitter. It mentions Justin Bieber. You'll want to tweet it. At least, my editors hope you do. My job might depend on it.
The Internet and social media have altered the face of journalism. Few media companies can survive selling ads in traditional newspapers and magazines that readers will see as they flip pages in search of content that tickles their fancy.
Online, which is where most of us get our news today, millions of readers click links on Twitter to go straight to the content they want. That means the specific article must sell the ad. In turn, the dollar (or cent) value of a story is measured in the eyeballs it attracts.
Thus, in order for a media outlet to make a buck in this new world of journalism, editors and journalists must fine tune their story selection and writing style to maximize its spread on Twitter. Social media researchers at Hewlett Packard have developed an algorithm that does just that.
"In principle, there is a formula, an algorithm, that you can apply to any news story you write [to maximize your exposure] on social media," Bernardo Huberman, a senior fellow and director of the social computing lab at Hewlett Packard in Palo Alto, Calif., told me Wednesday.
The formula is a mixture of three main characteristics: its source, subject matter and the popularity of the people mentioned. It predicts how many tweets a story will get with 84 percent accuracy.
Huberman and his team created the formula after examining data on story content from the news aggregator Feedzilla during a week in August 2011 and studying how these stories spread on Twitter. Interestingly, they note, the level of subjectivity in an article isn't a big factor in its popularity.
The most popular stories are those published by technology news sites, about gadgets and social media, and include gossip about well-known celebrities. By this reasoning, a scandal involving an iPhone and Justin Bieber posted on Mashable would do exceptionally well.
The bias toward technology-related stories and sources, Huberman notes, may be because people who use Twitter "are very, very keen on technology."
Overall, the formula matches what editors and journalists already intuitively know: Sex and scandals sell, especially scandals that involve somebody with name recognition. What surprised Huberman was the degree to which all of this is predictable by a computer.
This predictability could lead to a software program loaded on journalists' computers that examines every story they write and tells them how well it will perform on Twitter. It could also recommend ways to improve a story's Twitter score.
One of the concerns is that "if everyone starts using this algorithm, all news stories will start looking the same," Huberman said. Even more troubling is "stories that might be important but don't have these characteristics will drown. No one will notice them. That's sad."
But it is also possible that journalists can use the formula to jazz up a story that would likely drown by highlighting or incorporating elements known to make it a Twitter success.
An argument can be made that the role of journalism isn't about success on social media. Huberman, for one, agrees with that sentiment. But he is interested in what he calls social attention — how to get people to pay attention to whatever you want them to pay attention to.
"The success of a story, whatever the story is, depends on being attended to by people to read it and pass it on," he said. "You can have the most incredible thing in life, a story, or something to buy or sell, but if nobody notices it, you might not be able to do anything with it."
Findings are to be published in the Association for the Advancement of Artificial Intelligence. A pre-print is available from arXiv.org.
More stories on Twitter:
For more of our Future of Technology series, watch the featured video below.