A few months ago, I had the pleasure of attending a lecture by Carlos Castillo of the Qatar Computing Research Institute on news, and social media. I’d come under the impression that it would be about using social media to discover and generate news, but it was more interesting than that. Mr. Castillo’s presentation and research occupies an intersection between Computational Linguistics, social media, and news. He uses social media signals to identify the lifecycle of a news article, and its relevance to an audience.

It left me wondering how we can use the signals in both computational language analysis and social media analysis to keep people better informed. I don’t mean this in terms of volume of information, but accuracy of information. Right now, I can see the insights of Mr. Castillo’s work being mis-applied to increase the reach—and ad views—of a story rather than promote real journalism. To put it another way:

Truth has never been an essential ingredient of viral content on the Internet. But in the stepped-up competition for readers, digital news sites are increasingly blurring the line between fact and fiction, and saying that it is all part of doing business in the rough-and-tumble world of online journalism.

— If A Story is Viral, Truth May Be Taking a Beating

A number of the fake news stories in the New York Times piece I just quoted and linked to are “soft news” at best. They succeed in their mission of getting attention, measured in tweets, shares, likes, and the all important Page View, but they are not journalism in any legitimate sense.

But there is a desire among people to read real news. In Mr. Castillo’s lecture, he noted that long-form journalistic pieces have a long timeframe of relevance and traffic, up to a week, while breaking stories tend to have a lifespan of about nine hours with an intense first hour. For both, the amount of traffic and social media a news piece gets in the first hour is the best judge of its relevance. QCRI’s demo site provides a good visual explanation. Green bars show the predicted page views for an article of Al-Jazeera news based on existing traffic and social media signal. The articles are primarily hard news, as that’s the bulk of what Al-Jazeera produces, but the source doesn’t matter. The same algorithm would work for Huffington Post, MSNBC, Buzzfeed, or Fox News.

While people will share and click for hard news and soft news alike, soft news has the risk of spreading misinformation. This can’t be good for society, right? Well, to quote a quote, “Even if it’s fake, it’s real.” Is there value in the fake news, engineered for pure virtality, to spawn discussion? Potentially. I haven’t seen much discussion around viral news stories except for people complaining about the viral news stories in their feed. This could just be a function of the online circles I travel in—jaded and cynical tech people, often former news junkies themselves. [1]

The New Yorker recently published a piece trying to determine what stories go viral, and why. At the risk of spoiling the article, Aristotle may have had the answer already. “The answer, he argued, was three principles: ethos, pathos, and logos. Content should have an ethical appeal, an emotional appeal, or a logical appeal. A rhetorician strong on all three was likely to leave behind a persuaded audience.” I’m not so sure about persuasion in the Internet age. It’s entirely possible to live online in an echo chamber of voices that are similar enough to yours that almost nothing counter to your worldview can permeate.

It’s that “almost” that makes things interesting. If viral news stories have a spread that can transcend, or at least bypass, the social filters in our online lives, and they can spawn constructive discussion, we may be on to something. In the technology world, The Verge’s Fanboys piece is extremely viral, and the discussion surrounding it constructive. The Verge could be creating a template for a story that forces people to think about a contentious issue, and if it gets even one obnoxious online “fanboy” to think about their loyalties and behavior with a little more nuance, it’s a win.

What is clear, is that virality cannot be forced, but it can be engineered. Fanboys may not be as engineered for virality as the stories on Upworthy and Buzzfeed, though the clever layout tricks it employs show that a lot of thought was put into how people will see it. This brings me full circle, to the research from QCRI and Carlos Castillo. Predictive analytics can be a valuable tool to make sure that, should an editorial team with a focus on elevating discussion and making an impact want, they can engineer a story that can go viral and spawn real discussion. The cynic in me, however, expects it to only drive pageviews and increase ad revenue. There’s no reason it has to be either one or the other, though.

Either way, you’ll still get jaded former news junkies complaining. Just maybe not as many.