Random acts of the Twitterverse

A visu­al­iza­tion of how #GOP spreads across the Twitter network…looks a bit like what I’d imagine “preaching to the choir” might.

Last week a coworker tried to explain the ins and outs of Twitter to me with little suc­cess. I get the point, really I do — it’s just that I find the infor­ma­tion over­load issue impos­sible to cir­cum­nav­i­gate. “You just have to ignore some of it,” she said.

This idea of a finite atten­tion span forms part of the foun­da­tion for pre­dicting the spread of ideas on Twitter, according to new research from Northeastern’s Stern­berg Family Dis­tin­guished Pro­fessor, Alessandro Vespig­nani and his collaborators.

In an article pub­lished on the Nature Sci­en­tific Reports web­site last month, Vespig­nani and his team showed that the social network’s struc­ture, cou­pled with our lim­ited cog­ni­tive capacity, has a larger impact on the spread of ideas than the rel­a­tive impor­tance of those ideas or the people Tweeting about them.

Imagine that all the infor­ma­tion is equal,” said Vespgi­nani. ” Well, what you’d expect is that the infor­ma­tion will be selected equally so that all the mes­sages will have the same life­time and the same number of users or spec­ta­tors.” But this is not the case.

The team designed a com­pu­ta­tional model that mimics the so-​​called “Twit­ter­verse” based on real-​​world data from 1.3 mil­lion hash­tags and 120 mil­lion retweets from 12.5 mil­lion users. They stripped away all of the external fac­tors, such as main­stream media and world events, that could change the “value” of each Tweet. The model included two assump­tions: first, that the net­work in the vir­tual Twit­ter­verse looks like the real one and, second, that our atten­tion span is finite (mine is clearly much shorter than the average user, just looking at the feed breaks my brain). The prop­a­ga­tion of Tweets across this mod­eled net­work matched per­fectly with the real world prop­a­ga­tion of Tweets.

Visu­al­iza­tions of Tweets related to the Arab Spring and the March 2011 earth­quake in Japan

When they changed the para­me­ters, for example by adding rel­a­tive values to the vir­tual Tweets or using a random net­work instead of the social net­work struc­ture, the model strayed from actual data.

An article on the Atlantic’s tech­nology blog says “the research sug­gest that it doesn’t fully matter who you are or how many con­nec­tions you have, but what you’re saying rel­a­tive to the existing con­ver­sa­tion is what really mat­ters in spreading knowl­edge.” Before talking to Vespig­nani, I tweeted this (!), but actu­ally it’s kind of wrong. In fact, it doesn’t seem to matter who you are, how many con­nec­tions you have, or what you’re saying. It’s much more random than any of that.

It’s like a neu­tral evo­lu­tion, Vespig­nani said, wherein some­thing evolves not because it’s better fit to sur­vive, but because of random cir­cum­stances. “And in some cases things that are very fit may just disappear.”

That doesn’t mean that infor­ma­tion doesn’t have value, he said. “But actu­ally this is a model that tells you you don’t have to invoke those prop­er­ties to explain what you see in terms of the life­time of ideas. What we believe is that most of what we see in the social net­works in terms of rumors, ideas, or the spreading of knowl­edge, is due to the sto­chas­ticity in social networks.”

But why do we care about all this in the first place? Vespig­nani, who is the local PI for a col­lab­o­ra­tive NSF grant with Indiana Uni­ver­sity to explore the spread of knowl­edge, can envi­sion a time when we’ll use the Tweets leading up to events to pre­dict their future occur­rence. For example, could we have pre­dicted the Arab spring or the out­come of an elec­tion based on what was taking place on Twitter just prior? Vespig­nani believes yes, once we under­stand how the system works we will be able to use it to pre­dict outcomes.