A burst for Bursts

Photo via Thinkstock.

Photo via Thinkstock.

The other day I starred the fol­lowing head­line in my RSS feed: “Any Two Pages on the Web Are Con­nected By 19 Clicks or Less.” I didn’t read it imme­di­ately because it sounded like vaguely familiar old news that I could prob­ably return to later.

But this morning in our office’s daily edi­to­rial meeting, I kicked myself because it turned out that article was about the work of one of Northeastern’s very own guys, Albert-​​László Barabási. It’s my job to know what’s going on with our researchers and I was miffed that I let this one slip by me. I imme­di­ately pulled out my phone and stealthily read the first few lines of the news story, which appeared on the Smithsonian’s blog Sur­prising Sci­ence:

No one knows for sure how many indi­vidual pages are on the web, but right now, it’s esti­mated that there are more than 14 bil­lion. Recently, though, Hun­garian physi­cist Albert-​​László Barabási dis­cov­ered some­thing sur­prising about this mas­sive number: Like actors in Hol­ly­wood con­nected by Kevin Bacon, from every single one of these pages you can nav­i­gate to any other in 19 clicks or less.

But wait, hadn’t I heard that.…before? Wasn’t this old news? By now I was having some serious déjà vu. So I went back to my office and down­loaded the article that pub­lished on Feb­ruary 18th, 2013 with this so-​​called new discovery.

Net­work Sci­ence” appeared in a spe­cial issues of the journal Philo­soph­ical Trans­ac­tions of the Royal Society A on Monday, but it’s not so much a research article as a sum­mary of a talk that Barabasi deliv­ered back in 2010 at a work­shop on the new sci­ence of the Internet. In the talk he pre­sented what he believes are the four guiding prin­ci­ples of the Internet:

  1. It's held together by a few highly connected hubs, rather than a random assortment of equally strong nodes (or websites).
  2. It's a "small world," meaning each site can be reached via a finite number of links from any other site (this is the "19 clicks or less" phenomenon).
  3. It expands through the addition of new documents, which are more likely to link to highly connected documents than otherwise.
  4. It grows in a systematic way, with more linkable nodes winning out over less well-connected ones.

All of these stemmed from pre­vious work that Barabasi’s lab had pub­lished over the last fif­teen years. The talk, and sub­se­quent paper, were merely a recap. The 19 clicks result, which gained some steam after the Smith­sonian article went live (ABC News, Slate and Mash­able all picked it up), was actu­ally pub­lished in Nature Mag­a­zine in 1999. Not Philo­soph­ical Trans­ac­tions. And not on Monday.

So, why on earth am I even blog­ging about this? I’m not the Sci­ence Jour­nalism Tracker and people are right to think that Barabasi’s research is cool, so why would I call atten­tion to their being 14 years late to the party?

Because the very fact that the story has gained trac­tion again is itself a rep­re­sen­ta­tion of the work the story is describing! It’s so delight­fully meta, I’m still get­ting goosebumps.

I was imme­di­ately reminded of a story that another net­work sci­en­tist, Hamid Ben­brahim, told me a few months ago: In Sep­tember 2001, United Air­lines filed for chapter 11 bank­ruptcy. The Chicago Tri­bune wrote a story about it and seven years later that story appeared on the Miami Sun Sentinel’s home­page without a date for unknown rea­sons. That same week, Hur­ri­cane Ike was on track to strike the Florida coast. Pos­sibly because people were searching for info about their flights to Florida, the Sen­tinel story started get­ting thou­sands of hits.

On Monday, Sep­tember 8th, 2008, a jour­nalist working for Bloomberg News searched for news on recent bank­rupt­cies. The undated Sen­tinel story appeared in his results. Not really thinking about the cur­rent state of the air­line industry, which was fine, he grabbed the head­line and posted it to Bloomberg’s home­page. Eight min­utes later, United Airline’s stock plummeted.

If Barabasi’s 1999 Nature article could have somehow impacted the stock of some random air­line, the Smith­sonian story could have caused some serious, undue damage just like the Sentinel’s did in 2008. Luckily, all it did was give Barabasi a little burst of well-​​deserved attention.

And, hey, here’s a funny con­nec­tion: “Bursts” is the title of his 2010 book whose sub­title reads “The Hidden Pat­tern Behind Every­thing We Do.” In it he describes how humans are a “bursty” species. We eat in bursts, usu­ally around the begin­ning, middle and end of the day­light hours. We check our emails in bursts, rather than every five min­utes across the day’s 24 hours (well, most of do). We cook, sleep, make phone calls, and surf the web in bursts. As a result, arti­facts of humanity are bursty as well. The arti­cles we read go through little bursts of pop­u­larity, get­ting a whole bunch of readers on the first day of pub­li­ca­tion and then sub­se­quent bursts at seem­ingly random points throughout their even­tual history.

Barabasi and his team looked at bursts on a Hun­garian news portal called Origo​.hu. In Bursts, he writes the following:

…two researchers in my group at the Uni­ver­sity of Notre dame—graduate stu­dent Zoltan Dezso and post­doc­toral asso­ciate Eivind Almaas—set out to answer this simple ques­tion: how long is any par­tic­ular news item accessed at a given time at Origo​.hu? In other words, how long is each story’s fif­teen min­utes off fame really? To answer the ques­tion, they first deter­mined the number of people who clicked on a par­tic­ular article each hour. Not sur­pris­ingly, about 28 per­cent of traffic occurred during the ini­tial twenty-​​four hours after an article was pub­lished online. On the second day, there was a dra­matic drop-​​off in read­er­ship, accounting for a mere 7 per­cent of the article’s total hits.

This makes sense, of course: if a piece of news inter­ests you, you tend to read it as soon as you come across it while trawling your favorite sites. And by the third or fourth day every­body who had any interest in the piece had a chance to look at it, thus the vis­i­tors should have ceased arriving. The problem was that they did not. Instead most arti­cles on the portal con­tinued to be read many times over, many days after their ini­tial publication.

This was some­what mys­te­rious. First of all, how did people even come across these arti­cles, many of which had dis­ap­peared from the front page days prior? Second, why on earth would any­body be inter­ested in old news?

Yes, indeed. Why on earth would any­body be inter­ested in old news? For the details of Barabasi’s answer, you’ll have to read the book (shame­less pitch, but it’s worth it, I promise!). In the mean time, I’ll leave you with this thought: Isn’t it cool that every web­site on the web is con­nected to every other through just 19 clicks? Did you know that before today? Here’s an article where you can read all about it (it was pub­lished in 1999).