On 19 May, EgyptAir Flight 804 crashed into the Mediterranean Sea, killing all 56 passengers and 10 crew members aboard. The Wikipedia entry documenting the disaster went up within hours, and it will likely remain online into perpetuity. Human readers, however, lost interest after about a week. A pair of new studies reveals that’s common whether an aircraft crash kills 50 people or 500—a finding that reveals some surprises about our online attention spans.
With the vast knowledge of humankind mere touchscreen taps away, it’s fair to say websites like Wikipedia represent a kind of boundless augmented memory for humans. Yet in the face of these near-infinite data, human attention remains relatively short. Social scientists refer to this gradual loss of interest in a topic as attentional decay. “The internet offers limitless knowledge, but it doesn’t solve the problem of our limited attention spans,” says the study’s lead author, Taha Yasseri, a computational social scientist at the University of Oxford in the United Kingdom.
Yasseri was initially drawn to look at the online behavior surrounding aircraft crashes after Germanwings Flight 9525 went down in the French Alps last year. Investigators eventually determined the pilot committed suicide, taking the lives of all 150 people on the plane. Curious, Yasseri checked out the page views—a running tally of everyone who has visited the site—for the disaster’s Wikipedia entry and noticed that interest peaked in the days following the crash, then rapidly dropped off after about a week.
He began looking at other crashes both big and small and noticed similar page view patterns for all of them. Yasseri and his Oxford colleagues decided to formally analyze the data and built an algorithm that parsed data on all pages on aircraft crashes in Wikipedia’s two most popular language versions, English and Spanish. It then spit out a report on when each page was created, when it was edited, how many times it had been read, and when those page views happened.
The researchers found that when a crash involved fewer than 50 deaths, Wikipedia readers tended to pay relatively little attention. The researchers called these “low-impact events,” and found they were statistically insignificant to readership patterns. Crashes with more than 50 deaths, on the other hand, were identified as high-impact events. For these, readership jumped and scaled up according to how many people were killed and the continental region in which they occurred.
Crunching all the data, the team put together a table describing how much attention readers from each language version of Wikipedia paid to various high-impact crashes. A strong regional bias emerged: English Wikipedia readers were much more likely to read about North American and European crashes, and Spanish Wikipedia readers were more likely to read about Latin American crashes. Page views on English and Spanish Wikipedia for North American and Latin American aircraft crashes, respectively, were each about 50 times greater than for African crashes.
But as the researchers report today in the journal Royal Society Open Science, regardless of where the crash happened and how many people were killed, on both English and Spanish Wikipedia, page views dropped in half between 3 to 10 days after the event.
This happens for two reasons, Yasseri explains. First, there’s a “decay of novelty.” New information is appealing, and we naturally lose interest after it stops being new. And second, humans can only pay attention to so many topics at one time. As other things happen, newer events outcompete older events for our attention.
Although Yasseri says he expected to see attentional decay, he was surprised that the scale of the disaster didn’t seem to influence how long it took for people to move on. The fact that it was seen in both language versions hints at a universal rule for online information consumption—big disasters don't capture our attention any longer than smaller ones.
Maybe that’s not such a bad thing, according to Giovanni Luca Ciampaglia, a research scientist at the Indiana University Network Science Institute in Bloomington who studies collective social phenomena on the internet. “I refrain from making a value judgment over whether this short attention span online is good or bad,” he says. “But maybe it signals that we are able to just move on and pay attention to other things as they come up. In a certain sense, it’s kind of necessary.”