The Internet has changed our lives in almost every way. Wikipedia and other online knowledge repositories have had a significant impact on the way we learn. It has changed even the way that science is done. Social scientists are increasingly using online data to study individual or collective behavior on a scale that’s normally seen only in natural sciences.
We are far from having the large social science experimental data sets produced by CERN. But at least we have digital observations like those collected and analyzed for observational astrophysics. Online tools are used by millions of people every day. For example, Wikipedia is read 500,000 times a day.
Scientists call this ” Collective Memory“. It is the way members of a group remember an event from the past. Although collective memory is an important concept in sociology, there are very few empirical studies that have examined the topic. This is due to a lack of data. Scientists who study how people remember past events have traditionally spent a great deal of time and energy collecting data via interviews and surveys.
Plane crashes
In a study published in Science Advances by a team of a sociologist and two physicists, we used data from Wikipedia’s publicly available daily statistics on page views for all articles in the encyclopedia to study collective memories.
We looked at all aircraft accidents in aviation history (all the way back to Wikipedia). We chose to do this because these events were well documented and also because there have been many of them – which makes the statistical analysis robust.
We divided events into two categories: recent (2008-2016) and past (anything prior to 2008). Malaysia Airlines Flight 370, Malaysia Airlines Flight 17, Air France Flight 447, and Germanwings Flight 9525 are examples of recent flights. American Airlines Flight 587 and Iran Air Flight 655 are examples of past crashes.
Then, we used statistical methods to measure the increased page views of articles about past events one week after an event. This increase was called the “attention stream.” We wanted to know if there was a relationship between the similarity or timing of recent events and past events, as well as the increased attention paid to them. We wanted to find out if it was possible to predict how much attention will be delivered to past events when a similar event happens.
When the Germanwings Flight crashed outside New York City in November 2001, we found that people consulted Wikipedia to find information about the crash involving an American Airlines Flight. This page was visited three times more in the week following the Germanwings accident.
It seemed like a pattern. As a result, we consistently saw a significant rise in views for past events. In the week following an event, people viewed past events 1.4 times as much as they did recent ones. This shows that an event’s memory can grow over time, gaining more attention. Then, we tried to model the pattern, taking into consideration factors like the impact of recent and past events and the similarity of the events. We also looked at whether there were any links between the articles on Wikipedia.
What is the basis of our memory?
In the cases of American Airlines and Germanwings, the pilot’s role could have been a significant coupling factor. The American Airlines plane was destroyed by pilot error, while the German aircraft was deliberately crashed. We were even more surprised to see that these two Wikipedia articles had no hyperlinks between them. Our general results remained robust, even after we removed pairs of pieces that were connected directly by hyperlinks.
The original impact of a past event was the most important factor for memory-triggering patterns. This was measured using its average daily pageviews before the recent event. This means that certain past events are more memorable, and our memories of them are more easily triggered. Some examples of these events include the crashes that occurred in relation to the terrorist attacks.