Nicholas Davis and Florian Ramseger are guest blogging for the Forum. Nicholas Davis is Associate Director, Deputy Head Strategic Risk Foresight and Florian Ramseger is a Research Analyst from the Risk Initiatives team.
The role of the internet in general and social media in particular in driving protests in the Middle East and Northern Africa (MENA) has been a matter of some debate. Commentators such as Clay Shirky and Don Tapscott argue that social media has ushered in a “revolution in revolutions”. Others, such as Malcolm Gladwell disagree, arguing that age-old dynamics are driving current trends, with technology simply a new channel rather than a radical shift in behavior.
However, one thing is clear: the events that started in Tunisia in December and have since spread to many other countries are having an impact on the web browsing behavior in the region. And it is reasonable to assume, as Shirky and Tapscott do, that shifts in patterns of technology use that surround major events such as the current protests across the MENA region create feedback loops that influence behavior in new ways.
To investigate this, the Forum’s Risk Initiatives team looked at specific search-query terms that people in the MENA region and around the world entered into Google search over the last few weeks and months. By using Google Insights for Search, it is possible to limit search queries by country, giving us a local proxy for how populations responded to social unrest. A number of very interesting patterns are visible.
First, take Tunisia, were it all started (see Figure 1). The search terms “manifestation” (protests), “gouvernement” (government) and “Ben Ali” all saw a sustained increase in late December when demonstrations that had initially been confided to the city of Sidi Bouzid started to spread to other towns including Tunis. In the run-up to the resignation of president Ben Ali in early January these search terms really took off.
Figure 1 – Tunisia
In Egypt, too, internet search behavior reveals interesting trends (Figure 2). First the search term “الحكومة” (government) started to increase slowly in early January. Then around the 15th January the search term “مظاهرة” (demonstration) exploded, spiking on the first day of the Egyptian protests. What happens then to those variables is also highly interesting. For four days, the graphs drop to zero, most likely due to the widespread internet outage in the country. The fact that the search term “Mubarak” does not exhibit the same behavior is particularly suspicious.
Figure 2 – Egypt
Absence of a signal is a signal
The Egypt example, when combined with the commentary by Shirky and Gladwell, shows that the internet and social media are a double-edged sword for governments experiencing social unrest.
On one hand, they are a powerful tool for communication and ad-hoc organization, leading regimes to consider censorship as a tool of controlling populations. However, as Mark Turrell pointed out, in the case of Egypt it could have been the shutting down of the internet and phone systems that was the real tipping point for revolution. In network theory we talk about inactive and active nodes that create contagion and spread a signal through a system. In the domain of social protest, there are usually only a limited number of “active nodes” who get out on the streets and protest – students, minorities and other aggrieved parties. However when the Egyptian government shut down all internet and mobile phone communications to deprive these limited nodes from the means to communicate, they inadvertently activated a whole other series of nodes – the majority of people simply unable to go about their lives as usual and enjoy the freedom to converse. Reports that the employed middle class, children and the elderly were involved in protests was a strong signal that the protests were engulfing entire communities.
So, to return to the question of how important the internet is in driving protests across the MENA region, what we’ve seen to date indicates that Malcolm Gladwell is right – in the absence of the internet, age-old social dynamics are at the root of social protest. But the evidence also supports Clay Shirky and Don Tapscott’s views – without the existence of social media and the internet leading up to these events, we may not have seen as widespread a social reaction. And certainly, people in the rest of the world would have known a lot less about what was happening, as it happened.
The problem of motives
Of course simply looking at how often words like “demonstration” were googled in different countries, cannot infer what actually motivates people to search for these terms. The data do not distinguish between someone who is using the internet to organize a demonstration and someone who is just curious about what is going on. Nonetheless, the pattern seems to be that, in the lead up to major events, certain search terms see an increase above their respective long-term trends. Thus, the question arises whether it might have been possible to predict the Tunisian and Egyptian revolutions just by tracking those variables.
Who is next?
The logical next step is to see whether such search data might indicate which country will see the next overthrow of government. In Algeria, people increasingly googled the French words for “government” and “demonstration” in the run up to the first major demonstration (figure 3). But since the protest started these trends have abated again. Censorship could explain some of this, but we can’t say for sure. What does seem to indicate that the Algerian government is not off the hook, though, is the continued elevated level of the search query “Bouteflika”. The president of the People’s Democratic Republic might have mixed feelings about his sudden popularity on the internet.
The data gives more straightforward results for Morocco (Figure 4) and Iran (Figure 5) where the French and Persian equivalents of “demonstration”, respectively, have shot up in recent days to reach their highest ever levels. These are clearly countries to watch.
Figure 3 – Algeria
NB: Only weekly averages are available for Algeria
Figure 4 – Morocco
Figure 5 – Iran
NB: Only weekly averages are available for Iran
Hold your horses
Unfortunately no predictions can be made for other countries, such as Jordan, Libya, or Bahrain. The reason is that there simply is not enough data available. And even for those countries where we have data, we should stress that all figures are clearly to be taken with a pinch or more of salt. Since Google only makes relative search figures available (How often a particular word is searched relative to all other search terms, scaled in relation to its highest peak.) we can’t say how large the sample sizes are that we are talking about. Add to this the above-mentioned problem of censorship, and it becomes clear that you don’t want to bet your house on these trends. But one thing is clear: what is happening in the Middle East and in North Africa is happening on the web, as well as on the ground.
 It would certainly be interesting to investigate the actual motives in more detail, for example by looking at what other search terms were used together with the keywords that we looked at. Another possibility would be to see whether Twitter data can give one more insights on the context of users’ internet activities. The limited amount of publicly available data might make these more refined approaches difficult at the moment, but these are certainly ideas for potential exploration in the future.
 To be precise, searches that also include the other countries, Egypt and Tunisia, were excluded.