When the Arab Spring ‘officially’ began, in December of 2010, Twitter had approximately 50 million monthly active users (MAUs). Today, that number has jumped to over 300 million. That’s 300 million people around the world tweeting about their locations, movements, grievances, and plans, and that vast trove of data has not gone unnoticed. Social scientists have been looking into whether or not it is possible to mine Twitter for this exact type of information in order to predict mass protests for several years now, and even the UN recently got into the action by launching Global Pulse. If it is indeed possible to predict large-scale disruptions using publicly available Twitter information, then there may be a way for governments or international organizations to leverage that information to prevent major crises from occurring.
The answer to whether or not Twitter can accurately be used to predict social unrest seems to be a resounding ‘maybe.’ Optimists, like MIT researcher Nathan Kallus, argue that it is absolutely possible to predict crowd behavior by mining Twitter for certain key phrases, a claim he demonstrated by analyzing the 2013 deposition of then-Egyptian President Hosni Mubarak. Kallus and others like him want to believe that the publicly available information on Twitter can be harnessed to make predictions about the future: in 2012 the US government launched project EMBERS (Early Model Based Event Recognition using Surrogates), which tracks human behavior overseas, and Global Pulse has been involved in a variety of big data development initiatives since 2009. Recent human crises, such as the Baltimore riots and the unrest in Ukraine, have also been the subject of much analysis, with many researchers arguing that these events were easily predictable through Twitter mining.
If these protests are indeed predictable, then the next step would naturally be prevention. If a government was able to predict a large, disruptive riot and figure out what the trigger was – rising food prices, religious oppression, anger at a military action, etc. – then they could potentially find a way to mitigate unrest before it reached a tipping point. For example, ‘smart cities,’ which integrate information and communication technology to manage assets and, ideally, improve quality of life for inhabitants, could automatically dispatch police officers to areas with an uptick in negative Tweets. Countries could share data on trends within their borders with the global community, and international organizations could airlift supplies to areas struggling with food insecurity, or send in peacekeeping troops in order to prevent a military coup.
Yet for all the benefits that utilizing Twitter predictions to prevent large-scale protests could, there are also challenges that would inevitably result. The first problem is the accuracy of the predictions. The events being studied, like the Arab Spring and the riots in Ukraine, have already happened, and it’s far easier to know which keywords to flag or which date range to track when you already know the outcome. Moreover, it’s difficult to know who is telling the truth on Twitter and who is simply venting. Predictive analytics tools in the hands of repressive regimes also presents a challenge, as it would be easily possible to spread rumors about riots in order to make arrests or cause chaos. Alternatively, they could also intervene and shut down planned demonstrations before they even begin – not just before they escalate to violence – and stifle civic and political expression. An ideal tool would be able to distinguish between protests with the potential to be violent or disruptive and more peaceful protests, but it seems unlikely the technology has reached that level of nuance or will in the future.
Invasion of privacy is another concern: the idea that an international organization like the UN could not only keep an eye on public affairs around the world, but intercept protests and other movements affects the basic right of a nation-state to control its borders. On a more micro level, it would affect the basic rights of individuals to express themselves, and could easily lead to a ‘Minority Report’ scenario where someone exercising their fundamental right to freedom of speech by venting on Twitter suddenly finds themselves in prison.
As technology evolves and Twitter continues to grow, the fields of predictive and preventative analytics will grow along with it. We have now reached a point where it is necessary to think harder about how Twitter mining can be harnessed proactively to reduce the negative consequences of riots and protests on the global stage, while at the same time mitigating its associated challenges. Safeguarding individual rights to freedom of expression and privacy should be paramount, as should protecting nation’s sovereign rights. It’s a difficult balance to strike, but moving backwards away from these advances in technology would be even more difficult.
Image: 2011 Egyptian protests (credit: Essam Sharaf/Wikimedia Commons)