How Twitter Explains Egypt’s Bloody Politics

As Egypt’s political crisis has swelled in recent days, key actors ranging from ousted Islamist President Mohamed Morsy to the opposition Tamarod movement have taken to Twitter to stake out their positions in the conflict. Now, a team of researchers is mining hashtags on the microblogging service to monitor those very tensions.

Starting with a set of Egyptian Twitter users who self-identified as either secularists or Islamists, the researchers — Ingmar Weber and Kiran Garimella of the Qatar Foundation’s Computing Research Institute (QCRI) and Alaa Batayneh of Al Jazeera — tracked the repetition of hashtags and how closely they were associated with one group or another. For example, Weber told Foreign Policy by Skype from Doha, everyone uses the hashtag #FF, or “follow Friday,” which gives it a neutral polarity rating on their Political Polarization Index. #Coup, on the other hand, has been favored by jilted Islamists over the past week, while liberals prefer #revolution. Here are some other examples of polarized hashtags:

The researchers then plotted polarization over time, crunching the data for 17 million Egyptian tweets (the higher the number on the y-axis, the more polarized Egyptian Twitter users are).

The chart above shows the level of political polarization among Egyptian Twitter users from March 2012 through July 3, 2013. That first peak, in April, coincides with violent protests over military rule, the second with Egypt’s constitutional crisis in November. Starting about a month ago, you can see a crescendo of polarization leading up to the Tamarod protests on June 30.

The big question looming over the study is whether the data could have foreshadowed the contentious and, at times, deadly protests that erupted across Egypt over the past week and a half. As Patrick Meier, the director of social innovation at QCRI, wrote on his blog, iRevolution, “this index appears to provide early warning signals for increasing tension.” Weber, who worked on similar projects tracking online trends during the 2012 presidential campaign in the United States, said he’d like to see the model applied to other countries experiencing political upheaval.

Finding the right formula for turning social media into a crystal ball is a growing field of study (in March, for instance, Businessweek reported on research underway at Sandia National Labs to cull predictive data from the Internet). And while Weber is quick to point out that QCRI’s Political Polarization Index doesn’t predict events, it can demonstrate when tensions are running high — and when things are most likely to escalate.