A few days ago, the Monkey Cage Blog ran a guest post by sociologist Amy Austin Holmes that wrestled with the question of why Egyptian security forces perpetrated a massacre in broad daylight at Rabaa a year ago. The Rabaa massacre is just one part of a larger episode of state-led mass killing that began in Egypt after the 2013 coup and continues today. As Holmes describes,
Why would Egyptian security forces deliberately kill hundreds of unarmed countrymen, women, and children? One prominent school of thought on the causes of mass atrocities sees state-led mass killing as an instrumental response to threats to incumbents' power. Other things being equal, rulers are more likely to kill en masse when they believe that rebels or putschists might soon topple them. Consistent with this view, Holmes sees the security apparatus's post-coup violence as part of an unbroken pattern of efforts by the military to beat back challenges to its dominant position in Egypt's political economy.
On its face, the military's decision not to fire on protesters in February 2011 seems to contradict this argument. If security forces were so bent on sustaining the existing regime, why did they allow President Hosni Mubarak to fall? As Holmes points out, though, the fact that the military didn't get directly involved in violent repression in the early days of the revolution doesn't mean that repression didn't happen, or that the military didn't support it when it did.
Equally important, Hosni Mubarak was not the regime. In retrospect, it's easier to see the military's decision to let the president go as a sacrificial gambit that ultimately succeeded, or at least didn't fail. After Mubarak's departure, Egypt's military leaders took direct control of national government through the SCAF. When the candidate they presumably preferred lost the presidential election in 2012 to the Muslim Brotherhood's Mohammed Morsi, Egyptian military leaders didn't follow the script written by their Algerian counterparts in 1992. Instead, they patiently worked with opposition leaders to encourage and facilitate a wave of protests that gave the ensuing coup more diplomatic wiggle room and shaped insiders' expectations about the coup's likely success. With hindsight, the larger strategic objective behind all of this surface zigging and zagging seems clear.
Of course, the presence of a threat to elite power and the military's apparently calculated response to it do not rule out alternative explanations. The other main school of thought on the origins of mass atrocities locates the impetus to killing in the character of the prevailing political system. Authoritarian regimes led by rulers who espouse an exclusionary ideology and whose ethnicity is politically salient are seen as unusually prone to mass violence, especially when they are internationally isolated. In Egypt, rulers' ethnicity is not a political issue, and the country is not internationally isolated, but the Mubarak regime and the authoritarian "deep state" on which it rested have long toed an ardently anti-Islamist line. Viewed through this lens, Egypt also looks like a relatively risky case.
Ultimately, it's impossible to say with confidence which of these factors contributed exactly how much to the occurrence of this episode of mass atrocities in Egypt. What we can try to do, though, is learn which of these schools of thought—the "elite threat" and "bad regime" frameworks sketched here—does better at predicting future episodes and then infer explanatory power from predictive power.
In fact, that's one of the things we're trying to do with our statistical risk assessments, which combine forecasts from models representing both schools of thought, along with a machine-learning approach that amalgamates the two. As it happens, neither of the theory-specific models identified Egypt as an especially susceptible case in 2013, but the machine-learning amalgamation of them did. And, as a result, so did our top-line forecasts.
We shouldn't infer too much from a single case, but this instance only strengthens our belief that an ensemble approach—combining forecasts from multiple models that represent different points of view and use different methods—is the right way to go. There are many pathways to state-led mass killing. To anticipate them as accurately and early as possible, our theories and early-warning systems need to reflect that equifinality.