Much recent research has highlighted the success of non-violent protest. Erica Chenoweth and Maria J. Stephan’s data analysis had demonstrated that disciplined, non-violent protests succeed more often than violent ones, even in the face of repressive actions by regimes.
And yet recent events in Bahrain, Egypt, Syria, Libya, and Ukraine give one pause. While Tunisia offers the example of a relatively peaceful protest campaign that overturned a dictatorship, in Libya civil war seemed necessary to overturn a regime about to massacre peaceful protestors in Benghazi. In Egypt, the peaceful protestors who brought down the Mubarak regime were soon marginalized, with the Muslim Brotherhood now outlawed and suffering mass executions at the hands of a counter-revolutionary military regime. In Syria, the dictatorship responded to peaceful protests with brutalizing attacks and seems likely to have crushed the protests if they had not recruited defecting soldiers and become militarized (although to be sure, we do not know what would have happened if the protestors had stuck to non-violence). Finally, in Bahrain, the most massive peaceful protests seen in the region, as a percentage of the population participating, were crushed by the military. By contrast, in Ukraine, it was only after peaceful protestors were galvanized by more violent “ultranationalists” who attacked police and burned buildings that the ruler fled (although again we cannot be sure what would have followed if this turn to violence had not occurred).
No doubt non-violent protest can be surprisingly effective in causing a ruler’s security forces to defect and bring about the end of the regime. But obviously it is not always true that non-violence succeeds, or that violent protest fails.
One can conceptually divide regimes into three types: (1) regimes with major problems of loyalty among elites and security forces, such that any mass protests will likely lead to rapid defections that leave the regime defenseless; (2) regimes where ethnic or regional or professional loyalties of much of the security forces and of many business and administrative elites are so strong that they will fight effectively to defend the regime if it is threatened, but other security and elite groups would be glad to see the regime fall and so would support insurrection; and (3) regimes where the ties of loyalty of both the security forces as a whole and most of the financial and administrative and managerial elites to the regime remain quite strong.
In regimes of the first type, non-violent protest is in fact likely to be more effective for precisely the reasons that Chenoweth and Stephan make clear. Non-violent protests are likely to win sympathy and support from both domestic and external potential allies; they force the regime to delegitimize itself if it responds with violence; and they make it easy for the security forces to defect and abandon the regime as there is no immediate threat to the persons or property of groups close to them. By contrast, in this situation violent protests can alienate the broader population, antagonize the elites and provoke the security forces to act, and legitimate the use of violence by the regime to defend itself and others from danger. All of this is particularly true when the regimes in question have been trying to build a façade or even the true beginnings of greater democracy, and have built ties to the West; both of these elements make it more costly and delegitimizing for regimes to use deadly force against peaceful protestors, and more likely that regimes will be supported if they face violent pressures from below.
Yet the calculus is different in regimes of the second and third type. In the second type of regime, a protest is likely to lead to some defections from the military and elites, yet not enough to disable the regime, for a core of security forces and elites will still back it and urge it to fight hard to survive. In these cases, a fairly effective repressive response from the regime is likely to occur, and this response will not have the “backlash” effect of causing further elite and popular defections from its support. Instead, society will start to polarize, with some military and elite defections creating a potentially effective opposition capable of fighting and even overcoming the regime, but other members of the security forces and elites still prepared to fight to defend it. In this situation, violent opposition—guerrilla warfare or violent urban protests—are almost essential for the opposition to succeed. If they do not take up arms, the loyal and effective portion of the security forces will crush the revolt. But if they do take up arms and start to move toward victory, there are likely to be more defections and the exposure of the weakness of the regime. These will then pave the way for a successful revolution.
Finally, in the third situation, neither peaceful nor violent protests are likely to be effective. Either way, the regime will be able to isolate the protestors as a deviant minority, posing a danger to society, and call on loyal security forces to destroy the opposition.
One can readily call to mind examples of all three from recent events in North Africa and the Middle East. Type (1) regimes were found in Tunisia and Egypt; in both cases an unhappy military stood aside while a mass movement emerged to threaten the rulers, and non-violence triumphed. But in the second phase of the Egyptian revolution, when the highly politicized military sought to retake power (a factor absent in Tunisia), non-violent protest was brutally crushed by a military that had strong professional unity and loyalty and strong elite (and substantial popular) support. Type (2) regimes were found in Libya, Syria, and Yemen. All are far more tribal and sectarian societies than Egypt or Tunisia, and these cleavages were visible in the military’s response to protests. In Libya, the professional military stood aside, but tribal and mercenary forces (recruited from outside of Libya and therefore loyal only to Qaddafi) were ready to crush the protestors. Only violence—in this case greatly aided by NATO—could have reversed the tide. In Syria, a similar situation arose, but the professional military had stronger sectarian ties, and while mass infantry would not fight to defend the regime (many defected), air force and artillery units, reinforced by elite special forces, did so with vigor. NATO however, stood aside, and the Assad regime has been able to hold out and even start to win against what soon became a nationwide armed insurgency. In Yemen, we see yet another case, with civil war only averted by U.S. intervention, but still going on at a moderate level in certain regions and among certain sectarian groups.
Finally, in Bahrain and Iran, we see the third type of regime. These regimes were capable, in 2011 and 2009 respectively, of crushing very widespread non-violent protests. They did so because the Bahrain military (backed by Saudi Arabia) and the Iranian Revolutionary Guards and Quds militias were completely loyal to their regimes, willing to do whatever was necessary to keep their rulers in power. Popular protests in these countries provoked no widespread defection from either the security forces or the administrative and business elites (in Iran, even the opposition politicians in whose name the protests over election manipulation were made would not come forth to lead a call to overturn the regime).
I thus believe that the statistical result of Chenoweth and Stephan reflects the fact that type (1) regimes have become much more common since World War II, and that type (2) and (3) regimes have become less so. The decline of ideologically-based party regimes, the professionalization of militaries, and the reduction in regimes based on clan, sect, or regional loyalties have all contributed to this situation.
And it is certainly a wonderful trend! But it means that in assessing the odds that a non-violent protest will be successful, one cannot simply bet on non-violence. One has to also examine the kind of regime one is facing. For repression usually does not work—except when it does, as in the type (2) and type (3) regimes.