A “nice paper”:http://papers.ssrn.com/sol3/papers.cfm?abstract_id=1642040 by Ruth Kricheli, Yair Livne and Beatriz Magaloni, which uses signalling theory to argue that protests are less likely in regimes that use violence against protesters, but that when protests _do_ occur, the regimes are more likely to fall.
bq. why are more repressive dictators a priori less likely to be overthrown by civil protest, but they are more vulnerable to civil protest once it actually occurs? The underlying idea here is that the task of coordinating and orchestrating mass protests in more repressive regimes is significantly harder, which makes civil protest less likely. But if citizens are willing to protest even under a highly repressive regime, their taking to the streets generates a very powerful and informative signal about the general anti-government sentiment in society. The protest’s information revealing potential is maximized when it is very costly for citizens to signal their opposition to the regime. This information is thus more likely to induce many more citizens to take to the streets in the future.
The authors find substantial statistical evidence to support these contentions. This speaks to some of the issues raised in “Josh’s post”:https://themonkeycage.org/2011/04/what_the_costs_in_a_cost-benef.html about Syria a few days ago. Obviously, there is no guarantee _at all_ that the Syrian regime will topple. However, the likelihood of regime collapse may be somewhat higher than it would appear to be at first glance.