(Page 2 of 24) - Of Social Networks and Popular Rebellion authored by Siegel, David.

I Introduction

Popular rebellion is notoriously difficult to model. On the individual level,

motivations for or against taking part in such a risky action vary greatly from person to
person. While some might choose to take to the street or to take up arms because of a
strong moral obligation to match actions to beliefs (Kuran 1991), others might do so in
pursuit of more material selective incentives (Lichbach 1994). On the other hand, some
might choose to remain detached from the movement, perhaps because they possess a
prosperous career tied to the regime under assault, or perhaps because they are simply too
afraid of being killed/imprisoned/exiled (henceforth euphemized as “removed”) to rise up
in opposition to something or someone viewed as oppressive. Such variance makes a
model based on individual motivations difficult to come by.

On the group level, we run into the problem of data availability. Mass political

violence is a relatively rare occurrence, and observations of it tend to be clustered within
a narrow range of institutional structures, making statistical generalizations across more
than a few forms problematic. Further, institutions generally interact with individuals in
complex ways, engendering more complexity throughout. Combined, these issues imply
that even excellent theoretical and empirical studies of mass violence tend to be primarily
applicable to a restricted geographical area or historical period.

Despite this, occasions of violence and risky protest—defined here as those in

which there is a genuine risk of serious reprisal to the protestor for her actions—do betray
some generalities. Though rare events, they often occur quite suddenly, and often
unexpectedly as well. Popular rebellions, the study of which is the focus of this paper,
are frequently leaderless as well (Kuran 1991; Koopmans 1993; Lohmann 1994; Opp,
Voss et al. 1995). Since the backbone of any successful theory is an empirical reality, the
trick is to find a model that fits this one. To do so, it is necessary to create a system in
which rebellion is possible, though not necessary likely, and spontaneous, in that the
transition to rebellion occurs with some speed, despite the lack of a single leader
dragging everyone into the streets. This paper presents just such a system.

Of course, this type of effort is not novel. There is a broad and insightful extant

literature in this field, almost as varied as the phenomenon itself. Some studies focus on
the impact of large-scale structural or institutional factors such as economic growth or
decline, inequality, regime structure and its relation to other states, or ethnic and cultural
rifts on the nature of political violence (Gurr 1970; Skocpol 1979; Muller 1985;
Huntington 1996; Kaufmann 1996; Fearon 2003). Others look to build from the ground
up, making assumptions on the motivations of citizen and state in order to construct a
model that is illustrative of the behavior in question (Granovetter 1978; Grossman 1991;
Kuran 1991; Lohmann 1994; Fearon 1998; Chwe 1999; Diermeier 2001; Epstein 2002).

In the following sections I propose a model that incorporates aspects of both. The

fundamental cost-benefit analysis that underlies Granovetter’s model—individuals have
various desires to rebel, but all fear the costs of doing so—is paired with a basic form of
learning—individuals update perceived costs over time—to form the behavioral
backbone of the model. The abstract level at which this is done allows me to combine
this productively with the generalized institutions of social networks, oppressive
governments, and the mass media in order to examine the effects of these on both the
likelihood and the form of rebellion.