Поиск паттернов в динамике протестных кампаний: вычислительное моделирование и эмпирический анализ
Much of current research misses the dynamic nature of protest campaigns. Meanwhile, the analysis of the dynamics allows us to identify the role of various factors influencing their course. First of all, this applies to repressions: how does their use affect the development and, ultimately, the outcome of the protest campaign? In itself, the question of under what circumstances repression weakens and under what circumstances strengthens protest is the subject of a large number of empirical and theoretical studies. However, in most studies, the design of the study misses the dynamic aspect, which seems to be methodologically incorrect. After all, repression can be used in various strategies: either already at the first protest action, or when a certain number of protesters is reached, etc. They can vary over time: increase or decrease over the course of the campaign, etc.
The variety of different dynamics of the number of protesters and the dynamics of the use of repression gives rise to a variety of scenarios for the development of a protest campaign. In this regard, this paper raises the question of identifying dynamic patterns. At the same time, we consider both empirical scenarios that have already taken place in real protests, as well as “ideal”, i.e. arising in theory and capable of serving as guidelines in the analysis of real ones.
To obtain "ideal" scenarios, a theoretical and mathematical model was developed with various strategies for the reactions of the authorities to the protesters, which we implemented into the existing computational model of protest mobilization. Based on the data obtained in the course of computer simulations, firstly, by linear and logistic regressions, the effects of various decision-making mechanisms on the survival of protests were evaluated, and, secondly, using various methods of time series cluster analysis, we discovered a number of patterns. For verification, the same methods of cluster analysis were applied repeatedly on empirical data.