In a previous post, we described how going beyond probabilistic approaches enabled the mastery of propensities to emerge as the central concept of the Cindynics. On the other hand, even if it does not really describe the unreliability of probabilistic approaches, an article by Clayton Besaw and Jonathan Powell analysing the case of the failed coup in Libreville in 2019 advocates the need to combine qualitative and quantitative approaches: in doing so, they corroborate the contribution of the cindynic approach and the interest of the power analysis that can be achieved thanks to the most advanced cindynical models.
Besaw and Powell note that their forecasting models of coups, based on a dataset listing all coups for 50 years, failed to predict the 2019 Libreville putsch. According to this models, only two countries in sub-Saharan Africa were less likely than Gabon to suffer a putsch in 2019: Seychelles and Tanzania.
The OEF Research 2019 Annual Risk Of Coup Report1 points out that Jonathan Powell had noticed ex ante that modeling had arguably underestimated the risk of coup in Gabon. The report specifies a weakness of the modeling: "Our forecasts can approximate the conditions in which coup plotters may risk their lives to use military means to force leadership change, but it cannot ascertain the exact individual psychological dispositions and strategic thinking that drives them to act".
Joshua Lambert2 describes different phases of political forecasting research: after the failure to predict the collpase of the Soviet Union, political forecasting faced suspicion, and the experts' opinions were even compared to those of "dart throwing chimps". Then the Political Instability Task Force project, funded by the CIA, rekindled interest in the field, and paved the way for a third generation of forecasting using new methods, such as machine learning, which improve forecasting capabilities and thus increase confidence in forecasts.
For their part, Cindynics went through different phases of evolution: initially, going beyond probabilistic approaches made it possible to thoroughly consider actors of danger situations, to define the vulnerability and resilience of consensual situations, and to develop a prevention method based on vulnerability reduction and mastery of propensities. Then IFREI specifically studied non-consensual multipolar situations and developed second-order Cindynics, on the basis that observed actors in any situation are also observers. This makes it possible to study actors' prospectives, i.e. the situations that each of them would ideally wish, before any actual decision to take action. This approach enables early forecasting (and prevention) where other approaches have failed to produce correct forecasts, for example in the case of the Arab Spring or the collapse of the Soviet Union3 . Second-order Cindynics are therefore a response to the weaknesses pointed out in the OEF Research report since they help to precisely analyse the actors' prospectives, their dispositions, and the strategies that will lead to their decisions and behaviors.
The second problem described by Besaw and Powell is the failure of the Libreville coup: they quote the analysis of Naunihal Singh, for whom the success of a coup notably relies on the credibility of the message sent by the putschists. Naunihal Singh particularly analysed the coups which took place in Ghana between 1966 and 1983. He believes that during a putsch, the most important in an actor decision is to support the side that all the others will support4 . Hence the importance of the credibility of the putschists' communication: "For a successful coup, the content of the broadcast should convey a credible but exaggerated sense of the strength of the challengers relative to the loyalists"5 .
This analysis is in line with the latest developments in cindynical analysis: second-order Cindynics make it possible to describe the power of an actor as his ability to impose his prospective on other actors. The question of the perception of power is at the heart of the dynamics of collective action construction in general, which applies in particular to the dynamics of revolutions and coups. If you have ever found yourself in a situation of having to catalyse a collective action, you have noticed that most of the actors will only follow you if you have already succeeded in mobilising enough actors, and that you thus display a credible initial power. In practice, to take up a Gramscian reflection, your problem is therefore to first find actors "optimistic by willpower" who consider above all the content of the project and who will adhere to it, in order to be able then to be joined by the "pessimists by reason" who above all attach importance to not taking the risk of joining a project that they do not feel is sufficiently supported.
In other words, actors evaluate your power before joining your project, or not : this means that actors, like any cindynician, observe the situations observed by other actors. This leads to third-order cindynical models, where analysts observe a matrix of sets of relative situations observed by different actors. This transition to third-order makes it possible to analyse what happens in practice by considering each actor's ability to estimate actors' powers. Thus, third-order cindynical models open the door for an in-depth analysis of the dynamics of coups evoked by Besaw and Powell, and studied by Naunihal Singh. Limiting oneself to second-order models would certainly be simpler, but when it comes to coups, refusing to consider the strategic thinking capacities of actors would be somewhat unreasonable.
The failure of Lieutenant Kelly Ondo Obiang's coup is the result of the message he broadcast: by failing to display any credible existing support, and by calling on civil society to come out and demonstrate in support of the coup, he demonstrated his lack of power. Actors perceived this power insufficiency and therefore most of them did not join the desired mobilisation.