The approaches focused on the polity dynamics before and during the outbreak of civil conflict or civil war usually model the spatial and temporary distribution of events and people as spatial configurations of conflict‐prone regions or groups (Weidmann and Ward, 2010). The post‐conflict modelling studies often emphasize phenomena like the emergence of new state entities. A good example is Kosovo, which emerged as a new state entity after the military confrontations in the Kosovo war during 1998–1999.1 There are also cases of reintegration or reconstruction of the old state entity/entities into new one(s), either the same or different state entities. African polity cases are relevant examples, usually resulting from the violent civil conflicts and civil wars during the past half century.
The study of polity dynamics in civil and military conflict scenarios could reveal important details about polity’s behaviour during the conflict phenomena and after their completion. Moreover, prolonged conflicts have often resulted in repeated state‐failure scenarios, especially in northern Africa, where the cases of Somalia, Ethiopia, Sudan, Eritrea and Kenya illustrate repeated processes of state disintegration and reintegration after violent civil and military conflicts.
The model elaborated by Takuto Sakamoto and Mitsugi Endo (2015) is concerned with polity reconstruction after violent confrontations in armed conflict (from now on called ‘Polity Reconstruction Model’). The model is based on the conflict analysis in virtual states (CNVS) model (Sakamoto, 2013b) and on a GIS data model which enhanced the representation of conflict evolutions (Sakamoto, 2013a).
The virtual polity is an agent‐based model (Sakamoto, 2013a) of a state entity with territory, society, government and resources. The model replicates the political and military conflicts between the government and the insurgent groups for territory hegemony. As the conflicts have ethnic and/or religious roots, both the government and the insurgent groups are characterized by attributes which exclusively describe these characteristics of the individual inhabitants.
The model polity resembles the states in northern Africa, among which Somalia has been approached as a case study, in which the population is multi‐ethnic and civil conflicts emerge as the struggle for power and resources between the government and conflict‐prone groups.
The definition of the model concerns the structure of the polity, the types of agents and their goals and roles. The reconstruction of the polity in the same state entity or in several such entities is modelled as a macro‐emergent phenomenon based on the interactions (conflicts) unfolding at the micro level of the artificial polity.
The territory is modelled as a two‐dimensional grid of cells whose boundaries’ geometry closely resembles those of the African states studied. Each cell represents the spatial unit of territory representation.
There are two types of agents defined in the model: the population type (Population) and the ruler type (Ruler). Each type of agent is characterized by a set of socio‐cultural attributes (invariant) called traits, and a set of goals and roles is defined as operational procedures for each type of agent.
The Population type of agent is modelled as an individual agent characterized by three classes of socio‐cultural traits: ethnicity, religion and region.2 The ‘ethnicity’ class includes traits like ‘ethnicity’, ‘nation’, ‘nationality’ and ‘tribe’. The ‘religion’ class includes ‘traditional’ (i.e. systems of faith which are specific to various ethnic groups) and ‘universal’ (i.e. Christianity, Islam) religions. The ‘region’ class regards the regional administrative divides. These traits are considered invariant in the model. The individual agents inhabit a territory or a region. Each cell in the territory description is characterized by a number of individual agents and a certain amount of resources.
The Ruler type is a class representation of the groups which dispute the supremacy over a region/territory. The goals of a Ruler agent are to gain, extend and maintain control over the state by excluding all other competitor Ruler agents. The Ruler type of agent has two characteristics: traits and mobilization factors. The Ruler’s traits are considered invariant and indicate whether there is some political alignment between the Ruler and the agents or whether the Ruler shows indifference with regard to the Population agents’ traits (i.e. this could be the case when the Ruler gains control over a new region). The mobilization factors are considered variable and concern the human and material resources from both inside and outside the state which are acquired by the Ruler: the more resources it mobilizes, the more powerful it is.
The polity dynamics are traced at the level of the interactions of Population and Ruler agents. The simulations use the georeferenced GIS data acquired from the real state history (e.g. Somalia). The simulations generate at each run the population which inhabits each cell of the territory. Traits and resources are initialized with distributions of values which are afterwards updated by the interactions (conflicts) between insurgent groups. The initial Ruler faces challenges from other rulers/groups. As the conflict emerges, resource mobilization is performed by each competitor until one ruler gains (temporary) control over the territory. Population‐type agents prove loyalty or betray their rulers, thus modifying the dynamics of the conflicts and conditioning their outcomes. The conflicts usually result in territorial division until the state disintegrates and the initial government disappears. Each Ruler sooner or later faces the same fate. The analysis of the simulation results is based on a measure of state disintegration (disintegration index, DI) such that the system could keep track of the degree of territorial division as the effect of conflicts.
The model suggests a dimension of modelling which has not been pursued in the literature concerning computational and simulation modelling and artificial polity. The reconstruction dimension suggests that the development of speculative scenarios would allow the evaluation of potential outcomes which cannot be tested in reality and which, moreover, cannot be validated unless tested in the long run. Reconstruction provides a viable alternative to test possible situations and to evaluate the source of polity instability.
The model constructs various alternative scenarios to the original polity instability by systematic variation of several classes of parameters which were initially kept invariant: external resources, traits and even the territory. This allows the study of the long‐term territorial condition in two kinds of scenarios: the analysis of consequences by the replication of real phenomena and the prediction of the conditions which would favour or facilitate polity territorial stability in the long run. Each study concerns the results of 20 runs with appropriate variation in the parameters’ values. The simulated experiments revealed several kinds of polity stability and instability phenomena, where both ‘stability’ and ‘instability’ are defined in terms of territorial fragmentation.
The replication‐oriented simulated experiments developed two case studies.
In one case, the degree of territory disintegration is associated with the changing amount of external resources presumably received by the Ruler (i.e. government). The simulations revealed a persistent instability in terms of territorial fragmentation regardless of sustained ‘infusions’ of external resources. As the author notes, the model explains a prevailing tendency toward territorial fragmentation in the conditions of genuine clan‐based polity organization, a hypothesis supported by other studies as well.
In the other case, the degree of territorial disintegration is associated with values of gross national income (GNI) per capita. In these studies, several major phases could be conceived in the process of long‐term territorial fragmentation: at moderate levels of local resources, there is a tendency in the long run toward stability in the territorial structure, without excluding territorial fragmentation but making it develop at a slow pace. At increased levels of GNI per capita, there emerges a stable polity with no more territorial fragmentation even in the absence of external resources. The model thus identifies as sources of polity instability the level of resources (i.e. low in real polity) and the groups’ distribution over the territory (i.e. sparse in real polity).
The prediction‐oriented simulations introduced two main differences from the original model settings: the traits as well as the territory itself were systematically varied.
The alternative based on variation of traits results in the emergence of a significantly increased unity achieved by government re‐orientation in ethnic and religious matters. However, reconstruction based on increased unity achieved in the socio‐cultural traits proved that this alternative cannot compensate the low resources problem and in the long run territorial fragmentation cannot be avoided.
The reconstruction modelling thus proves useful in generating and analysing various scenarios which otherwise would be hard to investigate.
The Conflict Analysis in Virtual States Model, CNVS (Sakamoto, 2013a), represents a geographically situated polity with territory, population and resources. It combines an artificial society model with an agent‐based model of individual interactions at the micro level of the artificial society and also with a GIS for simulating the emergence of civil conflicts and their outcomes. Case studies address state failure in African countries like Sudan (Sakamoto, 2013a).
The state reconstruction model is based on an African case study, Somalia (Sakamoto, 2013b), where the state is viewed as a self‐organizing entity. Sakamoto and Endo’s polity model has three dimensions: territory, resources and traits.
Computational and simulation modelling of polity is concerned with territorial integration/disintegration caused by conflict and the use of traits and symbols in the polity modelling.
Yamakage Virtual Lab – Exploring Artificial Societies Professor Susumu Yamakage (University of Tokyo, Japan). The research platform called artisoc (from ‘artificial society’) is a general‐purpose simulator developed by Kozo Keikaku Engineering Inc. The copyright of artisoc belongs to both Professor Susumu Yamakage (University of Tokyo, Japan) and Kozo Keikaku Engineering Inc., Japan. The artisoc official website, called MAS Community, is at http://mas.kke.co.jp/. artisoc is not open‐source software. The software is sold by Kozo Keikaku Engineering Inc. Free rental is available for research use in a college/university. Inquiries should be addressed to the Innovative Information Technology Department of KKE, contact address at mas‐support@kke.co.jp.