System Narrative
How do rebel groups control territory and engage with the local economy during civil war? Charles Tilly’s seminal War and State Making as Organized Crime (1985) posits that the process of waging war and providing governance resembles that of a protection racket, in which aspiring governing groups will extort local populations in order to gain power, and civilians or businesses will pay in order to ensure their own protection. As civil war research increasingly probes the mechanisms that fuel local disputes and the origination of violence, we develop an agent-based simulation model to explore the economic relationship of rebel groups with local populations, using extortion racket interactions to explain the dynamics of rebel fighting, their impact on the economy, and the importance of their economic base of support. This analysis provides insights for understanding the causes and byproducts of rebel competition in present-day conflicts, such as the cases of South Sudan, Afghanistan, and Somalia.

Model Description
The model defines two object types: RebelGroup and Enterprise. A RebelGroup is a group that competes for power in a system of anarchy, in which there is effectively no government control. An Enterprise is a local civilian-level actor that conducts business in this environment, whose objective is to make a profit. In this system, a RebelGroup may choose to extort money from Enterprises in order to support its fighting efforts. It can extract payments from an Enterprise, which fears for its safety if it does not pay. This adds some amount of money to the RebelGroup’s resources, and they can return to extort the same Enterprise again. The RebelGroup can also choose to loot the Enterprise instead. This results in gaining all of the Enterprise wealth, but prompts the individual Enterprise to flee, or leave the model. This reduces the available pool of Enterprises available to the RebelGroup for extortion. Following these interactions the RebelGroup can choose to AllocateWealth, or pay its rebel fighters. Depending on the value of its available resources, it can add more rebels or expel some of those which it already has, changing its size. It can also choose to expand over new territory, or effectively increase its number of potential extorting Enterprises. As a response to these dynamics, an Enterprise can choose to Report expansion to another RebelGroup, which results in fighting between the two groups. This system shows how, faced with economic choices, RebelGroups and Enterprises make decisions in war that impact conflict and violence outcomes.

AncientS-ABM is an agent-based model for simulating and evaluating the potential social organization of an artificial past society, configured by available archaeological data. Unlike most existing agent-based models used in archaeology, our ABM framework includes completely autonomous, utility-based agents. It also incorporates different social organization paradigms, different decision-making processes, and also different cultivation technologies used in ancient societies. Equipped with such paradigms, the model allows us to explore the transition from a simple to a more complex society by focusing on the historical social dynamics; and to assess the influence of social organization on agents’ population growth, agent community numbers, sizes and distribution.

AncientS-ABM also blends ideas from evolutionary game theory with multi-agent systems’ self-organization. We model the evolution of social behaviours in a population of strategically interacting agents in repeated games where they exchange resources (utility) with others. The results of the games contribute to both the continuous re-organization of the social structure, and the progressive adoption of the most successful agent strategies. Agent population is not fixed, but fluctuates over time, while agents in stage games also receive non-static payoffs, in contrast to most games studied in the literature. To tackle this, we defined a novel formulation of the evolutionary dynamics via assessing agents’ rather than strategies’ fitness.

As a case study, we employ AncientS-ABM to evaluate the impact of the implemented social organization paradigms on an artificial Bronze Age “Minoan” society, located at different geographical parts of the island of Crete, Greece. Model parameter choices are based on archaeological evidence and studies, but are not biased towards any specific assumption. Results over a number of different simulation scenarios demonstrate better sustainability for settlements consisting of and adopting a socio-economic organization model based on self-organization, where a “heterarchical” social structure emerges. Results also demonstrate that successful agent societies adopt an evolutionary approach where cooperation is an emergent strategic behaviour. In simulation scenarios where the natural disaster module was enabled, we observe noticeable changes in the settlements’ distribution, relating to significantly higher migration rates immediately after the modeled Theran eruption. In addition, the initially cooperative behaviour is transformed to a non-cooperative one, thus providing support for archaeological theories suggesting that the volcanic eruption led to a clear breakdown of the Minoan socio-economic system.

…

This is a relatively simple foraging-radius model, as described first by Robert Kelly, that allows one to quantify the effect of increased logistical mobility (as represented by increased effective foraging radius, r_e) on the likelihood that 2 randomly placed central place foragers will encounter one another within 5000 time steps.

The Pampas Model is an Agent-Based Model intended to explore the dynamics of structural and land use changes in agricultural systems of the Argentine Pampas in response to climatic, technological economic, and political drivers.

This model visualizes gradient descent optimization - the fundamental algorithm used to train neural networks and other machine learning models. Agents represent different optimization algorithms searching for the minimum of a loss landscape (the “error surface” that ML models try to minimize during training).

The model demonstrates how different optimizer types (SGD, Momentum with different parameters) behave on various loss landscapes, from simple bowls to the notoriously difficult Rosenbrock “banana valley” function. This helps build intuition about why certain optimization algorithms work better than others for different problem geometries.

HOW IT WORKS

…

This model is an agent-based simulation designed to explore how climate-induced environmental degradation can contribute to the emergence of social violence in coastal communities that depend heavily on ecosystem services for their livelihoods. The model represents a coupled social–ecological system in which environmental shocks—such as sea level rise and marine ecosystem decline—affect local economic conditions, food security, and community stability.

Agents in the model represent individuals whose livelihoods depend on coastal ecosystems. Environmental degradation reduces ecosystem productivity and increases economic hardship, which can lead to the formation of grievances among agents. The model incorporates behavioral thresholds that determine how individuals respond to hardship and perceived injustice. Under certain conditions—particularly when institutional capacity and law enforcement effectiveness are limited—these grievances may escalate into violent behavior.

The simulation allows users to explore how different climate scenarios, levels of ecosystem degradation, livelihood dependence, and institutional responses influence the probability of social instability and violence. By modeling the interactions between environmental stress, socio-economic vulnerability, and governance capacity, the model provides a computational framework for examining potential pathways linking climate change and conflict in coastal social–ecological systems.

…

The “Descriptive Norm and Fraud Dynamics” model demonstrates how fraudulent behavior can either proliferate or be contained within non-hierarchical organizations, such as peer networks, through social influence taking the form of a descriptive norm. This model expands on the fraud triangle theory, which posits that an individual must concurrently possess a financial motive, perceive an opportunity, and hold a pro-fraud attitude to engage in fraudulent activities (red agent). In the absence of any of these elements, the individual will act honestly (green agent).

The model explores variations in a descriptive norm mechanism, ranging from local distorted knowledge to global perfect knowledge. In the case of local distorted knowledge, agents primarily rely on information from their first-degree colleagues. This knowledge is often distorted because agents are slow to update their empirical expectations, which are only partially revised after one-to-one interactions. On the other end of the spectrum, local perfect knowledge is achieved by incorporating a secondary source of information into the agents’ decision-making process. Here, accurate information provided by an observer is used to update empirical expectations.

The model shows that the same variation of the descriptive norm mechanism could lead to varying aggregate fraud levels across different fraud categories. Two empirically measured norm sensitivity distributions associated with different fraud categories can be selected into the model to see the different aggregate outcomes.

This version 2.1.0 of the uFunk model is about setting a business strategy (the S in the name) for an organization. A team of managers (or executives) meet and discuss various options on the strategy for the firm. There are three aspects that they have to agree on to set the strategic positioning of the organization.
The discussion is on market, stakeholders, and resources. The team (it could be a business strategy task force) considers various aspects of these three elements. The resources they use to develop the discussion can come from a traditional approach to strategy or from non-traditional means (e.g., so-called serious play, creativity and imagination techniques).
The S-uFunk 2.1.0 Model wants to understand to which extent cognitive means triggered by traditional and non-traditional resources affect the making of the strategy process.

This is an extension of the original RAGE model (Dressler et al. 2018), where we add learning capabilities to agents, specifically learning-by-doing and social learning (two processes central to adaptive (co-)management).

The extension module is applied to smallholder farmers’ decision-making - here, a pasture (patch) is the private property of the household (agent) placed on it and there is no movement of the households. Households observe the state of the pasture and their neighrbours to make decisions on how many livestock to place on their pasture every year. Three new behavioural types are created (which cannot be combined with the original ones): E-RO (baseline behaviour), E-LBD (learning-by-doing) and E-RO-SL1 (social learning). Similarly to the original model, these three types can be compared regarding long-term social-ecological performance. In addition, a global strategy switching option (corresponding to double-loop learning) allows users to study how behavioural strategies diffuse in a heterogeneous population of learning and non-learning agents.

An important modification of the original model is that extension agents are heterogeneous in how they deal with uncertainty. This is represented by an agent property, called the r-parameter (household-risk-att in the code). The r-parameter is catch-all for various factors that form an agent’s disposition to act in a certain way, such as: uncertainty in the sensing (partial observability of the resource system), noise in the information received, or an inherent characteristic of the agent, for instance, their risk attitude.

We construct a new type of agent-based model (ABM) that can simultaneously simulate land-use changes at multiple distant places (namely TeleABM, telecoupled agent-based model). We use soybean trade between Brazil and China as an example, where Brazil is the sending system and China is the receiving system because they are the world’s largest soybean exporter and importer respectively. We select one representative county in each country to calibrate and validate the model with spatio-temporal analysis of historical land-use changes and the empirical analysis of household survey data. The whole model is programmed on RePast Simphony. The most unique features of TeleABM are that it can simulate a telecoupled system and the flows between sending and receiving systems in this telecoupled system.