Working Paper Agent-Based Modeling of the Early Minoan Social Organization Structure Angelos Chliaoutakis 1 and Georgios Chalkiadakis 2 Abstract. In this work, we develop a generic agent-based model (ABM) for simulating ancient societies. Agents in our model are completely autonomous, in contrast to most existing ABMs used in archaeology. We employ this model to evaluate the impact of differ- ent social organization paradigms and agricultural strategies on pop- ulation viability and evolving spatial distribution of settlement loca- tions, at a particular region of the island of Crete during the Early Bronze Age (early Minoan civilization). Model parameter choices are based on archaeological studies, but are not biased towards any specific assumption. Interestingly, one of the social models exam- ined is inspired by a recent framework for self-organizing agent or- ganizations. Results over a number of different simulation scenar- ios demonstrate an impressive sustainability for settlements adopt- ing a socio-economic organization model based on self-organization; while the emerging “stratified” populations are larger than their egal- itarian counterparts. This provides support for archaeological theo- ries proposing the existence of different social strata in Early Bronze Age Crete, considering them a pre-requisite for the emergence of the complex social structure evident in later periods. Moreover, observed population dispersion agrees with existing archaeological evidence. 1 Introduction Agent-based modeling (ABM) 3 is increasingly used in Archaeology in recent years, as a tool for assessing the plausibility of alterna- tive hypotheses regarding ancient civilizations, their organization, and social and environmental processes at work in past ages [4, 14]. Its emerging popularity is due to its ability to represent individuals and societies, and to encompass uncertainty inherent in archaeolog- ical theories or findings. Indeed, the unpredictability of interaction patterns within a simulated agent society, along with the possibility of emergent behaviour, can help researchers gain new insights into existing archaeological theories; or even come up with completely novel paradigms regarding the ancient societies being studied. Now, ABMs should ideally be providing a higher level of ab- straction than the one offered by object-oriented systems [12]. Mod- eled agents should be capable of autonomous action, and of main- taining high-level interactions and organizational relationships with other agents, while being potentially “selfish” [17]. However, most mutiagent-based simulation models used in archaeology, simply do not define agents in the way these are defined in AI or MAS research. Unfortunately, “agents nowdays constitute a convenient model for 1 Technical University of Crete, Greece, email: angelos@intelligence.tuc.gr, secondary affiliation: GeoSat ReSeArch Laboratory - IMS/FORTH, Greece 2 Technical University of Crete, Greece, email: gehalk@intelligence.tuc.gr 3 ABM acronym denotes “agent-based modeling” and “agent-based model”. representing autonomous entities, but they are not themselves au- tonomous in the resulting implementation of these models” [7]. In contrast to most existing ABM approaches in archaeology, agents in our model are completely autonomous, and can build and maintain complex social structures. Furthermore, though our work here is inspired by existing case studies, our model itself is quite generic, and does not aim to prove or disprove a specific theory. In- deed, using agent-based models built on knowledge derived from ar- chaeological records and evidence, but not trying to fit their results to a specific material culture, allows for the emergence of dynam- ics for different types of societies in different types of landscapes, and can help derive knowledge of socio-ecological systems that are applicable beyond a specific case study. In more detail, in this work we have developed a functional ABM system prototype for simulating an artificial ancient society of agents residing at the Malia area of the island of Crete during the Early Bronze Age. The agents correspond to households, which are con- sidered to be the main social unit of production for the period [16]. The ABM allows us to explore the sustainability of agricultural tech- nologies in use at the time, and examine their impact on popula- tion dispersion. In addition, the ABM attempts to assess the influ- ence of different social organization paradigms on land use pat- terns and population growth. Importantly, the model evaluates the social paradigm of agents self-organizing into a hierarchical social structure, and continuously re-adapting the emergent structure, if re- quired. To this purpose, we developed and tested a self-organization algorithm that builds on the work of [15] on self-organizing agent organizations used for problem-solving and task execution. We note that this is the first time a self-organization approach is incorporated in an ABM system used in archaeology. Our simulation results demonstrate that self-organizing agent pop- ulations are by far the most successful, growing much larger than populations employing different social paradigms. The success of this (dynamic) social paradigm that gives rise to “stratified”, that is, non-egalitarian societies, provides support for so-called “manage- rial” archaeological theories which assume the existence of different social strata in Neolithic / Early Bronze Age Crete; and consider this early stratification a pre-requisite for the emergence of the Minoan Palaces, and the hierarchical social structure evident in later peri- ods [2, 9]. In addition, the observed dispersion of settlements is in line with existing archaeological data. 2 Related work In recent decades, archaeologists have used computer models to test possible explanations for the rise and fall of complex ancient soci- eties. They now begin to use multiagent simulations as a means for