Abstract: TBA

Abstract: TBA

Abstract.XXX

Abstract:TBA

Abstract: This presentation introduces GDAM — Generative Dance with Agent-Based Models — as a complexity-informed artistic research methodology grounded in the logic of agent-based modeling, with a particular focus on cellular automata. GDAM investigates how emergent choreographic organization, togetherness, and collective self-organization can arise from simple relational rules enacted by moving bodies in shared space. Cellular automata provide one of the conceptual and operational bases of GDAM. In computational terms, they are systems composed of discrete units whose states evolve over time according to simple local rules and neighborhood relations. Although each unit follows a limited set of instructions, the collective evolution of the system can generate complex, unpredictable, and emergent patterns. GDAM translates this logic into collective dance improvisation, where each dancer functions as an embodied agent whose movement decisions are affected by local perception, relational proximity, spatial orientation, rhythm, attention, and the actions of neighboring dancers. Rather than using cellular automata only as computational models, GDAM explores them as choreographic and methodological devices, named MCOs — Mathematical Choreographic Objects. In this translation, local rules become movement scores; neighborhoods become perceptual and spatial relations between bodies; states become embodied qualities, actions, orientations, or attentional modes; and iteration becomes the unfolding of choreographic time. Through MCOs, GDAM creates conditions in which complex collective behaviors can emerge without centralized control, including synchronization, divergence, distributed leadership, collective spatial patterning, and shared affective atmospheres. The methodology combines artistic experimentation, embodied practice, qualitative observation, participatory reflection, and computational imagination. Its aim is not to reduce dance to computation, but to use the generative principles of cellular automata and agent-based modeling to design research situations where human social complexity can be experienced, observed, and analyzed from within. In this sense, GDAM proposes a transdisciplinary bridge between complexity theory, choreographic composition, qualitative inquiry, and social systems research. A central methodological contribution of GDAM lies in its capacity to make visible processes that often escape conventional research methods: tacit coordination, embodied decision-making, mutual adaptation, relational sensitivity, hesitation, resonance, and the emergence of collective agency. By placing dancers inside rule-based yet open-ended systems, GDAM invites the study of how social organization is generated through situated interactions rather than imposed from above. This presentation argues that GDAM does not simply combine dance and computational modeling; rather, it choreographs an interaction between them. Cellular automata provide a formal logic for modeling emergence in dance, while dance provides an embodied field where this logic becomes affective, relational, and socially meaningful. Through this interaction, GDAM supports abductive thinking, allowing unexpected insights to emerge through practice, perception, and reflection.

Abstract: Max Weber launched the ideal type method, emphasizing the difference of social sciences from natural sciences. In social sciences, the natural science positivist methodology of verifying empirical phenomena against some universal law is inapplicable. The social researcher has, first, to deconstruct the phenomenon under analysis, laying out its principal components. Second, the researcher has to establish causation between the components, constructing the phenomenon’s functional representation. The established functional structure and causation is then projected on the real phenomenon, explaining it by measuring its distance from the ideal type. The ideal type is a subjective, researcher-bound, value-biased instrument that, nevertheless, explains the social phenomena under analysis better than positivist methods concealed behind the “objective truth” façade. For Weber, the ideal type is rooted in the very nature of human thinking that produces categories for understanding the world. Both the social actor and the analyst studying that actor think in ideal-type categories, thus bridging the gap between the social phenomenon itself and its analysis. Employment of Large Language Models (LLMs) – the currently dominant form of Artificial Intelligence (AI) – for social studies challenges this symbiosis. LLMs reason by predicting the next token – a lower-level data unit – from the statistical properties of the preceding sequence of tokens. However, the LLM token is not a category of the entity as a whole but a fragment of it. Humans grasp a real phenomenon by its ideal type and manage reality accordingly. LLMs decompose real phenomena into fragmented data, process them piecemeal, and then reproduce a construct of reality by assembling tokens according to next-token prediction. A mental gap between human social action and its analysis by LLMs thus exists and needs to be managed; otherwise, it becomes a principal source of the LLM’s thinking stupor and hallucinations in running social studies. The ideal type methodology provides the solution for the effective employment of LLMs in the analysis of complex social phenomena. The strategic organisations – large entities characterised by self-sufficient, purposeful behaviour such as sovereign states, cohesive social groups, political parties, and business corporations – are among them. I have developed “Strategy by AI”, the methodology that implements the Weberian Ideal Type method for the analysis, governance, and prediction of complex strategic organisations. The methodology operates through six modules producing a portfolio of documents composing the strategic world model – the ideal type of the subject – that functions as an axiomatic yardstick: a fixed coordinate system against which the subject’s observable conduct is measured and explained by the distance between the two. The methodology imposes the reasoning architecture that organises the LLM’s encyclopaedic knowledge into strategic intelligence; neither the methodology alone nor the LLM alone could generate such output. The methodology’s description and examples of its employment – strategic cases – are published on the website www.strategybyai.org. I will present the achievements and limitations of the methodology and compare it with the ideal type method as I employed it by traditional means in my pre-AI books and essays.

Abstract: TBA

Abstract: XXX

Abstract: This work presents a scientometric analysis of academic production in political science focused on Latin America, combining network analysis techniques and text-mining methods. A corpus of 5,409 articles published between 1980 and 2025 was constructed using data collected from the Scopus and OpenAlex databases. Based on this dataset, a citation network with more than five thousand vertices was developed, enabling the investigation of structural patterns within the scientific community dedicated to the study of Latin American politics. The descriptive analysis of the corpus reveals a strong growth in scientific production over recent decades, as well as an asymmetric geographical distribution of publications, with a predominance of journals based in the United States and the United Kingdom. The citation network structure exhibits typical properties of complex systems, including an approximately power-law degree distribution and the presence of a giant connected component concentrating most of the publications. Statistical tests on citation patterns were also conducted, indicating that articles tend to cite works published in the same language more frequently. Furthermore, a strong asymmetry in scientific dialogue between countries was observed, particularly highlighting the centrality of connections between Brazil and the United States. To investigate the thematic structure of the literature, topic modeling was performed using the BERTopic algorithm, which identified 18 central themes in the research agenda on Latin American politics. Additionally, Textual Forma Mentis Networks were employed to examine the conceptual organization of the texts, revealing structural differences between academic productions authored by researchers from the region and those produced outside it. The results contribute to a better understanding of the global dynamics of political science research on Latin America, highlighting structural inequalities in the circulation of knowledge and suggesting new directions for future computational analyses of scientific literature.

Abstract: TBA