Theia Vogel’s "Gas Town" simulation explores the intersection of social theory and agentic behavior. By prompting ChatGPT to summarize complex texts on state formation—specifically circumscription theory, which explains how environmental boundaries force social complexity—Vogel constructed a digital environment where AI agents act as town mayors. These leaders navigate a world hemmed in by computer constraints, mimicking the natural boundaries that historically drove human societies toward centralized governance and conflict. The simulation’s economy relies on an exploitation of resource scarcity. Instead of a demand-based system, Vogel allocated a fixed number of tokens—the basic units of data used by an LLM to process text—to specific towns. This created a zero-sum game where growth requires conquest. By introducing a "soldier" role, the simulation allows towns to defend their resources or launch raids to steal tokens from neighbors. This transformation of API limits into a strategic game mechanic highlights how constraints can foster emergent properties in multi-agent systems. Beyond entertainment, this project serves as a microcosm for studying how models interact in competitive, resource-constrained environments. It demonstrates the potential for using AI not just as a chatbot, but as an AI Agent capable of simulating sociological theories. As these agents navigate conflict and cooperation to secure their survival, they provide a unique lens into the dynamics of resource management and strategic planning within digital ecosystems.

Theia Vogel’s "Gas Town" simulation explores the intersection of social theory and agentic behavior. By prompting ChatGPT to summarize complex texts on state formation—specifically circumscription theory, which explains how environmental boundaries force social complexity—Vogel constructed a digital environment where AI agents act as town mayors. These leaders navigate a world hemmed in by computer constraints, mimicking the natural boundaries that historically drove human societies toward centralized governance and conflict. The simulation’s economy relies on an exploitation of resource scarcity. Instead of a demand-based system, Vogel allocated a fixed number of tokens—the basic units of data used by an LLM to process text—to specific towns. This created a zero-sum game where growth requires conquest. By introducing a "soldier" role, the simulation allows towns to defend their resources or launch raids to steal tokens from neighbors. This transformation of API limits into a strategic game mechanic highlights how constraints can foster emergent properties in multi-agent systems. Beyond entertainment, this project serves as a microcosm for studying how models interact in competitive, resource-constrained environments. It demonstrates the potential for using AI not just as a chatbot, but as an AI Agent capable of simulating sociological theories. As these agents navigate conflict and cooperation to secure their survival, they provide a unique lens into the dynamics of resource management and strategic planning within digital ecosystems.