Turbo Simmel is a symbolic AI system that combines signed graph analysis, structural balance, sociological theory and a small expert knowledge base to analyse social conflicts, alliances and coalitions.
Actors are nodes in a signed graph. Positive edges represent friendship, alliance or cooperation; negative edges represent conflict, hostility or competition; neutral edges represent an undefined relationship. The graph is the internal model on which the system reasons.
The theoretical model combines Georg Simmel's analysis of the triad (Simmel, 1950 [1908]), Theodore Caplow's theory of coalition formation (Caplow, 1956), and structural balance theory (Cartwright & Harary, 1956).
The system searches the network for elementary cycles, examines the pattern of positive and negative edges, and calculates structural balance. Positive edges have the value +1 and negative edges −1. If the product of the edge values in a cycle is positive, the cycle is balanced; if it is negative, it is unbalanced.
The same graph structures are also classified as sociological situations. In this way the program turns a network pattern into a social interpretation.
The graph-analysis engine is connected to a small expert knowledge base containing social configurations, coalition patterns, actor positions and strategy rules. The examples stored in the knowledge base are drawn from Georg Simmel's analysis of the triad.
The surviving program contains more than twenty named strategies, including Friend Mediates, Tertius Gaudens, Builds a Coalition, Divide, Joins Against Common Enemy, and Enemy's Enemy = Friend.
The system compares the current graph with patterns in the knowledge base and determines which strategies are applicable from the position of a selected actor. This is rule-based inference over a graph representation.
The reasoning chain is:
graph → elementary cycles → structural balance → social pattern → actor position → expert rule → strategy
A strategy represents a possible change in one or more relationships. The program can display the network that would result if the strategy succeeds, and the new graph can then be analysed again. The History function stores successive states, allowing alternative social dynamics to be explored.
The original source code has been lost, but the compiled DOS program, its help system and its example library survived. In 2026, I reconstructed the system with the help of ChatGPT: the surviving files were analysed, the help and examples were recovered, the graph-analysis and expert-system logic were documented, and this browser-based version was created.
The web version is a functional reconstruction, not a line-by-line recovery of the lost source code.
Caplow, T. (1956). “A Theory of Coalitions in the Triad.” American Sociological Review, 21(4), 489–493.
Cartwright, D., & Harary, F. (1956). “Structural Balance: A Generalization of Heider’s Theory.” Psychological Review, 63(5), 277–293.
Simmel, G. (1950 [1908]). “The Triad.” In K. H. Wolff (Ed. & Trans.), The Sociology of Georg Simmel, 145–169.
