In real-world systems, phase transitions often materialize abruptly, making it difficult to design appropriate controls that help uncover underlying processes. Some agent-based computational models display transformations similar to phase transitions. For such cases, it is possible to elicit detailed underlying processes that can be subsequently tested for applicability in real-world systems. In a genetic algorithm, we investigate how a modest difference in the concentration of correct and incorrect knowledge leads to radically different outcomes obtained through learning efforts by a group of agents. We show that a difference in concentration of correct and incorrect knowledge triggers virtuous and vicious cycles that impact the emergent outcome. When virtuous cycles are in operation, delaying the onset of equilibrium attains superior outcomes. For the vicious cycles, reaching equilibrium quickly attains superior outcomes. Our approach helps uncover simple mechanisms by which Nature works, jettisoning the yoke of unrealistic assumptions endemic in mathematics-based approaches. Our explanatory model helps direct research to investigate concentrations of inputs that obtains outcomes on the favourable side of phase transitions. For example, by tracking change in concentration of relevant parameters, scientists may look for reasons why cells cease to reproduce fit cells in organs. This can help design rejuvenation of organs. Further, in the world of physics, our model may inform in situations where the dominant Ising model falls short.
Download