I build GPU based particle simulations that model social contagion in large populations. Each dot begins in a random color state and at a random position. Once the simulation begins, each particle simply copies the color that is most common in its local neighborhood while it moves through the simulation environment.
While eventually the whole population always converges to a single color (opinion, language, behavioral strategy) the time it takes to reach convergence and the predictability of the winning color depends heavily on the way in which the particles move. I investigate three movement types: every individual moves in a constant (random) straight line, every individual moves in a random walk, and individuals have social schooling movement behavior (see Couzin '05 for schooling rules).