Social flow, viscosity, and turbulence increasingly help explain observations of collective social systems in which self-organization is driven by norms and beliefs. We propose a simple agent-based model of self-organization of human agents, adapted from seminal color sorting experiments of individual perceptions of color proximity, as representative of a fundamental class of social phenomena involving convergence toward a stable collective social structure. We define inverse social viscosity as the measure of the difference between agent situated beliefs and perceptions as the driver of collective action flow. We study convergence and reversions using a particular form of the equation describing nucleation processes in phase transition theory. Our analysis suggests that tolerance to imperfect compliance with norms and a degree of tolerance with own beliefs decrease coordination efforts. In addition, our research suggests that social viscosity is a proxy measure for the cost of social organization, which can in turn be used to inform the design of socio-technical systems. Breaking social isolation is a successful strategy to foster self-organization.