Emergent preeminence of selfishness: an anomalous Parrondo perspective

Abstract

A minimalistic multi-agent Parrondo’s game structure with branching dependent on local capital spread was previously introduced, indicating that stochastically mixing two losing games can produce winning outcomes with bounded capital variance among players. Using a similar game structure, we unveil further intriguing behavior that a bias toward selfish exploitative behavior, involving redistribution of capital from the poor to the rich, leads to counterintuitive superior capital gains than cooperative behaviors. Inter-agent interactions of exploitative nature not only maximizes capital growth in winning scenarios, but also expands the parameter space over which the Parrondo effect may manifest. These novel findings suggest a link between growth maximization and inequality that could be relevant to socioeconomic, ecological, and population dynamics modeling. We also present a theoretical framework for enhanced accuracy in the prediction of ensemble capital statistics.