In this monograph we bring together a conceptual treatment of evolutionary dynamics and a path-ensemble approach to non-equilibrium stochastic processes. Our framework is evolutionary game theory, in which the map from individual types and their interactions to the fitness that determines their evolutionary success is modeled as a game played among agents in the population. Our approach, however, is not anchored either in analogy to play or in motivations to interpret particular interactions as games. Rather, we argue that games are a flexible and reasonably generic framework to capture, classify and analyze the processes in development and some forms of inter-agent interaction that lie behind arbitrary frequency-dependent fitness models.