The macroscopic mechanical response of polymers can be traced down to the physics of the network by using a statistical approach for the description of the configuration state of the polymer chains. In this paper we present a micromechanical model to capture the macroscopic behavior of polymers by tracking the evolution of a distribution function describing chain configurations, more specifically the statistics of the end-to-end distance on the network chains. Damage, manifested in the softening and hysteresis under cyclic loading, is accounted for through the scission of chains, whose occurrence is evaluated on the basis of the probability of failure, also settled in the configuration space. The proposed micromechanical model can easily accommodate also the mechanics of dynamic network with reversible cross-links, thereby providing a general and physics-based approach to the study of polymers and polymer-like materials.