In equilibrium statistical mechanics, the macroscopic states of a system at thermal equilibrium are described by probability measures on the space of microscopic states that maximize pressure. For systems whose microscopic states are symbolic configurations from a subshift, Dobrushin, Lanford and Ruelle showed that under broad conditions, global and local equilibrium conditions are equivalent, that is, "equilibrium measures" are the same as (shift-invariant) "Gibbs measures". I will discuss some variants and generalizations of this theorem, in particular, a broad generalization to systems in contact with a random environment. Some nice symbolic dynamics issues arise. The underlying lattice can be any countable amenable group.