Standard accounts of convention include notions of arbitrariness. But many have conceived of conventionality as an all or nothing affair. In this talk, I develop a framework for thinking of conventions as coming in degrees of arbitrariness. In doing so, I introduce an information theoretic measure intended to capture the degree to which a solution to a certain social problem could have been otherwise. As the paper argues, this framework can help improve explanation aimed at the cultural evolution of social traits. Good evolutionary explanations recognize that most functional traits are also conventional, at least to some degree, and vice versa.