I got this seminar announcement:

We will introduce a family of partition-valued Markov processes called exchangeable coalescent processes, and we will discuss four applications. We will explain how these processes describe ancestral processes in a discrete population model, how they describe the genealogy of continuous-state branching processes, how they can be used to model the effect of beneficial mutations on a population, and how one example called the Bolthausen- Sznitman coalescent is related to Derrida’s Generalized Random Energy Models.

Now, I wonder how many people who do probability know Derrida the critical theorist also know Derrida the statistical physicist. And vice versa, of course. Perhaps someone (Sokal?) should try applying generalized random energy models to texts.