Understanding protein binding specificity, the tendency of proteins to preferentially bind some molecules and not others, is a general problem in many fields of structural biology: Binding preferences control the propensity of diseases to resist treatment, of antibodies to support immunity, and our own abilities to design therapeutic drugs. Specificity is a consequence of molecular shape and charge at sites where proteins bind to other molecules.  This talk will present a system for comparing solid representations of protein shape and charge at different binding sites.  We compare solid representations using Boolean set operations from Constructive Solid Geometry (CSG).  We demonstrate first that we can identify conserved and varying parts of binding sites that correspond to similarities and differences in binding specificity.  From this starting point, we advance our technique, describing how algorithms based on many CSG operations and on the statistical analysis of the resulting solids can systematically deconstruct binding sites.  We demonstrate how deconstruction reveals molecular components that influence binding specificity, reproducing observations established in existing experimental results.  The results and methods described illustrate a new kind of volumetric algorithm for the aggregate analysis geometric objects.