We have been working on a representation of the DDI 4 model in R, realized as a package. We now have an R-object-oriented class for each DDI 4 class with associated functions to validate and print objects, manage a registry of DDI 4 objects, manage DDI 4 URNs, and import and export DDI 4 XML. Our original goal was to enhance the ability of researchers to capture and report on metadata at the source, with the ability to embed references to metadata. In this presentation, we will discuss the intriguing prospect of computing directly on the metadata. What could be done with these metadata objects to facilitate comparison and harmonization? In what ways could the metadata be visualized? DDI 4 has powerful new capabilities in the collections pattern. For classes realizing a collection, operators could be defined to return their intersection, union and difference. Inner and outer joins could also be defined. Relationships within the collection could be visualized via network diagrams. These operators might provide efficient tools for harmonization. Operators could also be defined on pairs of objects of the same class. Similarity measures could be computed using corresponding attributes. These could be used to create visualizations (e.g., of similarities among variables).