Abstract: Suppose that M is a mathematician, and that M has just proved theorem T. How is M to know if her result is truly new, or if T (or perhaps some equivalent reformulation of T) already exists in the literature? In general, answering this question is a nontrivial feat, and mistakes sometimes occur. We will discuss several existing databases of theorems which assign a small, language free, searchable "fingerprint" canonically to their theorems. We will also address the question: "How can we canonically fingerprint all theorems and formulas?" which is currently an unsolved problem. Some of the motivation for this discussion came from recent research at the intersection of combinatorics and probability pertaining to peak sets of permutations and statistical processes on graphs called meteors, earthworms and WIMPS. Some of these results will be explained as examples of the main theme. This talk is based on joint work with Chris Burdzy, Soumik Pal, Bruce Sagan, and Bridget Tenner.