It is possible to interpret Frege's "sense" of a term A as an (abstract, idealized) algorithm which computes the denotation of A. With a rigorous notion of "algorithm", this view leads to a precise modeling of meanings and a simple calculus of synonymies, which, in turn, yields plausible, algorithmic explanations for some of the classical puzzles in the theory of meaning. This talk will focus on explaining this calculus and applying it, especially to puzzles which involve context, e.g., the difference in meaning between "I am tall" and "John is tall" when both these sentences are uttered by John.