Data types and codata types in programming languages are alternative and dual ways to define types and their inhabitants. Data types specify canonical values, whereas codata types specify canonical observations. When designing languages, one strives for a small set of orthogonal features, which we think can be in parts achieved by using the duality of data and codata types. We developed a highly symmetric language based on a system of Zeilberger [2008], using Girard’s concept of daimons to model extra-logical effects. In this talk, we would like to present this language and show some applications to the study of effects. It features a separation between intra- and extra-logical effects via built-in continuations and daimons, respectively. Furthermore it seems to be a good vehicle to study the duality between effects and coeffects by allowing concepts like delimited continuations to be dualized in a straightforward manner.