For as long as the TEI has existed, the needs of scholars to interpret and express the texts they encode has sat uncomfortably with the requirements of developers that the code they write be interoperable and usable beyond a single text or a single project. Fundamentally, software development for textual scholarship needs is just as much an act of interpretation as the encoding itself, and interoperability is just as problematic as any exact repetition of scholarly methodology ever is. This is far from being bad news, however! When code is interpretation, it is also scholarship. By breaking away from the concept of code-as-toolbox in favor of code-as-scholarship, we can make a true and decisive argument for the place of the digital in the humanities.