Computer-generated literary art has been developed for decades across languages and genres, a history we survey. In recent years, it has increasingly been produced by author/programmers: people who fully engage with literary and computational aspects and who sustain a practice in digital literary art rather than undertaking a single experiment. Focusing on those cases where computer systems produce output that is then published, we consider computer-generated literary work in poetry, prose fiction, and minor genres from the 1950s up though the beginning of the 2020s. We argue that such literature uniquely explores machine voices, computer cognition, and other aspects of computing within culture. The text generation systems that are the most hyped at the beginning of 2023 (ChatGPT foremost among them) are proprietary, opaque, and presented as entirely novel, with no historical precedent. In our discussion, we provide several ways to see how an understanding of computing and computer programs gives insight into the rich history of literary text generation and relates to important ongoing work in the area.