Abstract
Computational poetry has redrawn the limits that putatively define what poems can and cannot do. Because of the major poetic transformations that the union of poetry and programmable machines has wrought, many critics have assumed that the conventional modes of print-based literary criticism must undergo a similarly profound transformation before we can apply them to computational poems. I argue against this assumption. Although I acknowledge that the criticism of computational texts must reckon with their specifically computational elements, and indeed enact this acknowledgment by performing several readings of code, I show that literary ideas from classical antiquity are surprisingly relevant to contemporary network-based texts. In chapter one, for example, I demonstrate how we can use fragments of classical text as a heuristic to help us interpret a poem (Nick Montfort and Stephanie Strickland’s Sea and Spar Between) that is, on the face of it, quintessentially nonfragmentary because it contains over 225,000,000,000,000 stanzas. In the second chapter, I show how we can use the classical literary-rhetorical mode of imitatio to theorize neural network poetry. Imitatio was a Greco-Roman pedagogical tool whereby students would imitate canonical poets or poetic modes to improve their compositional skill; I argue that the process by which neural networks learn to write poetry is fundamentally similar, and that this similarity allows us to recycle preexisting critical insights from the theory of imitatio into the theory of neural network poetry. Finally, in the third chapter, I firstly interpret the full-FACE poetry generation algorithm through the modern philosophical apparatus of moral luck. Then I look at John Cayley’s The Listeners (an aural computational text narrated by Amazon’s voice-transactive AI, Alexa) through the lens of a classical poetic genre, georgic, that is centrally concerned with agrarian labour. I show how Cayley’s text can be thought of as a post-classical participant in the georgic genre because it too is centrally concerned with harvesting, though of a more abstract variety than usual; The Listeners explores the ‘harvesting’ of language data by semi-intelligent computational actors, and I argue that we can see further into the terrain being explored if we stand on the shoulders of previous georgic scholarship.
This thesis demonstrates that classical literary modes, when used in conjunction with the computational tools that programmable media additionally require, are usefully applicable to computational poetics. In doing so, I demonstrate that the relationship between human and computer-generated poetry is one of collaboration and continuity rather than one of radical separation.