Allison Parrish (decontextualize.com)
James Ryan (aleator.press)
Wendit Tnce Inf
7 x 4.5 x 0.5″
The asemic prose poems in this book were generated with a suite of generative adversarial networks (GANs). A GAN is a kind of machine-learning model that learns to produce images that resemble, but are not identical to, the images in a training dataset. To create a new poem, my program first samples word images from the latent spaces of several GANs that I trained (from scratch) on distinct datasets of bitmap images of English words: lowercase words, words with trailing punctuation, words with an initial capital letter, and words in a second typeface. Essentially, the sampling generates new words pixel by pixel. My program then places these words on the page, one after another, line by line, mimicking the process of typesetting English prose. The goal is to produce visual forms that call attention to the conventional appearances and operations of text: forms that afford “reading,” while remaining unreadable.
—Allison Parrish (poet)
Since the early 1960s, artists have been using computers to create printed matter that could not have been produced by others means. These works employ diverse poetics, take a variety of material forms, and emerge from an assortment of artistic tendencies. I am the proprietor of Aleator Press, a Minneapolis publisher and archive that aims to showcase the computer-generated book as a material cultural form with a long history and undeniable staying power. I am particularly interested in exploring (and subverting) aesthetic configurations that juxtapose (and synergize) the digital and the material. With Wendit Tnce Inf, Allison and I have produced a print artifact that is: computational, yet handmade; automated, yet labor intensive; cutting-edge, yet traditional; readable, yet unreadable. She generated the book contents using state-of-the-art AI technology, which I letterpress-printed and bound by hand.
—James Ryan (publisher, printer, binder)