A Twitter Bot That Generates Beautiful, Imaginary Moths

For centuries, devoted lepidopterists such as Vladimir Nabokov have painstakingly gathered and catalogued hundreds of winged arthropods, building impressive collections in glass cases that display moths’ myriad variations. A new project in the form of a Twitter bot has turned that tradition on its head, bringing it wholly into the digital age. Created by artists Loren Schmidt and Katie Rose Pipkin, @mothgenerator is a Twitter account that automatically produces images of make-believe moths, with the structures and colors of the insects’ bodies selected and combined by chance. Randomly generated nomenclatures also accompany each specimen so the resulting Twitter feed — which sends out images a few times each day — exists as a kind of online, Dadaist encyclopedia of moths, making it essential viewing during National Moth Week, which begins Saturday.

There’s the shouldered-signate two chholzi bufalli, a large insect with a zebra-stripe-patterned abdomen and violet-tinged hind wings; the granite-bogus moth mongusii amyelithophane, which has spiked magenta wings; and the sociable-anna moth brsalli approei, whose iridescent wings are attached to a delicate, skinny body. Pipkin said the pair chose moths as their subjects because they found their appearance especially fascinating. “They feel like living pieces of paper or tree bark,” they told Hyperallergic in an email. “They have such an unusual texture. I think they also seem strangely aware—more so than many other insects.”

Work on the bot began last month as a satellite project of inflorescence.city, Schmidt’s and Pipkin’s ongoing and evolving vision of an online metropolis that engages with random text and image generation. Written in JavaScript, @mothgenerator has a number of components that include processes to trace antennae and create wing shapes and markings, as Schmidt explained. The project is grounded in hand-drawing rather than in complex programming, however, with most of the work stemming from the pair’s continual refinement of their generated moths’ textures, colors, proportions, and patterns. To yield the mish-mashed names, Pipkin programmed a web crawler to collect around 4,000 real, English moth names and 10,000 Latin ones; while the program often matches whole English nouns with adjectives, the Latin words get spliced into phonemes, shuffled around, and pulled together so the linguistic bits form mostly gobbledygook.

Pipkin has been making other image-spawning Twitter bots for a year, including one that predicts changes in sea levels and another that forms bird’s migratory patterns from unicode. However, @mothgenerator is the first to churn out actual visuals. For them, the timeline is “both a platform and an artwork”; having an autonomous work exist in such a public forum allows for interactions that its makers didn’t anticipate and may not even witness.

“There is also a (relative) ease of access to a digital, and particularly social, artwork that may not be available with more traditional formats,” Pipkin wrote. “This is true both in the sense of being physically able to navigate an art space, but also in the structure of how one approaches engaging with a work.”

Although the moths currently exist primarily online, Schmidt and Pipkin have previously connected the generator to a receipt printer so people could produce moths and print out monochrome iterations of their favorite fictional arthropods to take home. They also have a private bug collection of a very different sort — screenshots of programming errors. “I might be fondest of those,” Pipkin wrote, “simply because after 10 hours into staring at the same articulated leg hair function, what one really needs is a laugh.”