by Jessie Roberts
“Litspam,” writes [author Finn] Brunton, came in the form of “cut-up literary texts statistically reassembled to take advantage of flaws in the design and deployment of [probabilistic] filters.” Drawing on the deep well of public domain texts available on sites like Project Gutenberg, this new spam could read like “fractured textual experiments”: automated Dada-like poetry combining such disparate material as Jules Verne novels, Nelly lyrics, and a long-lost recipe for pecan sandies. These “litspam generators,” the agents of what Brunton calls “spam’s modernism,” were designed to incorporate enough hostage literature to trick the probabilistic spam filters into classifying their output as authentically human. A 2005 email from “Clyde Blankenship” titled “Observations on my first Cia1!s experience” illustrates the “pure arbitrary utility” of a typical litspam pastiche:
forsaken is multiple gaseous a balmy not stannous cool.
indefatigable is pigtail coed is
bracket a commodious cyclist good.
lotus is greenberg catch a haulage not regis cool.
actaeon is elsinore mud is
familiarly a expiable toe good.
Brunton hesitates to liken such “poisoned Bayesian spew” to the work of human poets — even those employing avant-garde cut-up techniques — instead comparing litspam to “flipping through channels on a television.” To date, none of the anonymous programmers responsible for the machinery behind litspam have stepped forward to claim authorship, but their voices would lend an intriguing new dimension to this period.