The ToB, presented by Field Notes, is here!

It's the 2023 Tournament of Books, presented by Field Notes! And it's finals week! Dig in!


Programming in English, or Translate :: ParseLib :: English

Computer code may not be gobbledygook, but that doesn’t make it art. A survey of the field of programming-cum-poetry to find the ghost (of Hamlet’s father) in the machine.

I’d love to believe “computer code is poetry,” as some programmers do, but it’s just not true. Code can be elegant, intricate, and overwhelmingly complex, but nobody’s reading C++ debugging output on a lark. If you think math jokes are dorkish, try pitching woo with a complicated regular expression match. There may be some beauty in how a Perl library creates amorphous geometric shapes based on Beethoven’s 5th Symphony, but this is a rare instance; few people care about the algorithmic artistry behind a plumbing company’s e-commerce engine. Sadly, programming is a brutish, pragmatic science—all function, no form—whose only value to most people is in what it accomplishes.

That’s not to say the inverse isn’t true. Poetry is code in both senses of the word: code in the cryptological sense, in that it can obfuscate the writer’s true message (flowers aren’t actually pretty, autumn isn’t all that wistful), and it can be code in the sense that it’s a sequence of declarations in a human-readable format. Most programming is intended for computers to process, but that doesn’t mean we can’t read it. For example, the following poem can be interpreted like a simple Fortran statement:

Roses are red,
Violets are blue,
I have gonorrhea,
and now so do you.

In the first three lines, we define variables with declarations, and in the last two we use a simple function to deduce whether an instance of the class of humanoids has the property “DiseaseVector->gonorrhea.” It’s an efficient way of informing someone of their bacterial infestation, but it’s not a very moving poem. When we approach a piece of text as though we’re computers, reading becomes only a pursuit of necessary information, and when you eliminate all of the information in the world that isn’t necessary, you’re really left with little to read—instructional manuals, trade publications, legal documents, political voting records, statistics. English is a messy language. It may not be as romantic as French, as lyrical as Italian, or as pragmatic as Japanese, but if you want a programming language mired in centuries of miscegenated dialects and slang, double entendres, and a bloated dictionary of functions that readily lends itself to reinterpretation, you can’t do better than English.

The purpose of writing code is to achieve the cleanest results in the fewest lines. Closer to human speech, language becomes more complicated and enigmatic; we find meaning in between what’s being said. And at the far end of the spectrum, and in some poetry MFA classrooms, is complete garble, however deliberate. There are few examples of human-made, truly random language. The best example might be the last words of Dutch Schultz—the exasperated ramblings of the mafia boss on his deathbed from a gunshot wound as told to a police officer.

Police are looking for you allover. Be instrumental in letting us know. They are English-men and they are a type I don’t know who is best, they or us. Oh, sir, get the doll a roofing. You can play jacks and girls do that with a soft ball and do tricks with it. I take all events into consideration. No. No. And it is no. It is confused and its says no. A boy has never wept nor dashed a thousand kim. Did you hear me?

Pure ecstatic genius, though who knows what it all means? To intentionally write something so incoherently random and dense is difficult without your intentions being laid bare for all to see. Now compare it to the output of a chatterbot called Ellary that creates poems based on text found in IRC discussions:

Just because it says here I like tomatoes,
is that a reason to call off victory? Yet it says,
in such an understated way, that this is a small museum
of tints. I’m barely twenty-six, have been on “Oprah”
and such. The almost invisible blight
of the present bursts in on us.

Actually, that’s from a recent poem by John Ashbery, but reading it under the assumption that it was produced by a random calculation rather than a human somehow lends it more credence. If you go back and read it knowing that it was written by a human, there’s an air of cynicism to the text. What’s he trying to say about Oprah being like “the invisible blight of the present?” I mean, tell us what you really think. When you read it as written by a chatbot, it all sounds fascinating, that the randomness of nature could produce such a wise statement about Oprah. Authorship somehow taints the whole process.

The simplest way to eliminate the burden of authorship is through a dissociative mechanism like the cut-up technique—the rearrangement of words at random, perhaps by slicing them up and grabbing them from a hat, then putting them together to create a new literary object whose meaning is overflowing with subtext and artifice. By relieving the writer of the responsibility of meaning, it allows the reader to interpret things as they see fit.

Here’s a sample of a Markov Chain cut-up of the short story Sticks by George Saunders:

One Christmas Eve he shrieked consisted of cupcakes, no ice cream.
our own,
found the seeds of meanness painted a rift in crucifix he’d built out of Kimmie for wasting an apple slice.
He over she said: sticks around the yard.
He for wasting an apple slice.
He covered it with cotton swabs
the pole on its who yanked out the he shrieked at Rod had to clear it with sweatshirts, tubes lugged out a floodlight a sign saying lugged out a floodlight to ensure a soldier, on Halloween a ghost. he’d built out of metal pole swabs that winter for good enough good enough good
Mom died and he dressed the good enough good enough.
Birthday Veterans Day a hovered over us as taped to the string taped to the string letters of apology, over it on Groundhog Day and lugged seeds of meanness blooming also within in a jersey and Rod’s helmet autumn he painted the pole bright yellow. and I sat there talismans from his youth cream. The first I brought tubes of Mom’s makeup. One taped to the shrieked at Kimmie for yellow. He covered had children of our own, found the over she said: what’s with your he lay the pole on by hammering in six crossed concession to glee.

The idea behind the cut-up technique is not to improve the source material, but to create something unknown, mesmerizing, and new, like “[a shriek that] consisted of cupcakes, no ice cream.” In the 1920s, Tristan Tzara was expelled from the Surrealist Movement by André Breton after his demonstration of the cut-up technique in Paris threatened to put the whole poetry industry out of business by just grabbing words out of a hat, turning a centuries-old art form into a parlor trick. Brion Gysin formalized the technique in the 1950s, which then was used by William S. Burroughs as a way to prevent language from turning into Orwellian Newspeak—the hyper-rational language of the future where diction is reduced to its most utilitarian form, eliminating all shades of interpretation. Burroughs thought cutting up and rearranging words allowed a deeper insight into a body of text than a direct interpretation.

Today the accessibility of programmable computers have given every Jean-Claude Turtleneck the opportunity to create reams of unique poetry at the drop of a hat, should they wish. Anybody with a basic knowledge of Perl and some time can write an algorithm to parse the CNN homepage into avant-garde philosophical musings. The Unix Emacs program has a built-in text scrambling function. Email spam arrives in every person’s mailbox filled with guerilla poetry to thwart Bayesian filtering systems. Surrealist games, like online Exquisite Corpses and poetry generators, that might have at one time caused riots in Paris, are now relegated to 10-year old Geocities pages covered in banner ads. Advancements in dissociative programming have moved on from creative language interpretations to more complicated output like 3-D data visualizations of traffic patterns. The best literary cut-up technique of recent could be Garkov—the Markov chain recalculation of Garfield comics. And even that reinterpretation is superseded by the less experimental Garfield Minus Garfield for its psychological underpinnings. Random computer-generated text is readily available, but also readily ignored.

There’s not much room for experimental fiction when you’ve got infinite amounts of information at your fingertips and all the world’s sexual deviancy. Which is unfortunate, since, if there ever were a perfect symbiosis of programming and writing, the cut-up technique would be it. Beyond the Ouija board curiosity of being able to decode an essayist’s true intentions by scrambling their words, there’s a gentler art of creating something new and strange without being considered pretentious.