Quantcast
Channel: Technology | The Atlantic
Viewing all articles
Browse latest Browse all 7171

Moby Dick in 50,000 Meows and Other Tales That Computers Tell

$
0
0

A novel, if stripped of plot, characters, or details, is simply a collection of organized words. Which means anyone can "write" a novel by stringing words together—even if that novel focuses only on one word, repeated over and over. No plot, no deeper meaning, no matter.

That's the idea behind the National Novel Generation Month (NaNoGenMo). Created by programmer Darius Kazemi last year, NaNoGenMo is an offshoot of the popular National Novel Writing Month, or NaNoWriMo. In NaNoWriMo, participants spend the month of November writing a 50,000-word-plus novel; in NaNoGenMo, they write algorithms to generate novels.

Finland-based developer Hugo van Kemenade created two interesting, and quite distinct, entries for this year's NaNoGenMo: 50,000 Meows and Gutenstory. Both use algorithms that hone in on specific words or phrases from classic works available on Project Gutenberg. And both make little sense.

In the former, his code replaces every word in a novel with the Internet-appropriate word "meow," but preserves sentence structure in the process. "Just 50,000 plain meows would be too simple and rather dull," he tells me in an email. But he says that people probably aren't interested in the project because of its complexity. "[The generated novel] is probably popular because it's just such a stupid idea."

Take the iconic opening of Herman Melville's Moby Dick, for example:

CHAPTER 1. Loomings.

Call me Ishmael. Some years ago—never mind how long precisely—having little or no money in my purse, and nothing particular to interest me on shore, I thought I would sail about a little and see the watery part of the world.

In meow form, that's:

MEOOOOW m. Meeeeeow.

Meow me Meeeeow. Meow meeow mew—meoow meow mew meow meeeeooow—meeeow meeeow me me meeow me me meoow, mew meeeoow meeeooooow me meooooow me me meeow, M meeooow M meoow meow meoow m meooow mew mew mew meooow meow me mew meeow.

It's easier to see the effect in plain text (below): The meows replace the words with the same number of letters, rhythm (as much as you can get from the syllables "me" and "ow," anyway), and format of the original text.

50,000 Meows

Van Kemenade's other entry, Gutenstory, borrows from classic literature as well. Instead of replacing the words, though, his algorithm combed 3.6 million possible sentences provided by Project Gutenberg and extracted sentences containing select words or phrases, most related to fairy tales—things like "so," "but why?" and "and then!" The program than compiled the sentences and listed them into chapters, each beginning with a sentence with the phrase "once upon a time," giving it a fairy-tale-like format.

But again, Gutenstory does more than just string the words together; the algorithm sorts the sentences by length, to give the chapters "a poetic effect," Van Kemenade says, "and to give you a chance to see the pattern."

Chapter 8, on the word "so," demonstrates the nonsensical-but-poetic pattern-making:

Once upon a time, the Foxes were angry with Sun...

So Baldur was prevented from coming back to Asgard.
So I went forth before them crying out, 'Slack, my master!'
So he chose to feign dullness, and pretend and utter lack of wits.
So Saul died, and his three sons, and all his house fell together.

And so on. Every sentence after the first begins with the "so," creating a semblance of poetry and rhythm. There are over 600 possible Project Gutenberg books these sentences could come from, ranging from works by Aristotle to Oscar Wilde. In the above passage, the first sentence is from a Native American legend, the second from a myth, the third from The Book of the Thousand Nights and a Night, the fourth from the Danish story that inspired Shakespeare's Hamlet, and the fifth from the Bible.

"I needed a big corpus of text to work with," Van Kemenade says. "By pulling out abstracted sentences away from their origins, and linking only by a common phrase, you get this interesting mix."

Sure, there are presumably infinite possibilities for text produced by algorithms, but when writing these algorithms, Van Kemenade felt the need to retain a novel-like structure. Even without meaning, these passages sound like novels, with intonation and patterns and the occasional, serendipitous combination of sentences that form a story. "Once upon a time, a mouse, a bird, and a sausage, entered into partnership and set up house together," reads Chapter 1 of Gutenstory. "I say, Tit! I am happy. I hate him! I'll be off."

And of course, Van Kemenade was just one of more than 100 participants in this year's NaNoGenMo, which produced 85 completed entries in total. Among those entries were graphic novels, novels created out of spam comments, and novels based on dream interpretations. All combined words spit out by computers with a human touch—even if some of those words were just variations on a cat's meow.

This article was originally published at http://www.theatlantic.com/technology/archive/2014/12/moby-dick-in-50000-meows-and-other-tales-that-computers-tell/383340/









Viewing all articles
Browse latest Browse all 7171

Trending Articles