This ambitiously titled conceptual poem is generated from Twitter feeds, selected to produce an endless stream of rhyming couplets. As of this posting, the program (developed with MooTools) has generated 1,353,298 verses and continues to generate about 4000 verses each day.
How does one go about reading a poem of such magnitude? This kind of poem— like Dan Waber’s Sestinas, Jim Andrews’ Stir Fry Texts, and Nick Montfort’s Taroko Gorge— is too large for a human being to read in its entirety, even though they are conceptually within human grasp. There is a range of expression in a generative algorithm that leads to an expressive pattern, even if the dataset is inexhaustible, and prolonged exposure to the text leads to insight on the pattern produced by the algorithm. Seen from this perspective, the text of the poem is and isn’t just the generated couplets that we read, nor is it just the code that generates the couplets, nor just the Tweets it crawls to find suitable pairs. The poem is the conceptual space between all these texts: its linguistic and poetic patterns, its range of expression, and the moment it becomes tiresome to readers because it has ceased to produce fresh combinations.
Think about the algorithm used to match rhyming words at the end of the poem. Does it produce anything beyond perfect end rhymes in the last word of the tweet? What keeps rhyme interesting is its variations, near hits and near misses, its ability to go beyond the last word in the line, and to even have visual resonance (eye rhymes), and substitution with alliteration, and other devices.
Perhaps the longest poem in the world isn’t that long, after all.