Elegia Machina is a book of poems written in collaboration with a machine, more precisely a Markov chain. A Markov chain is a mathematical process that has been around for a long time and which has already been used, notably in music.
Some consider it to represent an embryonic form of artificial intelligence. Markov chains can be used to re- produce phenomena – texts, images, sounds, scents…– through respecting basic data, called “training data”. In the case of a text, the principle consists in making a list of all the words and all the punctuation marks – we’re talking about a “dictionary” – and calculating the probability that any given word or punctuation mark will follow another. Then, in using a random signal and the probabilities created in this way, we can create a text, an image, a sound “in the manner of…”.
For Elegia Machina, first of all we transcribed numerous poems by the authors we have always loved: Verlaine, Rimbaud, Apollinaire, but also Wilde, Tzara, not forgetting Aurélie Nemours. Putting Markov chains to work on each of these texts, we were able to create what we call a ‘gangue’: thus Verlaine’s gangue is a collection of 20,000 to 50,000 words which have Verlaine’s vocabulary, his flavour, sometimes even rhymes, but which have never been written by him. This creates numerous ‘epiphanies’: these are the little strokes of genius which we select and write down on strips of paper. From these strips, we see themes emerge, and we then assemble them in ‘songs’: that’s our contribution.
The songs were organized as a journey and they took on, to our amazement, a free, initiatory form. The way themes emerge is of course influenced by our own existence and travels.
Elegia Machina, book prototypes, artistic direction by Damas Froissart