When I was in my twenties, Italo Calvino’s 1979 meta-novel, If on a Winter’s Night a Traveler, was one of my favorites. Working in bookstores and libraries, I recommended it to anyone who seemed like they might be interested in literary puzzles and experimental narratives.
Both innovative and pleasurable, the novel repeatedly puts forth and discards formal devices (e.g. mystery novel, romance), hopping from form to form like an interdimensional traveler. The book has a kind of intentional short-attention span that uses its narrative structure as a tool to reflect on desire and curiosity, consumerism and austerity, voyeurism and exhibitionism, on closure and openness.
Recently I decided to reread the novel because I remembered that there was some kind of precursor to today’s LLMs in the book. Computers play a relatively minor role in the novel, but the specific ways that Calvino thought about the intersection between computational thought and literature always stuck with me.
We first hear about literary machines through Silas Flannery – a writer who is somehow both a satirical version of the tortured artist and also a mystical figure of literary creation. Flannery struggles to finish the draft of a novel but at one point he hands over the draft to be completed by machines:
Computers would be capable of completing it easily, programmed as they are to develop all the elements of a text with perfect fidelity to the stylistic and conceptual models of the author.
While the concept of a computer that could complete a story, matching its style and intention, was pure fantasy in 1979, it is closer to reality with tools like ChatGPT. It’s not hard to imagine that given the recent progress in AI, we are not far from a world where we can feed twenty pages of a novel and a few prompts into an algorithm and get something recognizable as fiction – probably something extremely weird!
Another computational model that Calvino explores in If on a Winter’s Night a Traveler is the concept of interpreting a novel by text-mining most frequently used words. In this passage, a character named Lotaria describes how she tabulates the most frequently used words to pick out themes and tendencies within novels without having to read them:
An electronic reading supplies me with a list of frequencies, which I have only to glance at to form an idea of the problems the book suggests to my critical study. Naturally, at the highest frequencies the list records countless articles, pronouns, particles, but I don’t pay them any attention. I head straight for the words richest in meaning; they can give me a fairly precise notion of the book.
Calvino even anticipates the process of filtering out stop words (such as articles and prepositions) in order to get at the “words richest in meaning.” It seems likely that we are to understand this as an impoverished method for consuming novels, that Lotaria’s mode of reading is not actually reading. But Calvino’s writings elsewhere suggest that he remained preoccupied with the question of how machines impacted our ways of reading, as I discuss below.
Later in the novel, we read Silas Flannery ruminating about the computational method of analyzing books.
Perhaps instead of a book I could write lists of words, in alphabetical order, an avalanche of isolated words which express a truth that I still do not know, and from which the computer, reversing its program, could construct the book, my book.
Again, Calvino anticipates a kind of generative AI that takes the raw material of a prompt and fashions a potential text. The novel swirls around these computational models with a frequency that indicates a level of curiosity that goes beyond satire. As part of the Oulipo group, Calvino collaborated with computer programmers, mathematicians, and other authors that sought to construct experimental literary works derived from unusual formal constraints.
Calvino’s 1967 lecture “Cybernetics and Ghosts” provides another articulation of the author’s preoccupation with the relationship between computers in literature. Among the various points, he claims that “Writers, as they have always been up to now, are already writing machines; or at least they are when things are going well.” Portraying the craft of writing as a way of cobbling together various elements and forms in a multitude of combinations – trying things out until they click – he depicts literary writing as a “struggle to escape from the confines of language” through permutation.
The vision of computer-aided imagination is not necessarily without the place of the human – but Calvino puts more emphasis on the human role of interpretation than on composition. For any machine can happen upon the text of Hamlet or Invisible Cities through enough trial and error – but it takes a reader to recognize that this particular set of marks has the power to change our perception of the world.
Literature can work in a critical vein or to confirm things as they are and as we know them to be. The boundary is not always clearly marked, and I would say on this score the spirit in which one reads is decisive: it is up to the reader to see to it that literature exerts its critical force, and this can occur independently of the author’s intentions.
The upshot of this claim is that the critical power of literature – its potential to allow us to imagine better worlds than the one we live in – is only implicit in the text itself and needs to be activated by the reader. This means it is not enough to passively consume the “lessons of literature” – be it produced by humans or machines – for it to have force, we need to actually think about it. As a result, Calvino encourages us to actively engage with the texts we read.