The False Idolatry of Data: Calvino’s Numbers in the Dark
Tech and Humanities Series
Published posthumously, Italo Calvino’s Numbers in the Dark (2002) is a provocative collection of stories written between 1943 and 1984 that explores the effects of digital technologies and data.
Calvino (1923-1985) was an Italian author, most known for his book, Invisible Cities and If On a Winter’s Night a Traveler. Although Numbers in the Dark is not his best work, there are many eloquent thought-provoking stories that are quite prescient. I’ll discuss a few below.
Data as an End in Itself:
“World Memory”
“World Memory” (1968) is a story far ahead of its time. It involves a narrator who has resigned from being the director of a project and who is speaking to his successor, Müller. The project involves amassing “the greatest document centre ever conceived, an archive that will bring together and catalogue everything that is known about every person, animal and thing . . . A general and simultaneous history of everything, or rather a catalogue of everything moment my moment.”
The project is being undertaken in “expectation of an imminent disappearance of life on Earth.” The narrator notes that “this material is being put through a reduction process that brings it down to the essential, condensed, miniaturized minimum.” “What we are planning to build is a centralized archive of the human kind,” the narrator states, “and we are attempting to store it in the smallest possible space.” He discusses what ultimately matters will be what is conveyed to some other life forms in the universe once humanity is gone.
The narrator wonders whether the data being collected is what really matters, whether the exclusion of mundane trivialities of life might be omitting what matters most. He states that “as soon as we concentrate on something, we immediately want to include it in our files; with the result, I confess, that I have often found myself cataloguing yawns, pimples, unhelpful associations of ideas, little tunes I’ve whistled, and then hiding them amongst the mass of more useful information.”
Another wrinkle is added to the story when the narrator discusses the introduction of lies into the data: “Only in a superficial sense can lies be said to exclude the truth; you will be aware that in many cases lies—the patient’s lies to the psychoanalyst, for example—are just as revealing as the truth, if not more so.”
He then concludes that sometimes “the lie is the real information we have to pass on.” He discusses how he described his relationship with his wife Angela as “a great love story” but that this is false; “our life was one long trail of sourness and subterfuge.” But, in “the world memory Angela’s image is definitive, perfect, nothing can taint it and I will always be the most enviable husband there ever was.”
The story takes a darker turn when he confesses to murdering Angela because what really matters is the “information-Angela who would live as long as the universe itself.” The real Angela was an impediment to the information-Angela who had to be eliminated. There’s one more twist, but I’ll stop the summary here.
The world memory project anticipates modern AI, trained on massive amounts of data, where the mass of data is viewed as more important than the individual ingredients. Judgments must be made about what data counts and what doesn’t.
The story captures the fetishization of data, how the project of amassing and analyzing data can transform from a means to an end in itself, with AI becoming a replacement for the reality it was created to understand and augment. The last segment of the story shows the dark implications, where real lives are replaced with data, where we prefer the digital representation of humanity to humanity itself.
The Limits of Data:
“The Burning of the Abominable House”
“The Burning of the Abominable House” (1973) was written at the request of IBM to craft a story using a computer—an ancestor to modern writing with AI.
The story involves a protagonist who is trying to use a computer to piece together the events in a house that burned to the ground, killing four inhabitants. The only thing that remains is a plastic folder that has a list of twelve “abominable deeds committed in this house.” The narrator concludes that there are billions of permutations as to what might have happened. But something must be sorted out for insurance purposes. The narrator imagines various happenings. “I can’t prevent the slow tentacles of my mind from advancing one hypothesis at a time, exploring labyrinths of consequence that magnetic memories would run though in a nanosecond.” The narrator notes that he “can arrange the facts according to a different logic . . . Bring the number of characters to six, introduce new galaxies of combinations and permutations.”
While the story is fun and playful, it lacks heft. It demonstrates that while computers can surpass humans in making calculations, they can’t equal the infinite possibilities of human creativity.
The Corruption of Data:
“Numbers in the Dark”
“Numbers in the Dark” (1958) involves a boy (Paolino) who tags along with his mother who cleans corporate offices at night. The offices are cold and impersonal, a place of paper pushing and number crunching. A man still in an office full of machines tells him: “There’ll come the day when machines like this do all the work . . . With no need for anybody, not even me.” The “machines didn’t produce anything, but managed all the company’s business, they looked after the accounts, they knew everything that happened and that was going to happen.”
Later on, the boy encounters an accountant. The boy asks him what he’s doing, and the accountant says he’ll never finish his work and that the accounts never come out right. The boy asks why the accounting is never right because “aren’t there electronic machines that do all the sums on their own?”
“All wrong,” the accountant replies. “Those machines make mistakes?” Paolino asks. The accountant says “It was wrong from the start.” He shows the boy a dim room in the company’s vaults where old ledgers are kept. He takes out a ledger with an addition mistake that the accountant noticed, made more than a century ago:
Over all these years, you know what that mistake of four hundred and ten lire has become? Billions! Billions! The calculating machines and electronic brains and whatnot can grind out numbers all they like. The mistake is right at the core, beneath all their numbers, and it’s growing bigger and bigger and bigger!
The story has a melancholy vibe. We feel the inevitability of machines taking over, the emptiness of the entire business enterprise. Even though the story isn’t about AI explicitly, it conveys a truth that underpins AI, which is that AI is only as good as its underlying data, and that data can be corrupted.
* * *
Calvino’s Numbers in the Dark still has relevance for our times. The stories are a meditation on how humanity is placing too much faith in data and the costs of this false idolatry.
Daniel J. Solove is the Bernard Professor of Intellectual Property and Technology Law at the George Washington University Law School. He is the founder of TeachPrivacy, a company that provides workforce privacy, cybersecurity security, and AI training to companies and organizations around the world. He is the author of 10+ books and 100+ articles.
You can follow his events, writings, training, cartoons, and resources by subscribing to his free weekly newsletter.
Thank you for surfacing these; World Memory reads like a GDPR nightmare written fifty years early. The accountant's 410 lire is the truest of the three from where I sit: every data-mapping remediation eventually finds its century-old ledger error, the mislabeled field or wrong lawful basis that has been compounding through every system built on top of it. We are now training models on those ledgers. Calvino understood that the mistake at the core does not average out; it scales.