[Chat GPT 5.2; prompt at bottom]
The Hollerith Revivalists
There is something in the human animal that refuses to let a thing die properly.
We cook pizza in wood-fired ovens long after thermostats were perfected. We develop film in basements lit like Cold War submarines. We grind coffee beans with iron cranks, churn butter by hand, bind books with linen thread. We call it craft. We call it heritage. We call it “slowing down.”
And now — inevitably — that instinct has migrated to the most frictionless technology of all: generative artificial intelligence.
A small but steadily multiplying cohort of enthusiasts, scattered across Discord servers and vintage computing forums, believes that ChatGPT and other generative systems produce markedly superior results if prompted using punched Hollerith cards.
Yes. The stiff, beige rectangles with holes in them.
“You Have to Respect the Machine”
“I’m not anti-AI,” insists Leonard Kravitz, 34, a self-described Prompt Archivist, as he feeds a deck of cards into a refurbished IBM 029 keypunch in his garage in Akron. “I just think we’ve become sloppy.”
Kravitz believes that typing directly into a laptop produces what he calls “noisy intent.” The act of punching a card, by contrast, introduces constraint. You get eighty columns. No emojis. No dithering. No “just curious but…” hedging.
“It forces clarity,” he says, tapping the machine with priestly affection. “The model respects that.”
When asked how the cards, once punched, are transmitted to the cloud, Kravitz gestures vaguely toward a high-resolution scanner and a Python script. “That part’s modern,” he concedes. “We’re not Luddites.”
The Theory of Perforated Cognition
The movement has developed a quasi-philosophy. Its adherents speak of Perforated Cognition — the idea that physical holes encode intention more purely than pixels.
“Look,” says Dr. Miriam Falk, a former UX designer who now runs a newsletter titled Punch & Prompt, “AI is statistical. It responds to pattern density. Hollerith cards reduce linguistic flab. Every hole is deliberate. The absence of a hole is equally deliberate. That’s semantic minimalism.”
Dr. Falk claims — though no peer-reviewed study supports this — that she receives “30 to 40 percent more coherent essays” when submitting her prompts via card deck. She produces spreadsheets. The spreadsheets are printed on dot-matrix paper.
“Digital prompts are too easy,” she explains. “Anyone can mash keys. But when you have to physically punch a card, you mean it.”
Ritual as Optimization
Skeptics, of course, suggest that the improvement lies not in the cellulose medium but in the cognitive ritual.
“It’s pre-typing meditation,” says an anonymous computational linguist at a major West Coast AI lab. “If you force yourself to draft, condense, and commit to eighty columns before sending a prompt, of course you’ll get better results. You’ve edited yourself.”
The Hollerith revivalists reject this as reductionism. “You’re assuming the improvement is human,” Kravitz responds. “What if the model senses the card?”
He is not entirely joking.
The Reddit Boards
The movement appears to be spreading slowly, largely via message boards devoted to “deliberate computing.” One thread titled CardStacking for Cleaner Output runs 287 posts deep. Members trade advice on optimal cardstock thickness and debate whether IBM beige or Univac gray yields superior tone in poetic requests.
One poster claims that haiku composed via card deck display “greater ontological humility.”
Another insists that JSON output is “crisper when perforated.”
There are already splinter factions. The Purists demand manual punching only. The Hybrids allow laser-cut cards. The Neo-Jacquards are rumored to be experimenting with loom-encoded prompts woven into fabric.
A Visit to the Archive
In a Brooklyn co-working space decorated with Edison bulbs and visible brick, I meet a trio of twenty-somethings who call themselves The 80-Column Collective. They gather weekly to compose prompts on vintage hardware.
“We’re not Luddites,” says Sasha Nguyen, 27, adjusting a stack of cards with geometric precision. “We use AI constantly. We just don’t want it to feel like microwaving a Hot Pocket.”
Her colleague Marco adds: “It’s about friction. Friction creates authorship.”
They demonstrate by preparing a prompt requesting a satirical essay about municipal zoning laws. The drafting takes twenty minutes. The punching takes fifteen more. There is discussion about column alignment. A card is rejected for aesthetic imbalance.
When the result appears on screen — after scanning, OCR, and upload — the group reads it aloud reverently.
“It’s tighter,” Nguyen says. “You can feel it.”
The Irony Engine
What makes the Hollerith revival particularly exquisite is its timing. In an age when AI systems operate at planetary scale — trillions of parameters, data centers the size of suburbs — a small band of enthusiasts believes that salvation lies in cardstock rectangles designed for the 1890 census.
They have reintroduced delay into instantaneity.
They have reintroduced scarcity into abundance.
They have reintroduced holes into the seamless.
Anthropologists might see in this yet another cycle of modernity’s digestive process: first we automate, then we romanticize what we automated away. We install self-checkout lanes, then open “heritage markets” with hand-written price tags. We invent generative AI, then decide it works better if addressed as though it were a 1963 payroll system.
“It’s Not About Efficiency”
Before I leave Akron, I ask Kravitz whether he truly believes the AI “performs better” because of the cards.
He pauses.
“No,” he says finally. “I think we perform better.”
Then he smiles.
“But it’s more fun to pretend the machine cares.”
And perhaps that is the real heart of the movement. The Hollerith revivalists are not optimizing the model. They are optimizing the myth — the story that somewhere inside the vast probabilistic machinery, there is a faint appreciation for craft.
They are punching holes in cardstock not because the future demands it, but because the past feels reassuringly solid in the hand.
In a world where you can summon an essay in three seconds, it is oddly comforting to believe that excellence still requires a deck of computer punch cards, squared neatly on a desk, waiting their turn.
##
##
Prompt
Write one of our humorous alternate reality essays. Began by describing this human eccentricity of adoring the old-fashioned way of doing something which had been considered obsolete and bringing it back. Now let’s tie it to the modern and very new passion of interacting with chat bots with generative artificial intelligence. Let’s develop them as a scattered, but apparently slowly growing group multiplying by means of reddit boards and such, which believes in interacting with Chatbot and generative, artificial intelligence, the beliefs they get better results if they give the input on hollerith cards. Your essay wiill introduce this Looney topic and include typical journalistic tropes such as interviews with a few characters.
