A morning Habr digest featured an article from the sandbox — about Georges Papy's paper "minicomputer": children aged 7–9 added numbers by moving buttons across cardboard cards with sectors marked 1, 2, 4, and 8. The author promised a continuation, the article broke off mid-stream, and there was exactly one comment beneath it. Seemed like a charming episode from the history of pedagogy, a four-minute read.
But dig into the primary sources and the story unfolds into a drama with three fronts. First, behind the "cardboard abacus" stands cosmologist Georges Lemaître — yes, that one, the father of Big Bang theory: his 1954–1956 papers on machine calculation are explicitly listed in the bibliography of the children's methodology. Second, in the USA a federal project called CSMP was built on this minicomputer with a budget of "millions of dollars over many years" — and the verdict that officially buried the reform wasn't "the children couldn't handle it" but "the teachers didn't have time." Third, the pattern of resistance itself — "this is too complex, children won't understand anything" — is being reproduced in full today around neural networks in the classroom.
In short: I found myself looking at a story about how binary logic was twice given to six-year-olds — and twice killed by adults. I dug in — and I'm sharing.
It all started October 2, 1967 in a Brussels classroom. Frédérique Papy-Lenger — mathematician, wife of Georges Papy — hung a square card on the blackboard, lined into four colored sectors: white bottom left (1), red bottom right (2), purple top left (4), brown top right (8). Then — tokens. One token on white — one, on brown — eight. Two tokens in one sector collapse into one in the next higher sector: physical embodiment of binary carry.
Remember this moment, because it contains the entire idea: a machine without a single part. All the electronics — inside the child. The child doesn't "study" the binary system — the child becomes it. Frédérique later wrote that the exercises "produce harmonious cooperation between the intelligent human and the real machine that the minicomputer is in the pupils' eyes." In the eyes of six-year-olds, the cardboard with buttons is a real machine. And in some sense they're right: in their hands is a living one-bit processor that adds numbers simply by moving tokens.
The chronicle of lessons reads like an adventure novel. After two weeks children demand a third board — numbers crossed the hundred mark, and a boy named Didier declares: "Madame, I need another board!" — and triumphantly builds a hundred himself. After six months the same class tackles negative numbers through a sea battle game: black and red tokens "destroy" each other, and the children spontaneously invent the additive group of integers. Then fractions — a child asks for a board on the right and independently calls it "the board of small numbers," and the red line between boards becomes the future decimal separator. Dividing a hundred francs among three yields an infinite decimal tail — and one of the children remarks: "We'll be sitting here until morning." This is age six, Belgium, 1967–1968.
Frédérique Papy herself acknowledged: the minicomputer was "inspired by some of Lemaître's works." A few years ago I would have written this off as a polite gesture toward a famous compatriot. But in the bibliography of her 1970 article, Georges Lemaître's works are listed, and this isn't fiction: "Comment calculer" (1954), "Pourquoi de nouveaux chiffres?" (1955), "Le calcul élémentaire" (1956), and the book "Calculations sur machine" (Louvain, 1954).
So the father of the Big Bang in the mid-fifties — parallel to cosmology — systematically wrote about how to calculate and whether humanity needs new numerals. The man who in 1927 hand-calculated models of an expanding Universe, and in old age became fascinated with numerical analysis and "computer science," according to the Louvain archive, shared with Brussels educators the thought: binary representation is not exotic machinery, it's a way of thinking. Inspiration flowed from cosmology into children's didactics through one and the same Brussels circle — and that's perhaps the most beautiful part of this story.
Georges Papy — not just Frédérique's husband but a mathematician at the Free University of Brussels, moonlighting as a Belgian senator — led the country's "modern math" movement. In May 1965 Minister of Education Henri Janne, described by historians as "Papy's political friend," issued a circular declaring: from September 1, 1968, the new mathematics becomes mandatory for all secondary schools.
What followed was a real "battle of Hernani." Historian Dirk De Bock from Leuven, analyzing digitized Belgian newspapers of that era, identifies two camps: "Papists" and "anti-Papists." The war was fought not in academic journals but in the public sphere — with participation from parents, technical school teachers, opposition politicians. The new government in 1968 tried to give dissenters a delay — and thereby "gave wind to the opposition." But the finale was unexpected: September 1, 1969, all first-graders in Belgium, without a single exception, in general and technical sections, found themselves in a world of sets, arrow diagrams, and the minicomputer. Papy's reform is the only one of the "big" mathematics reforms of that time that won politically. De Bock formulates the diagnosis carefully but harshly: this was "a political, not a content-based dispute."
The American branch is the most instructive. In 1966 Burt Kaufman launched CSMP in Illinois — Comprehensive School Mathematics Program, a K–6 course for regular classrooms built on four "nonverbal languages": Papy's minicomputer, arrows (relations), strings (classification), and calculators (patterns). The method was Socratic: the class "buzzes," the teacher walks from desk to desk and "torpedoes" wrong theories with counterexamples. Frédérique Papy worked on the project personally from 1974–1980 — American Wikipedia calls her "the most influential figure in the project."
The final evaluation report (McREL, April 1984) opens with a killer phrase: CSMP became "the last of the great dinosaurs of federal mathematics curriculum development — an anachronism whose momentum and promise dragged it through one ordeal after another." An advisor predicted back in 1969: "It will never be widely used. It's an experiment — how good a program can we build without thinking about implementation." Nevertheless, by 1984 the program was alive in 150 school districts in 42 states, passed through by about 55,000 children, and federal support continued until 2003.
And here's the main discovery. In surveys of coordinators from the 150 districts that implemented CSMP, the first problem named was not mathematics and not children, but teacher retraining: "too great a change in teacher behavior and philosophy, not enough time or authority to train and monitor teachers, training cannot be conducted adequately." Half the responses on these points fall into the highest scale gradations — up to "decisive effect that causes rejection and cannot be overcome." Then come "back to basics" and the movement for testing and accountability.
The mechanics of failure are prosaic to the point of pain. The main constraints on training were "time and money": districts allocated two to three days per year for teacher professional development — and even that reluctantly. CSMP differed from everything teachers had learned in education schools. Coordinators honestly wrote that training requirements and the low emphasis on computational skills "will prevent the program from gaining wide distribution" — and they were right. The program didn't die from being too complex for children. It died from lack of two weeks over a teacher's entire career to master that complexity.
The formulation "children won't understand anything" from the Habr article is not a protocol quote but a pop legend grown from a real verdict. But the coordinators' actual verdict is much harsher and much more honest: they saw the problem not in children but in adult preparation. That's precisely why K–2, where the material was most "revolutionary" and most Papy-esque, turned out to be "the grades teachers liked most."
The Soviet branch is the most unexpected. Frédérique and Georges Papy's book "Children and Graphs. Teaching Mathematical Concepts to Six-Year-Old Children" actually came out at Pedagogika publishing house in 1974, with a foreword by Academician A. I. Markushevich — vice-president of the Academy of Pedagogical Sciences, the man who built the entire Soviet supplementary math education system in the 1950s–60s. Print run — 75,000 copies, a format worthy of any modern bestseller. The story about "at Kolmogorov's insistence" from the Habr article found no direct documentary confirmation — but the context of Kolmogorov's school mathematics reform of the 1970s, which was fiercely criticized for set-theoretic ideology, makes the mathematical elite's interest in Papy's books plausible: graph theory, sets, relations — that's exactly the "new mathematics" that reformers tried to bring into the classroom.
Yuri Neretin's preprint "Kolmogorov's Reform. A History of Errors" describes the collapse of that reform with jeweler's precision. September 1972: everyone studies the new "Planimetry." Resistance builds over years — "teachers were accused of insufficient qualification (with a significantly harsher subtext), they could be ignored. Ignoring millions of parents was impossible. In 1977 the wave reached universities." December 22, 1977, the CPSU Central Committee and government issue a resolution about "program overload with excessive information" — and counter-reform begins: Pogorelov, Atanasyan, Alexandrov. Children here too aren't the jurisdictional problem, but what adults argue over.
And now — a continuation that still brings joy. Papy's methodology didn't die: it mutated. In 1998 Tim Bell, Ian Witten, and Mike Fellows from the University of Canterbury in New Zealand released "Computer Science Unplugged" — a collection of offline activities where computer science is taught without computers: binary numbers on cards with dots, finite automata in treasure hunts, graph coloring in the unfortunate cartographer's problem. Materials — "cards, string, crayons and other common household items." Activity #1 — "Count the dots" — direct evolution of the 1-2-4-8 cardboard squares: five cards with dots, a child flips them, showing numbers, uniqueness of representation revealed as a magic trick, and the smallest result — a trick question that almost everyone answers "one," with the correct answer being zero.
Irina Derevianko, translator of CS Unplugged into Russian and author of that very Habr article, writes directly: "Papy's legacy today is experiencing a second birth within the global CS Unplugged movement." She is also retranslating Georges Papy's books — "detailed minicomputer guides were never translated in the USSR." Seventy years later, cards with sectors still await their full return to the Russian-language classroom.
The final chord is contemporary. Fall 2025, EdWeek Research Center surveyed 693 American educators: 50% received at least one AI training, versus 42% in 2024 and 13% in 2023. Progress looks rapid — but Pat Yongpradit from Code.org immediately clarifies: "A third of teachers attended exactly one one-off [session], and it could have been quite basic." The most effective professional development is "not one and done," but the budget and time schools are willing to allocate for this is exactly the same two to three days per year as in 1978.
The formula hasn't changed by a single bit. Technology arrives in the classroom → children master it faster than adults → the system hits the only real bottleneck: teacher preparation and motivation → politics starts arguing about whether "children will understand" → the decision is made behind children's backs. Papy, CSMP, Kolmogorov's reform, and the current hype around AI — it's one and the same knot, just retied.
I got carried away into the history of pedagogy, but what it smells of here isn't chalk — it's engineering. First: any educational reform has exactly one bottleneck — not students but adults. Belgium rubber-stamped the reform politically and "won" it. The USA funded pilots for decades and couldn't find money for two weeks of teacher retraining — and buried the result in archives. The USSR accused teachers of being unqualified and was punished by millions of unhappy parents. In all three cases, children, judging by the chronicles, were fine: they demanded a fourth board to build a thousand, and complained they wouldn't finish dividing a hundred francs by three until morning.
Second: innovation lives not in the patent but in the chain of carriers. Lemaître thought about computability — Frédérique Papy turned it into cardboard — CSMP formalized it into a program — CS Unplugged relaunched the idea in the era of personal computers — Derevianko translates it for the Russian-language classroom. Each carrier contributed something of their own, and none could bring the idea to mass implementation alone. This contradicts the myth of "the brilliant methodology everyone ignores": the methodology simply moves to where there's energy for it.
And third — the most unpleasant. All participants in this story share one attribution error: when reform stalled, children were appointed culprits ("it's too complex"), teachers ("they didn't understand"), politics ("wrong ministers"). But the real mechanism is always one: adults weren't given time and money to figure it out themselves. Until this mechanism is challenged, every new technology — from minicomputer to neural network — will arrive in the classroom twice: first time as revolution, second — as an archival find that someone will translate and republish again for the next generation of children.