There is a moment, before the concert begins, when the orchestra performs a gesture that no one in the audience considers worthy of particular attention. The principal oboe sounds an A, and, for a few seconds, fifty or more people cease to be separate musicians: they listen, adjust, turn keys and tuning pegs by an imperceptible fraction. It is a ritual so firmly established that it seems part of the very nature of music, as though that A had always been there, waiting for someone to find it.
That A, however, has an administrative history. In February 1859, the French government passed a law setting standard pitch at 435 Hz, to halt a rise that had been going on for decades: orchestras were gradually raising their pitch in pursuit of a brighter sound, meanwhile making life difficult for singers, forced into ever higher tessituras. In 1939, an international conference moved the reference pitch to 440. In 1955, ISO ratified it in a technical standard, ISO 16, using the same procedure by which the pitch of a screw thread is standardised. Many orchestras today play at 442 or 443, because the race never really stopped.
The detail I find most instructive, however, concerns the past. Ensembles performing Baroque music tune to 415 Hz, a semitone lower, and listeners have the impression that they are finally hearing the right sound, the one Bach would have heard. In the eighteenth century, in fact, no such 415 existed: there were Venice, Leipzig, Paris and London, each with its own pitch, sometimes more than a semitone apart, and the same city might have one pitch for the organ and another for the theatre. The 415 is a recent convention, a compromise devised in the twentieth century to make it possible to perform together a repertoire whose origins were scattered. We invented a standard to represent an age that had none. Those two numbers, 440 and 415, conceal a double history. The first is a convention that, over time, ceased to look like a choice. The second is a still more curious convention: a modern choice that helps us imagine a past that never truly existed in that form. It is this second mechanism, more than the first, that touches many aspects of our lives.
When a choice stops looking like a choice
The process by which a practice arising from a choice comes to be perceived as an immutable fact was precisely described in 1966 by two sociologists, Peter Berger and Thomas Luckmann. At first, someone solves a problem in a particular way and then repeats that solution, not least because repeating it costs less than deciding everything afresh. Then the repetition involves others, and at some point it survives those who began it as a custom: those who come later find the practice already entrenched and receive it as a feature of their surroundings, on a par with the climate or the language spoken there. The final stage is the quietest, because the practice loses every visible trace of its artificial origin and begins to behave like a property of the world. Once it reaches that point, a convention no longer appears as a choice, but as the natural way things are. It becomes the point from which we observe the world, and that is why it is so difficult to recognise.
There is a further depth to this, familiar to anyone who teaches an instrument: much of what we know is never conveyed through words alone. A pupil learns the correct position of the hand on the instrument because, after a few hundred correct repetitions, the movement of their arm eventually no longer requires attention: the body has made it its own. No verbal explanation can replace that process. When a convention reaches this depth, it ceases to be merely a shared idea and becomes a way of inhabiting the body. That is why it is so difficult to question through words alone. The violin’s chin rest, which seems as much a part of the instrument as the bridge or the tailpiece, is actually an invention by Louis Spohr dating from around 1820; the shoulder rest came later still, in the twentieth century. No violinist, while playing, feels they are using two modern accessories attached to an object that had managed without them for two centuries. They simply feel they are holding a violin.
The second origin
So far, we have followed the way a convention forgets its own birth. The next step is still more surprising: a convention that loses the memory of its origin does not remain without one: it ends up acquiring a new one. My favourite example concerns the classroom. A rectangular room, desks in rows, the teacher’s desk along one short side: this is the shape almost all of us picture when we think of a classroom, and there is an explanation ready to justify it. That arrangement, we are told, reproduces the factory: it was supposedly conceived in the industrial nineteenth century to accustom children to obedience and repetitive work, pupils lined up like parts on a conveyor belt, with an overseer at the front. It is a story that has circulated for decades, and today it is heard chiefly from those proposing to reform schools.
Historians of education treat it with suspicion and describe it as a rhetorical device, useful for presenting school as a relic of a bygone age: it is used more out of a desire to shape the future than to explain the past. The actual constraints were less ideological and more physical. Mid-nineteenth-century cities had to educate numbers of children on a scale never managed before, and dividing them by age into separate rooms was the simplest way to do so. Then there was the light: before electric lighting, a classroom could be only as deep as daylight could penetrate, which determined its proportions, and the windows were on one side only, the left, so that the right hand would not cast a shadow on the page. The shape of the room in which we spend many years of our school lives thus depends on the sun and the fact that most people write with their right hand. It is a humble origin, almost embarrassing in its concreteness, and indeed it is not the one we recount. The factory story belongs to a particular category of errors: those that survive because they do a job. It attributes an educational design and a vision of society to the classroom, where there were only the number of children to accommodate and the trajectory of the sun.
I would call this movement the tuning fork effect. Just like the A at 415 Hz, which is a twentieth-century decision heard as the sound of a distant age, many conventions forget their contingent origin and acquire another, grander than the one they actually had. The substitute past need not flatter: the factory is, to all intents and purposes, an accusation, and it works just the same, because it attributes intention where there was merely an accumulation of practical constraints. We no longer remember that these were reasonable choices; we begin to believe they could have been different.
Every convention that lasts long enough ends up with two origins. One is real, the product of circumstances and decisions; the other is narrated, older and more intentional. Over time, we forget the first and defend the second as though it had always existed.
When the first origin resurfaces, surprise is almost inevitable. Even today, much of the world measures education through teaching hours, academic credits and time spent in the classroom. Yet this criterion derives not from a theory of learning but from the Carnegie Unit, introduced in the United States in 1906. At its origin lay a very concrete problem: Andrew Carnegie’s foundation had to determine which universities could access a pension fund for professors. A simple, uniform and easily verifiable measure was needed. Time was chosen: roughly a hundred and twenty hours of teaching per year for each subject. Over time, that administrative solution became one of the most widespread ways of thinking about and measuring education. Today, we consider it almost natural, and every attempt to move beyond it encounters the same resistance we reserve for things that seem always to have existed.
The second origin often, though not always, takes shape without anyone deliberately constructing it: the Carnegie Foundation’s records have always been public. The story arises almost of its own accord, because a practice that has lost its rationale creates a void, and the mind tends to fill it with the explanation that seems simplest and most plausible, or simply nearest to hand. The cost emerges only over time. As long as a convention remembers its origin, it can be modified like any other decision. Once it has acquired a second origin, however, every change appears as a violation of the natural order of things.
Organisations learn to repeat
The phenomenon is easier to see when we move from people to organisations, because there it leaves written traces: a form, a procedure, an assessment criterion. A practice created to solve a problem thus ends up behaving like an autonomous structure, capable of reproducing itself even when that problem no longer exists.
Organisations learn, but they also learn to repeat. If a procedure worked at a particular stage, the likelihood grows that it will be retained even when the next stage calls for something else. Changing it seems costly, especially when the new solution is still uncertain. Routine thus becomes a memory that has stopped examining itself and asking questions.
Here the mechanism repeats itself in administrative form. The procedure remains identical, but the story sustaining it changes. When asked why something is done in a particular way, the answer is no longer: “because it once solved a specific problem”. It becomes: “because it is the correct, serious and professional way”. At that point, the convention parts company with the purpose that generated it and is treated as a principle. Questioning it seems almost to call into doubt the seriousness of the person doing the questioning.
Schools know this dynamic well. Teaching hours, the division of time into lessons, the mark: each of these tools arises to solve a concrete problem and, over time, comes to appear as the only possible way of organising learning.
This is the observation known as Goodhart’s law, which in its most common formulation runs as follows: when a measure becomes a target, it ceases to be a good measure. The indicator takes the place of the reality it is meant to represent. Thus the mark risks counting for more than learning, teaching hours for more than the time actually needed to learn, or attendance for more than participation. Little by little, the institution stops measuring what counts and begins making what it measures count.
What is happening now
The conventions examined so far are all at least a century old, which makes them convenient to observe: time has erased the memory of the reasons that brought them into being. It is more interesting to observe what is happening today, while the process is still under way. A convention is truly visible only when it is young.
A process of this kind is under way right now. In just a few years, formats that did not exist in 2020 have become established: the way a generative system is queried, the typical structure of the texts it returns, the question–answer–reformulation sequence that has become the standard form of assisted intellectual work. None of these formats derives from a theory of knowledge; they derive instead from design choices, technical constraints and commercial decisions made by a handful of companies. In a few decades, they may appear to be the natural way to collaborate with a machine, and someone will explain that things could not have turned out otherwise.
A study by Microsoft Research and Carnegie Mellon University, presented in 2025 and based on a sample of just over three hundred professionals, offers a useful finding: those who place greater trust in the generative system report exercising less critical thinking, while those with greater confidence in their own abilities exercise more. The study also shows that critical thinking does not disappear, but shifts: from production to verification, from doing to checking that the result has been properly constructed.
As a teacher, one observes something more specific, and less dramatic than it is often made out to be. Students use these tools chiefly to reduce the effort of the intermediate stage, that opaque stretch between question and answer in which one tries a route, abandons it and tries another. It is precisely there that understanding takes shape. A piece of work can be excellent and correct yet remain cognitively alien to the person submitting it; the difference emerges only later, when the problem changes and there is no longer an answer to reproduce.
Relying on knowledge we cannot verify directly is nothing new. Every society also lives by trust in experts. Today, however, this dependence takes a different form. Increasingly, the source from which we receive information can express itself with great naturalness, but this fluency does not necessarily coincide with competence. Recognising the difference requires precisely the critical thinking we risk exercising less and less.
Retuning
None of this implies distrust of conventions as such: a society without conventions would be unable to coordinate itself, transmit knowledge or even build institutions. The difficulty begins when we stop distinguishing them from the realities they organise, and the substitute past is insidious precisely because it reassures. It allows us to believe that a practice is older and more necessary than it really is, and makes us forget that someone, in a particular year and for particular reasons, made a decision.
Recognising this mechanism requires more than mere opposition, which often does no more than replace one convention with another equally invisible one. It requires returning to the origins of practices: what problem they were meant to solve, when they stopped being tools and began to seem the only possible way of doing things, what story has been built around them in the meantime. It is a rare exercise, because it means questioning the premises on which our actions rest, not just the actions themselves.
I return to the oboe at the beginning, because it contains something I had overlooked. What is interesting about the A is not only its arbitrariness, but the fact that it is re-established every time. Fifty or more professionals who have already tuned in the dressing room or backstage do it again in the hall, before every concert, and no one experiences the repetition as an admission of weakness: they know that instruments go out of tune with heat, humidity and the time elapsed since the last check.
I do not know many institutions that do the same. I wonder, then, what might resemble a tuning session for a school or an organisation: a public, recurring moment in which to ask again where what we are about to do really comes from, before we begin doing it.