Those who remain

Let me open with a confession of method, which in this case is also the subject. I have not read Giovanni Orsina’s book, Controrivoluzione. I have read the review written by my friend Flavio Felice — a friendship that took a thirty-year pause and that I am lucky to have found again, together with his writing. My access to Orsina is therefore second-hand: knowledge by description, not by experience. In a word I am about to steal: hearsay. Anyone writing about epistemology should always declare this; I declare it and make it my starting point. This is not an intervention in the historical-political debate, which is Flavio’s trade and not mine, and it will not become mine for having read a review. It is a scherzo — in the musical sense: a movement with a rigorous form that declares its own lightness — that an economist allows himself to play for a philosopher, built around a single word that is worth the price of the ticket.

1. The borrowed word

Orsina, Felice tells us, describes the fracture running through Western societies as a conflict between the hearsay of abstractions and the seen-with-my-own-eyes of concrete experience. He uses it as a political category. I take it and carry it home, because that pair of words is pure epistemology: it is the distinction between explicit and tacit knowledge that Michael Polanyi placed at the foundation of his work sixty years ago. We know more than we can say. There is a knowledge that lives only embodied, situated, in the first person — the seen-with-my-own-eyes — and a knowledge already detached from the experience that generated it, transmissible, stackable, quotable: hearsay. They are not degrees of the same thing. They are different regimes of knowing.

Hold on to that distinction, because now I am going to change subject abruptly. Or so it seems.

2. The hearsay machine

A large language model — the technology we today call, with semantic generosity, artificial intelligence — is trained exclusively on text: on what has been said and written. That is, on knowledge already made explicit, already separated from the experience that produced it. The seen-with-my-own-eyes is not in the dataset. Not because of a temporary technical limit, not because sensors or data are missing: by definition. The moment an experience is written down, it has already become hearsay. The machine does not have an experience deficit that more parameters could fill: it operates in an epistemic regime that structurally excludes first-person experience.

An LLM is, quite literally, the hearsay machine. That is not an insult: it is a definition. And it is an extraordinary machine — thirty years ago I was working on neural networks and I know how to recognise the greatness of what has been built. But its greatness sits entirely inside the perimeter of the already-said. This is what I mean when I say, in Kuhn’s vocabulary, that human and artificial intelligence are incommensurable: there is no common unit of measure, because they do not work in the same place.

Where the other place is, we shall see in a moment. First, a passage from the bigger picture.

3. Those who exit

The crisis Orsina describes in political terms can be reread, by constructivists, as a crisis in the social production of meaning. An order that progressively delegated the construction of meaning to formal systems — law, market, technology — discovered that those systems process information but do not generate inhabitable meaning: meaning an ordinary person can make her own within her daily experience, without having to convert to a vocabulary that does not belong to her.

And faced with uninhabitable meaning, the majority did not make a revolution. It did something quieter: it exited. Albert Hirschman taught us that people facing an organisation that disappoints them have two roads: voice or exit. Populism is voice — the noisy part, the part historians study. But the majority share of the phenomenon is exit: abstention, disaffection, resignation from the collective process of making meaning. And anyone who has run an organisation knows it: when customers stop complaining and simply leave, you have lost the most valuable signal you had — and you usually notice too late. Elites mistake exit for quiet.

It holds at the scale of a civilisation and it holds, in miniature, inside every company: the organisation that delegates the production of meaning to systems — yesterday metrics and dashboards, today artificial intelligence — does not harvest the revolt of its people. It harvests their inner exodus: execution without adherence, presence without participation. The silent majority exists in org charts too.

4. Those who remain

And here is the reversal, which is the real reason I am writing. Faced with the very same void — a world that offers no meaning — there is a figure who gives the opposite answer to exit. She does not protest and she does not leave: she remains. And she remains not because she has found a meaning others cannot see, but because she is master of a different process: giving meaning to something that has none. Vasco Rossi made a song out of it that every Italian knows; Joseph Schumpeter made an economic theory.

Because the Schumpeterian entrepreneur is exactly this: he does not answer an existing demand, he creates it. He does not discover combinations, he invents them. Creative destruction is the demolition of existing frames of meaning and the imposition of new frames that, before the entrepreneurial act, existed nowhere — not in the market, not in the data, not in any text. The entrepreneur acts before meaning exists, and by acting brings forth the environment to which meaning can later be given. The biologists of cognition — Maturana and Varela — call this movement enaction: the living being does not represent an already-given world, it brings it forth by acting. Meaning is a product of action, not its premise.

And here the hearsay machine meets its boundary — not of performance, but of nature. The machine operates only where meaning has already been deposited: text, patterns, the already-said. The entrepreneur operates in pre-semantic territory — where there is nothing yet to read, because significance will arrive after the act, as its consequence. It is not that artificial intelligence does this work badly: it is that for it this territory does not exist. There is no dataset of the not-yet-meaningful. The radical uncertainty Frank Knight wrote about is not insufficient information: it is the very absence of the distribution over which to compute. There, calculation is not weak. It is undefined.

Coda

The liberalism of the limit that Felice writes about finds here, it seems to me, its economic and anthropological counterpart. The limit of the machine is not a lack to be filled: it is the line that makes visible what only a person does. And what only a person does is not a nostalgic residue — it is the generative function of the entire economy: to remain where meaning is missing, and manufacture it. For others as well as for oneself: because the meaning an entrepreneur produces is either inhabitable by a community — colleagues, customers, territory — or it does not hold.

In an age divided between those who protest and those who exit, the most subversive figure is the one who remains. The scherzo ends here; the question it opens does not. And the ball, Flavio, is back with you: tell me where the economist played out of tune — it is what remains of my intellect, but it is all in play.

---

The prompt comes from Flavio Felice’s review of Giovanni Orsina, «Controrivoluzione. La storia politica del nostro tempo» (Marsilio, 2026). The debt for the hearsay / seen-with-my-own-eyes pair goes to the book’s author; the improper use I make of it here is entirely mine.