Let us now return to the chronological thread. In 1985, I did not yet have the words I would use today to describe what was happening. I did not speak of the data economy, platforms, disintermediation, algorithmic markets, or information infrastructure. Some of that vocabulary simply did not exist.

Yet something was already there. An extremely physical intuition: when information circulates poorly, people work more, machines run needlessly, miles are traveled empty, money disappears, and those who possess the information retain an advantage that others cannot even measure.

So I did not enter computing because I loved computers. I entered it because I had understood that information itself could become matter on which one could act. It had to be taken from wherever it was scattered, structured, transmitted faster, and then we had to observe what this new circulation changed in the real world.

The electronic freight exchange, which takes us back to 1985, belongs to that founding moment. At first, the problem was almost trivial to describe.

Every day, France was crossed by hundreds of thousands of trucks. The large carriers already had sophisticated organizations, internal networks, and regular shippers. They were far better able to optimize their flows.

But several tens of thousands of small carriers, sometimes men alone with their truck, lived a different reality. They drove south loaded and returned empty. Or the reverse. They could travel hundreds of miles without freight simply because the information that would have allowed them to connect with a shipper along their route did not reach them at the right time.

That was what interested me. The truck was not the problem. The driver was not the problem. Neither was the freight. The problem was the absence of a connection between two pieces of information that already existed: here, someone needed to transport something; there, someone had an available truck.

Between the two, the information vacuum produced a physical vacuum. An empty truck. Fuel burned for nothing. Hours of work wasted. Margin destroyed. A road congested without any corresponding economic production.

I looked at these carriers as men condemned to hard labor. The word stayed with me. These men could work fifteen hours, sleep in their cabs, cover thousands of miles, and yet remain economically fragile because they did not possess the same quality of information as the large organizations.

Something was already there that would become a constant throughout my life: computing interested me when it could reduce an asymmetry. An economic asymmetry. A geographic asymmetry. Later, a legal asymmetry. Then a historical asymmetry in the Art Market.

But in 1985, I did not yet have that retrospective vision. I saw only a concrete problem and a new possibility. Make the information circulate quickly enough for supply to meet demand before the truck had already left. That is exactly the logic of an electronic marketplace before the term became commonplace.

I was not transporting anything. I was not becoming a carrier. I was not buying trucks. I was building an information space that allowed those who needed transportation and those who had available capacity to meet directly.

The freight exchange was less a transportation company than a system designed to reduce an economic asymmetry through electronic matching.

That shift may seem obvious today because we live surrounded by platforms. The phone in our pocket knows our location, aggregates supply, compares prices, connects strangers, and updates availability almost in real time. In 1985, none of that was natural. Everything had to be conceived from within a world that remained largely analog.

French telematics, however, gave us an extraordinary advantage: Minitel. France had a lead there that would often be underestimated after the arrival of the Internet. For an entire generation, Minitel was a school of connectivity. It taught the French that information no longer had to be locked inside a book, a directory, or a local file. It could be called up remotely. Queried. Updated. And above all, it could be different tomorrow morning from what it was tonight.

This notion of continuous updating was a conceptual revolution. Paper had accustomed us to editions. An edition appears, then grows old. The telematic database introduced something else: knowledge as flow. I immediately understood the difference.

When information becomes a flow, the person who structures it is no longer simply a publisher. He builds a living system. He must know what enters it, where it comes from, how it changes, how the user retrieves it, and how long that information remains relevant.

The freight exchange was an ideal laboratory for understanding this new temporality because the value of the information disappeared extremely quickly. An offer of freight in Marseille for a truck heading back toward Lyon does not have the same value four hours later if the truck has already left. Available capacity today may be useless tomorrow.

Unlike case law or an auction result, which can acquire historical value, freight is perishable data.

We therefore had to begin thinking about the specific time of each piece of data. This discovery changed the way we processed data. Not all information ages in the same way. Some must be preserved forever. Others must be continuously replaced. Some retain value for decades. Others for only a few hours.

We would encounter this distinction in every database we built afterward. But here it was learned in the urgency of reality. The truck is waiting. Or rather, it is not waiting. If your system is too slow, the physical world moves on without you. The data then becomes useless before it has even been properly processed.

I believe this explains part of my later obsession with speed, machine power, response times, and scaling under load. Speed was not a technological fetish for me. It was part of meaning itself. Information that arrives after the decision it was supposed to enable is no longer truly operational information. It becomes an archive.

So we began working on machines, networks, and architectures capable of circulating this new matter. We had to learn everything no one teaches you when the professions themselves are still being born.

How to enter data quickly.

How to avoid duplicates.

How to identify a carrier.

How to qualify a route.

How to code a point of departure and a destination.

How to handle schedules.

How to connect freight with capacity.

How to prevent information that is already obsolete from continuing to be offered.

How to secure access.

How to ensure that a system does not become unusable precisely when many people begin using it.

Every technical problem actually concealed a structural question. This is where my thinking began to shift. I no longer saw simply information. I saw objects and the relationships among them. A carrier, a truck, a city, freight. Availability, a direction, a deadline, a rate.

The system had to understand these objects well enough to establish the right relationships. A database was therefore already no longer, for me, a vast digital warehouse. It was a geometry. We had to construct the space in which information could meet.

That is exactly what we would continue to do for fifteen years in worlds that appeared unrelated. Employment. Law. Legal notices. Government. Judicial sales. The Internet. Then Artprice.

Looking back, I see the same pattern recurring with almost disturbing regularity: identify a sector in which information is fragmented.

Understand the asymmetry produced by that fragmentation. Create a database. Normalize the information. Connect the actors directly. Let this new circulation change the market.

At the time, I did not theorize it that way. I worked. I tested. I watched where the system broke. I tried to understand why. This empirical dimension governed my entire method. I have always had an ability to project far ahead, but I do not believe in visions that remain pristine because they have never touched reality. Reality soils concepts. That is a good thing. It forces us to discover what we had not anticipated.

The freight exchange taught me that the circulation of information could have an immediate economic effect. It also taught me that when you reduce an asymmetry, you necessarily alter a relationship of power.

Regional freight exchanges, traditional intermediaries, and established networks had their usefulness, their habits, and their economics. Direct electronic matching could gradually make part of that way of operating less necessary.

I do not want to write retrospectively that I alone eliminated the old freight exchanges. That would be historically too simple. The broader evolution of technology and the market obviously contributed to their transformation. But our platform demonstrated concretely that electronic circulation could be faster, broader, and more efficient than certain traditional regional models.

And this is where a lesson begins that would follow me everywhere: a technological innovation does not encounter only technical problems. It encounters interests.

The person who benefits from scarce information does not always welcome with enthusiasm the person who makes it accessible. The person whose role is to connect two actors does not necessarily regard as progress a machine that begins performing part of that connection directly.

Yet I had no vocation to kill intermediaries. That vocabulary of economic warfare is too impoverished. The real question was always the same:

What value does the intermediary provide when information circulates differently?

If he provides expertise, advice, trust, responsibility, his role remains. If his only advantage is control of access, then technology will probably end up shifting that advantage.

I encountered exactly the same phenomenon with legal publishers. Then in judicial sales. Then in the Art Market. The freight exchange was one of its earliest manifestations. Above all, it taught me to look at data as economic matter. I mean matter. Because it has properties. It can be scarce or abundant. Fresh or obsolete. Verified or doubtful. Local or national. Private or public. Easy to reproduce but expensive to structure.

Raw data is almost never the finished product. The value lies in the work that gives it a place, a temporality, and a relationship with other data. This is the idea that would later be industrialized within Groupe Serveur.

But before Groupe Serveur, there had to be this almost primitive period when each problem forced us to invent a way of thinking. We were in the era of the first machines in the almost physical sense of the term. Computing was not yet transparent.

The machines had a presence. They occupied space. They generated heat. They broke down. Storage capacity was precious. Bandwidth was minuscule compared with what we have today. You had to know the machine in order to ask a great deal of it.

That physical proximity probably spared me a mistake I often see today: believing that the digital is immaterial. Nothing is immaterial.

Behind information that appears instantly on a screen are infrastructures, energy, protocols, people, programs, procedures. Dematerialization never eliminates matter. It relocates matter. And it creates a new matter: data.

That formulation may seem paradoxical. Yet it corresponds exactly to my experience. From those years onward, I began to work with information as I would later work with iron and steel. I looked for its structure. Its resistance. Its possible assemblies.

Poorly qualified data is like a badly machined metal part: it will contaminate the entire assembly. A badly constructed category produces errors on a large scale. A duplicate can alter a calculation. The wrong temporality can make information false.

So a precision is required that has nothing to do with the romantic image of the Internet as a space without constraints. It is an industry. I believe the great good fortune of that period was precisely that we had to learn the industry of data before the term became fashionable.

We did not yet have Artprice. We did not yet have Groupe Serveur in its future form. We did not yet have the Judicial Server. But a discipline was taking shape. Collect. Qualify. Normalize. Distribute. Update. Keep the machine running. And always observe what the new circulation of information produced in the physical world.

I was not interested in a computing demonstration that worked in front of three engineers. I wanted the truck to stop returning empty. Perhaps that sentence sums up everything. Technology had to produce a measurable human consequence. Fewer unnecessary miles. Better use of the working tool. Greater access to markets for the small carrier. A reduction in the advantage held by the person who alone knew the information.

This is where computing became, for me, a form of political action in the broadest sense: not party politics, but a change in the concrete distribution of possibilities. Giving someone information sometimes means giving him a power he did not have. That idea would never leave me.

Minitel would amplify it. Legal databases would amplify it further. Artprice would turn it into a global doctrine. The Internet would transform the scale. And artificial intelligence, forty years later, would raise exactly the same question with infinitely greater power:

Who can access what knowledge, with what depth, and at what speed?

But in 1985, all of this still fits within a much simpler scene. A truck is driving. Somewhere, freight is waiting for it. They do not meet because the information does not circulate. I look at that absurdity and understand that there is a problem here that the machine can solve.

This is the beginning. Not the beginning of my interest in technology. Something deeper. The beginning of a conviction that would organize everything that followed: when information begins to be structured and set in motion, it ceases to be mere knowledge. It becomes an active force. It becomes matter.