The Network France Built First

The world it was born into

In the early 1970s France had one of the weaker telephone networks in the developed world. Getting a line could take months. Within a decade, a vast programme of investment had transformed it, and by the late 1970s the state telecommunications administration, the DGT, had a modern network and the ambition to use it for more than voice.

In 1978 two senior officials, Simon Nora and Alain Minc, published a report for the president on the “computerisation of society”. It argued that the merging of computers and telecommunications, which they called télématique, would transform the economy and that France needed a national strategy to avoid dependence on American firms. The report became a best-seller and a guiding document.

Several countries were then building videotex: systems that displayed pages of text and simple graphics on a screen, delivered over the telephone network. Britain had Prestel, Germany Bildschirmtext. Most struggled, because few households would buy a terminal before there were services, and few providers would build services before there were households.

The bet

France solved that problem with a single, expensive decision. The DGT would give terminals away. The justification was practical: the paper telephone directory was costly to print and update, and an electronic directory would replace it. Trials began in 1980, the electronic directory launched in Brittany in 1983, and distribution spread nationwide. The terminal, a small beige box with a screen and a fold-down keyboard, was called the Minitel.

The second decision, in 1984, created the economy. The kiosque system let users reach services through a code, with charges added to their phone bill by the minute. The DGT kept a share and passed the rest to the service provider. Anyone could set up a service without building a payment system. The bet was that free access and simple payment, together, would produce a national information market.

Why it made sense

It worked, and spectacularly. Thousands of services appeared: train and airline reservations, banking, news, weather, mail order, games and, notoriously, chat services, some of them erotic, which drew very heavy use. By common estimates, Minitel peaked at around nine million terminals and some twenty-five million users, in a country of under sixty million people, with tens of thousands of services. In the early 1990s, when most Americans had never been online, a French household could book a train, check a bank balance and send messages from the kitchen table.

The design choices made sense for France at the time. A single operator meant reliability and simple governance. Standard terminals meant no compatibility problems. The kiosque meant providers were paid, which solved a problem the web still struggles with. In its own terms the system was a success.

How it lost

The internet grew differently. It had no central operator, no standard terminal and no built-in payments. It ran on general-purpose personal computers that could do anything, rather than on terminals that could only display pages. In the early 1990s it was slower, harder to use and more expensive to set up than Minitel. For the things French users already did online, Minitel was better.

That was the trap. Each year in the mid-1990s, the decision to stay with Minitel was easy to defend. It made money, for France Télécom and for thousands of businesses. Users liked it. The internet’s advantages lay in what it might become, not in what it did that year. Meanwhile the web was compounding: every new site and user made it more valuable, and Minitel, which could not run new kinds of software, could not follow.

By the late 1990s French internet adoption lagged comparable countries. In a speech in August 1997 the prime minister, Lionel Jospin, acknowledged that Minitel, a technological success, risked holding back the development of new applications. Minitel declined for another fifteen years, kept alive by loyal users and services that never migrated, until France Télécom switched it off on 30 June 2012.

Explaining both the same way

It is tempting to say the internet won because it was open and Minitel lost because it was state-planned and closed. That captures something, but a symmetrical account needs more.

Minitel was designed for a world in which the terminal was the expensive part, the network was the scarce resource and payment was the hard problem. In that world, central design, free terminals and kiosque billing were excellent choices. The internet was designed for a world of cheap, general-purpose computers on which anyone could experiment. When personal computers became cheap and powerful, the conditions Minitel was optimised for disappeared. Its virtues of standardisation, central control and a single billing relationship became constraints.

The same reasoning explains why France was early and then late. The decisions that made it early were the ones that made it late. That is what a success that becomes a trap means.

What the loss teaches

Minitel’s deepest lesson concerns loss aversion at scale. A profitable, working system creates a constituency of operators, providers and users who all stand to lose from change. No single decision to delay is wrong. Together they turn a lead into a lag.

It also shows that the losing system can be right about something the winner gets wrong. Minitel’s kiosque was a working model for paying online creators, decades before app stores, and it avoided the advertising economy that came to dominate the open web. The internet won the platform contest. It did not win every argument.

Sources and limitations

Figures for Minitel’s peak terminals, users and services vary by source and year; the numbers here are common estimates, given as approximate. The history draws on Mailland and Driscoll’s scholarly account, the Nora–Minc report and contemporary coverage of the shutdown. Jospin’s 1997 remark is paraphrased rather than quoted. How much Minitel slowed French internet adoption is contested, since other factors, including the pricing of internet access, also mattered.