US scholars often liken Minitel to a gated community or walled garden, but these comparisons offer a distorted picture of the Minitel platform.1 Likewise, attempting to demonstrate the benefits of openness by pitting the French model of dirigisme against an American model of venture capital overlooks the many features of Minitel that enabled experimentation and entrepreneurship.2 In fact, Minitel was not a completely closed system but rather a computational and economic platform that displayed many features of openness. The production of new Minitel services was not directed or micromanaged by the State. Rather, the development of Minitel followed a pattern similar in many respects to that of the commercialization of the World Wide Web: public subsidy programs created a platform and drove its exploitation by private industry. We will now discuss four ways in which the State supported openness on the Télétel platform, which in turn catalyzed a creative and dynamic private sector for videotex.
A Novel Hybrid Architecture
The development of a vibrant, independent Minitel enterprise during the 1980s depended on a platform architecture that combined elements of centralization and decentralization. Some degree of centralization was necessary for features such as the routing and addressing systems, micropayments system, and privacy protections. Decentralization was also essential, however, to open up the ecosystem to third-party developers. Although it did not quite follow the “end-to-end principle” characteristic of the present-day Internet, the design of Minitel nonetheless afforded freedom and autonomy to the nodes operating at the edges of the network. This combination of centralized and decentralized architectural features has so far been largely ignored—if not altogether denied—by the relevant literature on Télétel.3
The push for decentralization was a departure from the French tradition of centralization, both from a liberal ideological standpoint and as a practical means to develop industry. Policy experts Simon Nora and Alain Minc explicitly advocated the transition, and convinced the president of the republic to launch a mass-scale telematics project. They made their case in an influential report titled The Computerization of Society, which we excerpt at length:
Telematics … allows the decentralization or even the autonomy of basic units. Better still, it facilitates this decentralization by providing peripheral or isolated units with data from which heretofore only huge, centralized entities could benefit. … It reinforces the competitiveness of the small and mid-size business vis-à-vis the large enterprises. …
It would, however, be unrealistic to expect computerization alone to overturn the social structure and the hierarchy of power that governs it. The traditions and the cultural model we have inherited from our history favor centralization and administrative proliferation, hierarchic rigidity in big business, and the domination of small business by big business. Our traditions stand in the way of the initiative and adaptability required by a society based on communication and participation. Only a deliberate policy of social change can both solve the problems raised by telematics and utilize its potential. Such a policy implies a strategy based on the balance of powers and counter-powers and on the capacity of the government to favor development rather than impose it. …
Authorities will develop tools to make their policy work … by restricting their action and decentralizing when the needed changes require other groups to take the initiative. …
Anything that increases access to information facilitates dialogue on a more flexible and personal level, encourages increased participation and more individual responsibilities. … [W]ill we know how to enhance adaptability, freedom, and communication in such a way that every citizen and every group can be responsible for itself? …
The authorities will no longer be able to call upon the old methods and objectives, which are almost certain to fail. Preparation for the future implies inculcating a freedom that will cause even the most deep-rooted habits and ideologies to lose their validity. This requires an adult society that can enhance spontaneity, mobility, and imagination. … [I]t also requires a government that—while openly exercising its prerogatives—acknowledges that it can no longer be the only star of the social drama.”4
With the caveat that foreign server manufacturers benefited much more than French ones, the decentralization of Télétel hosts to the edges of the network achieved just what Nora and Minc foresaw. By allowing private industry to freely exploit the public platform, the telematics program enabled the government “to favor development rather than impose it.”5 In practice, the decentralization of Minitel fostered the development of private industries in hardware, software, and content. Hardware manufacturers were doubly bolstered by this arrangement. First, the State’s plan to provide millions of Minitel terminals to the general public created economies of scale in the production of terminal equipment, and second, the requirement that service providers maintain their own servers stimulated the demand for host computers.
In refusing to enter the content industry, and leaving it to the service providers to create, organize, and manage the servers and databases however they wished, the State fostered a new industrial culture characterized by spontaneity, mobility, imagination, and individual responsibility. This was the transformation that Nora and Minc envisioned. The hybrid network design, part centralized and closed, part decentralized and open, contradicts the reductive depictions of Minitel as a top-down, closed, controlled, and inflexible dirigiste system that suffered no nuances. The hybridity of the Télétel architecture is especially pronounced when compared to the architectures of contemporary videotex networks developed in other countries, both by public investment and private enterprise.
Télétel and Its Videotex Contemporaries
In the realm of public systems, the monopoly PTTs in Britain, Germany, and Switzerland, among others, all opted for fully centralized architectures. Videotex content was hosted on servers that were owned, controlled, and managed by the PTT itself. Space on the servers was leased to content providers, which would periodically upload pages on the state-run servers to be published on the state-run network.6 This arrangement provided no opportunity for a third party to develop novel software or services for the system.
In contrast to Minitel, the decision to centralize videotex elsewhere in Europe created demand for a few large mainframes as opposed to many smaller host servers.7 The reason for centralization was simple and political: both the hands of the British and German PTT were guided by mainframe maker IBM.8 Indeed, in the mid-1980s, sociologists Renate Mayntz and Volker Schneider wrote that the German system as a whole was “designed and implemented by IBM,” whose strategy was “to create [and host] one single big database.”9 From the point of view of a mainframe manufacturer, complete centralization was merely common sense. This had been the standard approach to institutional computerization throughout the postwar period.10 In fact, when the French Direction générale des télécommunications (DGT) designed its only online information service, the electronic phone book, it also centralized all the content on a few PTT-run servers. But Télétel was different. Instead of sponsoring one hardware manufacturer that would also maintain all the databases, the French government aimed at giving the means to multiple hardware manufacturers to compete and thrive.
The decentralization of Télétel unleashed private innovation in the development of content and services. “The most striking difference between Télétel and the other systems,” concluded Mayntz and Schneider, was “its complete decentralization” at the level of content providers. This difference in design had “far ranging implications for the flexibility of adapting to the users’ changing needs.”11 It is a lot easier for a service provider to constantly update its content when it has control over its servers than when it needs to request that the PTT implement changes on its behalf in a central IBM mainframe.
The autonomy of service providers is particularly important for enterprises that succeed or fail based on the timely distribution of information. The bottleneck on content creation hardwired into the architectures of centralized videotex systems certainly played a large role in their failure to develop a private industry.
The problem with the British, Swiss, and German videotex systems was centralization, not necessarily public control. This subtle distinction is illustrated by a comparison between Télétel and a failed attempt by France Telecom to implement a similar system through a private corporation in the San Francisco Bay Area in 1991 under the name 101 Online. There are many reasons why this experiment failed, but a major problem was that France Telecom abandoned the hybrid architecture of Télétel. Instead, the implementation of 101 Online more closely resembled Prestel, BTX, and US videotex systems such as CompuServe and the Source.12 Instead of distributing the servers to the edges of the networks as it had done in France, and thus fostering independent entrepreneurship, 101 Online centralized all content on a mainframe it controlled.
Figure 4.1 Comparison of 101 Online and Télétel network architectures. Julien Mailland, “101 Online: American Minitel Network and Lessons from Its Failure,” IEEE Annals of the History of Computing 38, no. 1 (January–March 2016): 6–22.
The design of the 101 Online network was very much operator centered. Third parties were welcome to create content for 101 Online, but they were afforded little autonomy. Indeed, every time that a service provider wished to make a change to their content, someone had to physically travel to the 101 Online office in downtown San Francisco and ask the 101 Online editor to make the changes in the mainframe on their behalf. This was not quite a conducive environment for spontaneity, mobility, imagination, and individual responsibility, to once again borrow from Nora and Minc. As decrypted in retrospect by a lucid Jean-Eudes Queffélec, 101 Online’s director of sales and marketing, “We did not create an ecosystem enabling anyone but us to make money.” In contrast, in France, the DGT was part of something bigger: a plan decided by the president of the republic himself to boost private enterprise—hence, the decentralization of the hosts to the edges of the network. This comparison is particularly relevant because the network designer is roughly the same in both the Télétel and 101 Online cases: the DGT in France, and a France Telecom affiliate in the United States.
Beyond the French servers operating at edges of the network, Télétel also provided gateways to data networks, public and private, emerging elsewhere in the world. This interconnection contradicts the portrayal of the system as an exclusively national network.13 The design of Télétel followed a clear goal of the Nora and Minc report insisting that French telematics policy should “improve [France’s] ability to compete [globally] and open new markets.”14 From an export perspective, Télétel was accessible to users in foreign countries using either a Minitel terminal or a PC outfitted with a software emulator through a dedicated French phone number. From an inbound perspective, the Télétel ecosystem was open to and interconnected with foreign public data networks, and open to foreign content. The short code to access the international gateway portal was 3619—from there, one would type in the region and then the specific code of the service. For example, one could access the nascent Internet from a French Minitel for 2.19 francs per minute by dialing 3619 ETATS - UNIS + USNET.15 Newspaper Libération even published a joke article giving a tutorial for finding the phone number for the Pizza Hut of Wichita Falls, Texas, through the 3619 Siriel international gateway service.16
Figure 4.2 Expanded diagram of the Télétel platform including gateways from Transpac to Telenet and Tymnet, two commercial public data networks based in the United States. Intelmatique, Télétel Videotex, February 1982. Source: Orange/DGCI.
Let us not overstate the simplicity of becoming a participant in the ecosystem, however. While Minitel was completely plug and play for the end user, it was somewhat more complex for service providers. Daniel Hannaby evokes his experience actually connecting his servers for his 3615 SM site to the Transpac network:
It was a complicated process. It took a few months because … we were not in this business before. … We went to see France Telecom [and] they had to explain [to] us how it worked, so we went to buy these boxes to connect Transpac in a certain [place] in Paris that was the only reseller of these boxes. And well, it took a few months.
On the economical [side]—the nice thing about France Telecom and Minitel was that they gave this [away] for free. [That] was fantastic because [they were] able to create a nice ecosystem of companies that were able to [make] money [from] it.
It was, for the companies, a bit complicated because they had to create these softwares, specialized software, to connect to Transpac. So, we had at one point black boxes that were in front of the computers to translate the X.25 coming from the lines into something that the computers would understand. But beside that, the architecture was really something—it was a choice between this and, of course, IP, TCP/IP, but it did a really good job. So, I can’t complain.17
All in all, the French content industry was highly successful domestically, and decentralization of the servers to the edges of the network was a key part of that success. The benefits were not limited to the service providers, though. Next, we examine how France Telecom leveraged the hybridity of the architecture to promote the development of private hardware and software industries.
Catalyzing the Development of a Domestic Computing Industry
Nora and Minc chose to portray IBM as the prototypical US villain. Decentralizing the Télétel servers to the edges of Transpac and giving operators “free rein” to assemble their own host systems was meant to catalyze the development of a vibrant French computing industry.18 The plan worked, insofar as industry development is concerned. But while domestic enterprises flourished in the development of software and consulting services, foreign manufacturers like IBM quickly dominated the server industry.
Challenges Facing the Domestic Hardware Industry
In 1989, Bull was the flagship of French hardware manufacturing. It was the leading manufacturer in Europe and the seventh-ranked manufacturer worldwide.19 And yet at the same moment, Bull was conspicuously absent from the Minitel scene. As summarized in 1988 by Marie Marchand, a France Telecom researcher,
Steps were taken to promote what French industry had to offer by using Honeywell Bull’s Mini 6 computer, Intertechnique’s Réalités 2,000 and 5,000 models, and Thomson’s Microméga. France’s Agency for Data Processing, the ADI, had taken a similar tack by subsidizing the first videotex operations using French equipment. But despite such backing, French manufacturers would not be able to hold on to the host systems market, from which they have all but disappeared in 1987.
From the outset foreign manufacturers carved out a share of the market. IBM, DEC, Hewlett-Packard, ATT/Olivetti, and McDonnell-Douglas supported systems and software houses in their approach to videotex. They knew how to adapt to market trends; they had a range of marketing techniques and product lines that meshed with the needs of service providers. When in the early going, high-capacity host computers were called for, they had just what the doctor ordered. When around 1982–83, microcomputers were needed, they were able to meet the demand, and in 1984 they supplied the megamachines necessary for seeing through a mass-market strategy.20
Beyond marketing, another set of factors was driving the adoption of US computers: the preferences of the programmers and system administrators tasked with building Minitel services. During the 1980s, a generation of software engineers and system administrators were trained on campus computer systems running Unix, and they carried a preference for this “open” operating system into their professional lives.21 As US computers tended to accommodate more operating systems and programming languages than their French counterparts, US manufacturers came to dominate the field of Télétel hardware. Hannaby, for example, recalls choosing AT&T servers to host his site specifically because the French-made Bull systems did not handle Unix: “We had to go onto something that was using Unix because our system was Unix based. We found that AT&T was using these 3B2 and 3B5 computers, Unix based, and they were sold by Olivetti [the Italian company] at that time.”22 Taranis, another consulting and hosting firm, similarly relied on Unix-compatible minicomputers from Texas Instruments—namely, the BS 300, 600, and 800 models—for its infrastructure.23 In contrast, Marchand notes, “Other host systems remained tethered to the [Bull] Mini 6 or equivalent hardware [that was not Unix compatible] and dropped out of the consumer videotex market one by one.”24 By allowing service providers to build their own host systems, the State may have stimulated the growth of third-party services, but it was up to French hardware manufacturers like Bull to exploit this opportunity by meeting the demand for Unix among technical professionals.
Many Minitel services were built by consulting firms specializing in videotex rather than by in-house developers. As a result, the preferences of busy consultants had an outsized impact on the market for Minitel hardware. Jean-Louis Fourtanier was one such consultant who preferred Unix “both in terms of quantity and quality.”25 Fourtanier’s consulting and hosting business, CTL, was used by many of the largest Minitel sites, including Crac, Aline, Anabelle, the public television station FR3, and eventually all the properties of Le Nouvel Observateur, one of the top promoters of services.26 Since Fourtanier swore by Unix, he “invested in hardware that operated under Unix”—in this case, Hewlett-Packard.27 Once a machine gained a reputation as a reliable platform for a Minitel service, it was more likely to be used in the future. Hannaby remembers noticing that Olivetti started selling many Unix-compatible AT&T 3B15 minicomputers after he used one to build a Minitel server.28 Large institutions were even more likely to stick with familiar brands. When Crédit Foncier de France, the first French bank to offer its services over Minitel, had to choose a system, it went with something “muscle-bound … an IBM 3081 CPU with two front-end communications units and 400 access points.”29 Apparently no one was ever fired in France for buying IBM, either.
Opportunities for the Domestic Software Industry
In contrast to their hardware industry counterparts, independent French companies in the realm of software and database consulting as well as those (often the same) that offered hosting services were able to reap the benefits of the openness features of the Télétel platform. In addition to the traditional powerhouses such as SAGEM and Capgemini, a plethora of midsize and small software consulting houses thrived in that market. A ten-page France Telecom directory from April 1986 listed over 230 such firms.30 By April 1990, that directory’s pages would number seventy-seven.31
The hybridity of the Minitel platform, with decentralized hosts operated, designed, and maintained by private parties, exhibited a kind of openness that was not present in the other videotex systems of the time. By operating as an open platform, Minitel supported the development of dynamic and competitive hardware and software industries. Let us now observe how the State, far from being dirigiste, applied a venture capitalist approach to catalyze what would become a booming and wealthy private content industry.
Venture Capitalism à la Française
With Minitel, the State built a generative platform for developers through the combination of the kiosque billing system and the distribution of free terminal equipment. By threatening to discontinue the phone book and provide an electronic phone book service instead, the State forced free terminals onto the populace and the pump was primed on the user side of the market. The resulting base of potential consumers was enough to attract service providers to the platform. But the State also spared no expense in priming the service provider side of the market as well. The goal was to catalyze the development of a domestic content industry to rival the global hegemony of the US media industries. The DGT supported potential service providers in four ways: intensive, large-scale testing; openly sharing design research and expertise; sponsoring professional education in the realm of videotex; and subsidizing select services in ways that resembled the public–private partnerships that led to the rise of Silicon Valley and subsequent dot-com industry.32 All in all, the State acted much like a venture capitalist, offering financial support and organizational guidance to budding Minitel service providers.
The Circulation of State-Sponsored Research
Following the executive decision in 1978 to launch the telematics program, the DGT initiated two live experiments. First, in the region of Ille-et-Vilaine, it provided twenty-five hundred users with various experimental terminals hooked up to an online phone directory. Second, in the cities of Vélizy, Versailles, and Val de Bièvre, in the suburbs of Paris, it recruited another twenty-five hundred users who would connect to eighty services providing a total of thirty thousand pages of videotex content. This project was named Télétel 3V, or T3V, after the three cities.33 The two experiments would provide fruitful lessons in computer ergonomics and human–computer interaction. For example, some of the keyboards were laid out in alphabetic order to test the hypothesis that users unfamiliar with typing would find it easier to input commands. As it turned out, new typists were agnostic to the layout, whereas experienced users found it more difficult to use, as they were accustomed to an AZERTY layout.34
The experiments in graphical user interface design were published in the form of a book, Communiquer par Télétel, which targeted potential service providers and gave them advice ranging from organizing the logical tree structures of sites to best practices in color coding.35 It is worth noting that where the DGT had single-handedly imposed the choice of the X.25 network protocol, in contrast, it never required the service providers to use a particular structure or interface design at the application layer. Unlike, say, the 101 Online venture, service providers were free to organize their services and content however they liked. But France Telecom offered its recommendations as a means of helping the new, independent industry succeed. France Telecom then produced a number of free brochures on a variety of topics, including becoming a service provider, designing a messaging system, and building an interface as well as general information and statistics regarding the T3V experiment.
Over the course of the Minitel years, France Telecom published a large number of sophisticated brochures, printed on appealing glossy paper, covering a range of topics from recommendations on graphical user interface standardization to full technical specifications for every piece of the platform: terminals, PAVIs, protocols, and so on. France Telecom also published a quarterly newsletter, La Lettre de Télétel (Télétel letter), informing industry participants of the latest technical improvements to the platform and business experiments by service providers. This large corpus of reference material and market research was available free of charge to support third-party service operators.
Figure 4.3 Technical documentation for designers of videotex services. Direction générale des télécommunications, Spécifications vidéotex de visualisation et de codage (Paris, France: Ministère des Postes et des Télécommunications, circa 1980). Source: Orange/DGCI.
Figure 4.4 France Telecom documentation included detailed examples demonstrating how to design videotex services that would be easy for users to navigate. In this picture: the tree structure and use of function keys suggested for Télétel sites. Ministère des PTT, Recommandations aux partenaires Télétel, (Paris: Ministère des PTT, 1986), 11. Source: Authors’ collection, donated by Jean-Luc Beraudo De Pralormo.
The combination of centralization and decentralization in Minitel enabled France Telecom to develop and circulate expertise regarding the design and maintenance of videotex operations at a pace that was not possible elsewhere. First, it was easy to gather information about services and their users on a large scale via the PAVI. Second, because the network operator was not competing with the service providers, it openly shared the results of its analyses. This openness and cooperation stood in contrast to the contemporary situation elsewhere on both counts. For instance, the myriad of small-scale videotex services in the United States did not benefit from the experience and know-how that France Telecom was able to develop through lengthy tax-funded experiments and mass-scale adoption by users. Even when they did gather information on a large scale, companies like the Source or CompuServe had no incentive to share it with their competitors. In this respect, then, the centralized, state-funded nature of the Minitel platform enabled an element of openness with regard to design expertise and directly contributed to the blossoming of private, for-profit operators. And because every Télétel connection generated revenue for the DGT, the State stood to benefit from the success of any service using the platform. Venture capitalism, à la française.
In addition to openly publishing research, recommendations, and statistics about its experiments, the State supported a number of initiatives to directly train and guide would-be Minitel entrepreneurs. France Telecom created a school for Minitel developers called the Centre national de formation aux usages de la télématique (National Training Center for the Use of Telematics). It provided training that closely resembled what an executive graduate program looks like today, including courses in the fundamentals of network economics, and practical training in business plan development, legal aspects of telecommunications networks, and applied electrical engineering and computer science. It located the school in the Cesson-Sévigné industrial zone that was already home to both the CCETT and DGT research centers as well as major industrial players such as Thomson and Transpac.36 The DGT itself occasionally acted as an ad hoc incubator. During trade shows, for example, the DGT hosted small service providers in its exhibition area, and supplied all participating service providers with hardware and other equipment, free of charge.37
Public–Private Partnerships in France and the United States
Beyond the training and free research, a number of Minitel service providers received a direct public subsidy. Financing for these entrepreneurs was supplied by France Telecom itself; the Caisse des Dépôts et Consignations, the French development bank that invests in private ventures “in support of the public policies implemented by the State”; and public banks such as the Crédit Lyonnais.38 These monetary infusions further primed the pump on the service providers’ side of the market, creating a positive feedback loop on the users’ side: more services attracted more users. This form of State intervention did not involve central control over the creation of content, however, nor did it otherwise bear the mark of dirigiste micromanagement. Quite to the contrary, public subsidies for new service providers supported the development of an unbridled private industry, from mainstream information applications to the frequently edgy messageries (chat rooms).39
From the standard Silicon Valley perspective, it is tempting to assume that because third-party service providers were subsidized by the French State, then the overall Minitel ecosystem must have been State-controlled or strangled by bureaucracy. On closer inspection, though, the public–private partnership driving the development of Minitel looks very much like the public–private partnerships that led to the commercial exploitation of the packet-switched Internet in the United States.40 Indeed, the rhetoric surrounding US information infrastructure projects of the 1980s and 1990s mirrors many of the ideas developed by Nora and Minc. When then US senator Al Gore explained the significance of the High-Performance Computing and Communication Act in 1991, he emphasized the importance of information infrastructure for the future of US hegemony: “The nation which most completely assimilates high-performance computing into its economy will very likely emerge as the dominant intellectual, economic, and technological force in the next century.”41 This rhetoric justified the allocation of billions of federal dollars to the technology sector and the creation of a venture capitalist investment firm by the Central Intelligence Agency that financed over 180 private firms.42 Evan Sutherland, vice president of Sun Microsystems Laboratories, later acknowledged that the thriving tech industries were facilitated by “decades of consistent federal investment in information technology research,” and Marc Andreessen, founder of Netscape, bluntly stated that if development had been left to the private sector, “it wouldn’t have happened.”43
Certainly, the DGT was more hands-on than the Advanced Research Projects Agency (ARPA) or the National Science Fountation (NSF), as it required Transpac to use a network protocol that it had developed itself rather than leaving the decision to either the preferences of researchers or forces of market competition. But extending the meaning of this intervention to say that the whole Minitel ecosystem was subject to dirigiste forces, or worse, was “doomed by … excessive culturalist policies,” as one scholar put it, is inaccurate.44 Likewise, using the fact that the taxpayer paid to build the Transpac network to distinguish the US and French models is a fallacy; the US taxpayer subsidized several networking projects including ARPANET and NSFNET. Indeed, the hybrid architecture of Télétel requires a more nuanced description of State intervention than simply calling it dirigiste. While intervening à la française at certain levels, for example, by picking the network protocol and centralizing certain aspects of the platform, the State also very much acted like a venture capitalist, investing in emerging service providers and thereby enabling private sector innovation at the edges. And in spite of its investment, the State did not guide the hands of private developers but instead left them free to create and promote their own content and services (to wit, the rise of the pink messageries).
Walled Gardens, Gated Communities, and Shades of Openness
Minitel has been dubbed a gated community, and compared to “walled garden” services like CompuServe and America Online.45 While there certainly are elements of closedness in the platform—chiefly, the State gateways and censorship regime—the description of Minitel as “open” or “closed” requires some clarification. Just as the technical architecture of Télétel was not strictly end-to-end but still fostered innovation from the edges, content circulated on Télétel was censored and open at the same time. In spite of the censorship regime, Minitel service providers certainly enjoyed greater freedom than their US videotex counterparts—perhaps even more so than later entrepreneurs working within twenty-first-century walled gardens such as the Apple App Store.46 The comparison between Télétel and US-based walled gardens makes clear that the mere existence of a censorship regime did not mean that Minitel was in thrall to a totalizing form of authoritarian state control, as is often believed. Indeed, the practical experience of content regulation on Minitel underscores the need to approach the theoretical concept of “openness” with nuance.
There is no technical definition of walled garden (also known as gated community) in the realm of information networks.47 Yet in the United States, walled garden is a colloquialism that refers to an information system in which the service provider exercises final authority over the content and services accessible to users. In that sense, a walled garden is a censorship regime (where the prior approval of the network operator needs to be secured before information is published) as well as a regime where a central operator can kick content out of the gated community ex post facto, at will. This filtering may be based on legal grounds (e.g., the removal of unlawful content), but it may also be based on any number of criteria such as quality or business grounds elaborated in a “terms of service” agreement. Facebook’s behavior in this respect is a good illustration of a platform that, albeit seemingly open at the user level, exhibits features of gated communities.48
The walled gardens model is also associated with centralized systems such as the videotex networks emerging outside France at the same time as Minitel, such as Prestel, BTX, and 101 Online. In the 1980s, CompuServe centralized the hosting of services that it had picked as part of business negotiations with external content providers, and such negotiation often led to certain content being removed or replaced. In 1983, for example, the online version of the twenty-volume World Book Encyclopedia was simply removed from the CompuServe walled garden and replaced with the Grolier electronic encyclopedia.49 Overnight, CompuServe users lost access to the World Book content. Likewise, the Source (the online arm of Reader’s Digest) hosted all its content in a central facility in McLean, Virginia. In March 1983, the Source announced a new policy for curating the content on its platform: “THE SOURCE notes that new products are receiving close scrutiny based on likely long-term usage rates, as opposed to ‘attention-getter qualities.’ This relatively new policy has led to decisions to drop a number of previously anticipated new products, such as an electronic encyclopedia.”50 By taking an active editorial role licensing and soliciting content and services, early online services like the Source acted more like newspapers or radio stations than platforms.
The walled garden model and associated censorship regimes are not unique to early online systems. Since opening the App Store in 2008, the Apple iOS ecosystem has exhibited a hybridity similar to Minitel’s. Third-party developers build applications for Apple devices that can only be accessed through a centralized platform. Nearly all third-party apps are sold through the App Store where Apple acts as an intermediary, handling payments and taking a percentage of all sales. As part of the platform operation, Apple vets all potential applications and reserves the right to unilaterally remove any app or developer.
While the App Store has created a new commercial opportunity for small-scale software enterprises, Apple’s autocratic control of the platform routinely leads to conflicts with developers. In 2010, Apple removed the WikiLeaks app from its App Store, making it more difficult for iPhone users to access content published by WikiLeaks. The decision by Apple to block access to WikiLeaks through its platform was not motivated by any legal mandate—WikiLeaks was protected by the First Amendment—but rather by public relations considerations. Vice President Joe Biden had called WikiLeaks founder Julian Assange “more like a high-tech terrorist than the Pentagon Papers,” and Apple did not wish to be associated with that image.51 Unlike Minitel, with its implied commitment to the French public interest, privately run platforms like CompuServe and the Apple Store are governed by an opaque, centralized form of authority—gated communities à l’Américaine.
To understand Minitel as a kind of a walled garden, it is useful to directly compare Télétel with US walled gardens along four dimensions: the motivation for censorship, implementation of censorship, recourse available to censored service providers, and censorship of individual users.
First, the only element of the Télétel platform common to a walled garden is the censorship regime. In practice as well as policy, prior authorization from the executive branch of government (and later France Telecom) was required before a service could be connected to Transpac. The prior authorization regime, however, was not part of the original design of Télétel and was imposed solely because of pressure from the existing media industries, which were concerned (rightfully, it turns out) by the competition from electronic publishers. Originally, the DGT wanted to apply common carrier principles to the network and implement the policy we now refer to as “net neutrality”—that is, to not discriminate against packets based on the identity of the speaker or the content at stake. Until the election of Mitterrand in May 1981, internal notes from the prime minister’s office focused on common carrier and content neutrality principles as cornerstones of the nascent network’s organization. F. Froment-Meurice, of the government general secretariat, asserted in 1980 that “we do not find any legal basis under which the administration could impose to TÉLÉTEL’s ‘service providers’ a ‘code of conduct’ containing clauses relative to the content of data transmitted, to a right of reply, etc.”52 Similarly, one month before the election of Mitterrand, Pierre Huet, a State Council justice and the head of the ad hoc Telematics Commission, stated, “The administration must also respect the principle of equality of users.”53 In June 1981, debriefing the new, Mitterrand-appointed PTT minister, Huet reasserted and detailed the implications of this principle, noting, “The PTT Code applies: it grants the telecommunications administration a monopoly of transmission of mail correspondence, and in turn the administration is bound by public service obligations (freedom of access, mail secrecy).”54 The analogy used by the justice rested on the traditional-mail common carrier principle: a monopoly is granted to the carrier, which in turn cannot look into the content being transmitted over the network, and cannot discriminate based on content or user group.55 Ultimately, the pressure of the press lobby was such that the executive branch, and then Parliament in 1982, imposed a censorship system in order to reach a truce with the lobby and be able to move on with Télétel.56 The only walled garden aspect of Télétel, therefore, was not the result of a desire to control the network on the part of the State, as is often maintained. Quite to the contrary, the censorship regime was the result of lobbying by one private industry sector trying to prevent the emergence of competition from another, emerging private industry sector.
Second, because of the requirement of due process to which the State is bound, even with the censorship system in place, access to the Minitel platform was much more open than in the typical walled garden scenario. In the United States, because the censorship takes place in the private realm, there is no legal recourse for service providers that are shut out of a walled garden. If Apple does not want an app to be part of the iOS ecosystem, the developer of the app cannot sue Apple. In contrast, because Télétel was a public platform managed by the State, French service providers were afforded due process. Only illegal content could be rejected, not content that the platform operator did not like. Thus, the public management opened the platform up in ways that have not been realized elsewhere. In contrast, apps like WikiLeaks get rejected by Apple based on business considerations, even though their content is protected by the First Amendment.57 The administrative process through which the censorship system was administered, although complicated, was also transparent. If a service provider was not satisfied by the outcome of the administrative process, it could sue the State in court. So while the platform was not fully open and chilling effects certainly existed, due process principles ensured that legal content could eventually make it through—a level of openness that has never existed on the prototypical walled gardens built in the United States.
Third, once the service provider was approved, it could not get kicked out of the ecosystem lest the content turn out to be illegal. This again is an element of fairness—not to mention openness—that contrasts with US gated communities.
Finally, in the context of the messageries, once the provider of a messageries service was approved by the DGT, all the users of that system became de facto content producers. Publishing, in this sense, included all the messages posted in the chat room from thousands of individuals connecting via their Minitel terminals. Individual users were not subject to State censorship, and as a result, Minitel became an important platform for fringe political and cultural activity.58 For instance, during the massive antigovernment student demonstrations of 1986, Minitel became “a peerless tool for information and communication” used to organize the protests.59
As these examples demonstrate, the Télétel platform exhibited various characteristics of openness that exceeded walled gardens such as CompuServe or Apple. The contrast between these French and US illustrations provides yet another counternarrative to the portrayal of Minitel as a dirigiste, closed, parochial, and centrally controlled platform. In practice, Minitel was open in ways that should push us to think with greater nuance about the “shades of openness” in information and communication platforms.
Conclusion
Cesson-Sévigné may not be as sunny as Silicon Valley, but the public–private partnership that undergirds the French telecom capital is not all that different from the military-industrial complex that initially spurred the growth of the US tech capital. Over the years, many scholars have come to see Minitel as an icon of centralized State planning, the epitome of dirigisme, and the ideological opposite of US entrepreneurialism. Not only is this portrayal inaccurate—dirigisme implies a certain micromanagement that is absent from the Minitel case—but it also overlooks important decentralized characteristics of the Minitel architecture that are useful for thinking with greater nuance about the openness of the systems we rely on today. Indeed, while some aspects of Minitel were kept under centralized State control, the provision of end user services was left to independent enterprises positioned at the edges of the network. The State further stimulated the growth of this private sector by providing billing and account services, high-quality technical documentation, training in the design and management of information resources, and detailed research on large-scale public videotex experiments. By combining elements of centralization and decentralization, the architecture of Minitel invited mass-scale participation as well as encouraged widespread entrepreneurship and experimentation.
The plan to create Minitel was motivated in part by a vision of telematics put forth in 1978 by Nora and Minc. The two researchers hoped that by decentralizing the provision of services, French telematics might enable a kind of entrepreneurial “freedom” to unsettle long-standing political traditions in the country, causing “even the most deep-rooted habits and ideologies to lose their validity.”60 On reflection, it seems that while a certain amount of freedom may have been realized at the edges of the network, the system could not wholly undermine the French tradition of centralization.
In 1994, Gérard Théry was tasked by the French prime minister with studying ways in which France could benefit from the development of the international information superhighway. In his report, Théry opined that the Internet could not successfully develop in France without State sponsorship because it lacked many of the centralized features of Minitel—namely, a comprehensive directory of users along with a reliable system for billing and payment.61 Instead, Théry envisioned that there would be two Internets: “one Franco-German internet [driven by the French and German governments], and another one [the legacy of ARPANET and NSFNET].”62 Théry’s failure to foresee the commercial success of the Internet suggests that he did not believe that a network ecosystem could be viable unless it was designed and managed by a top-down state-like entity. Rather than challenge the political tradition of centralization, the Minitel experience seemed to reinforce a belief that centralization was a necessary precondition for the success of the Internet in France.
Freedom—in the Nora and Minc sense of the term—was felt most strongly at the edges of the Télétel network, among those users and service providers that built and occupied the services that gave life to Minitel. The hybrid architecture offered a generative platform for these private stakeholders to experiment and play with the possibilities of widespread telematics. In the next chapter, we examine the creativity of the private entrepreneurs who introduced online services, software, and hardware that extended Minitel in ways that the designers of the network could not have envisioned.
