Soft Culture Machines

In the 1960s, the architectural avant-garde was excited about information technology. To many, computers and networks contained in their very structures the libertarian ideals of freedom, play, and non-hierarchical organization necessary for utopian social reform. Architecture, by adopting the forms and structures of information technology, seemed poised to reclaim its role as one of Western culture’s privileged instruments of social organization. Among the small number of built projects that emerged from these techno-utopian discourses, the Centre Beaubourg—later renamed the Centre Georges Pompidou—in Paris seemed the most promising. The brief for international competition held between 1970 and 1971 announced from the start that “the entire Centre has been inspired by an original perspective, that of constantly renewing information.” It went on to describe a complex institution whose metabolism was fueled by information exchange, deploying contemporary discourses in which information systems had the power to transgress boundaries between social groups, expert disciplines, and geographic spaces. The brief argued that the building would need to be permeable to information flows both to and from its environment, achieved by various means such as visual transparency, electronic screens, exhibitions, television broadcasts, publications, and a remotely accessible electronic library catalog would extend the Center's reach to the whole of France and outside. In short, the building was a vast information-processing machine, both in metaphor and in actual performance.

The winning entry by Renzo Piano, Richard Rogers, Gianni Franchini, and Ove Arup—a “live center of information” consisting largely of a steel and framework covered with information displays, antennae, and satellite dishes—was the only scheme that in jury’s view successfully translated the philosophy of the Centre into an architectural proposal, which the presentation punctuated using aphoristic diagrams of information networks with Beaubourg at their hub. Peter Rice later described their scheme as a “large, loose-fit frame where anything could happen [...] loosely based on Archigram, Price, Littlewood [...] an information machine with movable floors, like the Fun Palace.” At the time, Rogers said, “[w]e are after a rich, flexible matrix. And we hope that people will understand, learn how to use it. In all this we want to mix as many modern gadgets as possible. Electronic gadgetry is not just a fad. [...] We see communication, information, as vital.” So when the winning scheme was announced it seemed to many that the wait was over—that the plug-in architecture of the Archigram Group, the cybernetic schemes of Cedric Price, and the indeterminate megastructures of Yona Friedman might finally be realized, and in the historic center of Paris, no less.

But when Dennis Crompton, a member of the Archigram group, visited the newly opened “live center of information” in 1977, he had expected to see “an intensity of information which [was] not only real and useful, but also gratuitous and entertaining.” Instead, he responded, “well, where is it?” All that remained of the media screens and other information-age hardware was their bare supporting frame. Crompton wasn’t naive, and recognized that to ask “where is it?” was in many respects a mark of success because, as he put it, “information is only perceptible when it is in the process of being transmitted. It really is not a load of hardware hanging from the ceiling, flashing out of capsules or sprouting off pylons; these are only symbols, which we use on drawings, of a transient activity.” Despite this concession he concluded that the elimination of the apparatus shown in the competition scheme was a missed opportunity. Crompton’s equivocation was symptomatic of architecture’s changing relationship to information technology. Optimistic experiments in the socially transformative capacity of information technology so characteristic of the experimental architecture of the 1960s seemed by mid-1970s to have been entirely abandoned. Indeed, the Beaubourg competition entries by Friedman, Maymont, and Archigram seem mere traces of a period of experimentation long ago exhausted. Why did these experiments into the imagery and forms of the information age come to such an abrupt end at precisely the moment when the information society was starting to fully materialize? Beaubourg was born into a moment of transition in the emergence of the information society, one in which machine-age computer hardware was in the process of ceding to the intangibility and invisibility of software and networks, a condition demanding new modes of architectural response and new kinds of technical expertise.

The prepare the competition brief Pompidou put together a team of skilled technocrats to ensure that his project would not simply languish in the domain of demagoguery, as had André Malraux’s Musée du XXe siècle. Heading the planning team was François Lombard, an unknown 30-year-old, Berkeley-educated civil engineer working for the culture ministry's Direction d’architecture on programming methodologies for the planning of the Villes nouvelles around Paris. In late 1969, Lombard recruited Patrick O’Byrne, a friend who was working for a small engineering firm in Montréal with the imposing name l’Institut de recherches et de normalisations économiques et scientifiques, one of a group of firms supported between 1963 and 1976 by the Ford Foundation’s Educational Facilities Laboratory. The Ford research applied methods of programming introduced by American firms in the late 1950s to the development of flexible, open-plan public schools that paved the way for what became known as the Performance Concept. The practice of programming that was emerging in postwar research was much more than a list of functions and areas: it merged behavioral psychology and analytical methods of systems design into a set of techniques for controlling the all aspects of a building’s performance. Programming specified a total system, with which the avant-garde utopianists of the 1960s would have sympathized, yet it offered no corresponding totality of architectural solution because, as a species of systems engineering, it specified desired performance while leaving the possible set of solutions as open as possible. Lombard saw that the programming methods developed in the Ford research, until then limited to anonymous, industrialized building types, were applicable to Pompidou’s prestigious, one-off cultural center, and so he persuaded O’Byrne to return to Paris with him.

In preparing the brief, the programming team developed a formal method in which the institutional department was interpreted as a set of social functions, which Lombard later described, in characteristic fashion, as an “ensemble mathématique.” The first task of the programmer was the identification of the set of subsystems and relationships comprising this ensemble and the modes of exchange by which they cohered. The programmer identified a particular class of exchange (such as document reference, or viewing artwork), followed by the beneficiary of the exchange (researcher, museum visitor), the object of exchange (document, painting), the manner in which the exchange was to be transacted (consultation with computer terminal, walking through an exhibition), the necessary personnel to support the exchange (librarian, docent), and the spatial context of the exchange (reading room, gallery). The architectural program conceived in this way went far beyond the description of floor areas and the correct juxtaposition of functions to describe a complex system of interactions.

This approach challenged available modes of architectural representation. In response, Lombard’s team developed a diagramming method that offered a unified language for the representation of people, spaces, objects, documents, flows. It was difficult to imagine a mode of representation that did not bring with it a host of prior solutions, and in the brief, even seemingly banal terms like “library,” “civic center,” or even “city” were considered risky because they imposed prejudice and bias on positivist problem-solving. Lombard accordingly warned architects that they should avoid interpreting such architectural terms as “library” or “hall” as spatial types and instead treat them as scenarios constituted by precisely defined performance criteria. The design of a complex system could only be achieved “by developing an entirely new taxonomy of problem formulation.” At Beaubourg, programmers attacked this problem by defining a taxonomy of interactions such as desirable and undesirable views, flows of people, goods, and documents, sound isolation, etc. They then analyzed each activity, identifying the interfaces and flows by which it cohered. Once each activity was sufficiently well defined, the programming team described interfaces between one activity and another through a graphic system for mapping the flow of objects and information between activities.

The resulting diagrams borrowed heavily from graph theory and interaction matrices, which allowed system designers to visualize complex relationships between components. In a 1970 article in Architectural Design, Jean Cousin referred to the application of graph theory to architecture, first practiced by Christopher Alexander a decade earlier, as a “topological” as opposed to “geometrical” approach to spatial organization. In place of the imprecise humanist language of earlier utopians, the topological approach offered a theory of representation for a new positive science of design, free of appeals to intuition and perception. The connected graph described a finite set of nodes (represented by circles, which might symbolize any object in particular domain) and a finite set of connections between them (represented by lines, which could be labeled with a weight, qualitative or quantitative, and which could be made into arrows in order to indicate a direction for the relationship). The graph made apparent relationships that were much more difficult to identify in tabular representations. First-order relationships were obvious enough and could generally be identified by talking to users and other stakeholders, but second-order relationships, for example one in which a department might be related to another within an organization via a “middleman,” were much harder to discern since nobody on the ground had a sufficiently elevated viewpoint to apprehend them. At Beaubourg, where interdisciplinarity governed, second- and third-order relationships were crucial. Like industrialized systems building, this relational approach elevated the interface between subsystems to a privileged place. Where systems building located interface in the invisible contractual space between manufacturers of components, the program diagram located the interface at the shared boundaries between activities. The form and material that made up this boundary was a matter for the architect; indeed, the programming team went to great pains to make it clear that walls were only one way of reifying the boundary surfaces shown on the diagrams. The diagrams are therefore to be understood less as plans than as three-dimensional matrices since a connecting arrow between functional elements might be satisfied by a door, window, corridor, stair, or elevator.

The visual language of the program diagrams is their most striking feature. This language was only hinted at in the brief, but was fully developed during the initial design phases immediately following the competition. In drawing them the programming team consciously avoided rectilinear forms, both so as to not compete with the architects and to assert the autonomy of their disciplinary contribution. As a response to Banham’s challenge that architects of the 1960s run with technology or be left behind, these curvilinear forms were altogether different from the curves and bubbles of 1960s techno-utopianism: they were neither an index of a computational process nor the formal representation of the liberation of form (and by extension of material existence) brought about by information technology. Here, curvilinear forms are merely an outcome of a technique whose computational logic is intermediary and invisible, its relationship to architectural form indeterminate since both rectilinearity and curvilinearity are viable formal outcomes of the graph’s topological logic. Positioned halfway between topological and geometric approaches, the diagrams assert a disjunction between their particular visual language and the resulting architectural forms, and it is here that their intelligence lies.

Programming at Beaubourg operated in this gap between political intention and architectural form. Many of programming’s inherent paradoxes, such the need for top-down control within a system whose raison d’être was to engender freedom and accident, echoed those of the project’s political context, a paradox that Richard Rogers pointed out in his infamous critique of Programmation delivered during a working meeting with Lombard’s team. No amount of flexibility could mitigate the irredeemable openness of the system that the program was trying to control. By 1973, Horst Rittel and Melvin Webber had observed that complex systems—and particularly social and environmental systems of the type described by Lombard—were inherently undecidable. In the end, Programmation remained suspicious of activities at the edges of the closed system, such as the non-programmed events and happenings so close to the hearts of the architects. Fifty years earlier, the French management theorist Henri Fayol characterized the complex organization as an abstract legal entity based on rational rules and structures of authority: while Taylor had focused at level of the individual, Fayol was concerned primarily with reorganization at the top, a view of organizational reform echoed by Beaubourg’s sponsors. The programming team thus faced an aporia: a Fayolist approach to critical reorganization entailed the reinforcement of departmental identity and therefore of boundaries, while the new humanist religion of interdisciplinarity demanded their erasure. Techo-utopians of the 1960s had by-and-large sidestepped this problem, since their libertarian theories for the most part refused all organizational categories larger than the individual or family. But for administrators at Beaubourg, the “department” was integral to the critical discourse of organizational reform: at issue was not the elimination of departments and other broad organizational categories but rather in what ways might new relationships between them be put into play.

If Piano and Rogers’ winning scheme so perfectly translated the informational aims of the brief it did so with only one foot in the 1960s. Indeed, while visiting the newly opened building, both Price and the members of Archigram found it “too static and consistent.” Compared to the more conventional megastructures submitted by several other competitors the winning design was a minimalist black-box in which any activity could unfold. Unlike the utopian schemes of the 1960s, it neither mimicked in its organization the topologies of networks or computing machinery, nor treated architecture as an ever-receding backdrop to a new technological environment. Both of these earlier positions assumed a distinction between hardware and software, as was typical of postwar information paradigms in which information was stuff that flowed through a complex hardware of pipes and gates. Put simply, in Banham’s megastructure the static structural framework was its hardware while its everyday modes of inhabitation were the software. In this way, the winning scheme provided a loose-fit frame inside which Lombard, who was an influential member of the jury as head of its technical committee, could implement his methods.

Ultimately, programming challenged the megastructure model. In 1978, Simon Nora and Alain Minc’s L’informatisation de la société—a government report that became an unlikely bestseller in France—argued that such unhelpful dichotomies as “hard” and “soft” were bring made obsolete by what they called télématique—the convergence of computers and telecommunications technologies that started some time in the mid-1960s. According to Nora and Minc, télématique would soon, in their words, “alter the entire nervous system of social organization.” They observed that over the course of the preceding decade in France, American corporations (among which IBM had acquired a mythical status) had encroached on the traditional cultural roles of the state. The challenge, as they saw it, lay in finding a new role for government that would involve all possible levels of state action, from top-down decree, through looser practices of regulation, to complete laissez-faire withdrawal. Unlike the postwar computing machinery analogized in the architectural forms of the 1960s, both the hardware and software of telematic systems were virtual, the logic of their operation hidden in new kinds of invisible and diffuse architectures. Programming at Beaubourg was thus a response to the problem of building a live center of information at a moment when, in the words of Nora and Minc, “the seemingly natural boundary between hardware and software begins to fade away.”