Pneumatic Structures, Cybernetics and Ecology: Toward Ecostructures for Habitation of People and Other Lively Systems


This is the transcript of a talk that Blair Hamilton, one of the people affiliated with ET&T gave an architecture conference (where Nicholas Negroponte was also presented). It sheds some light on how the Johnson Quarry community brought cybernetic vocabulary to the problems of architecture and ecology.

Pneumatic structures have a technology that Is important in its potential for development of a responsive architecture. As with each advance in the field of structures, the first use of pneumatic structures has been to imitate previous structures. Now that we are beginning to become skilled at using lightweight structural materials, we can pursue their unique possibilities.

Present building materials and their technologies fill our environment with architectural forms that tend naturally toward the freezing of time, the destruction of their environment, and the de-humanization of their inhabitants. This architecture encourages the designer to shut out and screen out the environment so that he can reasonably create a controllable sub-environment to inhabit. We have learned to value this isolation from our surroundings, unaware that it, like most of the rest of our technology, does not include any sense of ecology.[1] Architecture as shelter has been built of hard materials, dependent on their rigidity or insularity (structural, thermal, acoustical, visual) to separate our living spaces from their surrounding environment. I visit too many modern buildings where I am isolated in such a conditioned space that I am completely unaware of what is going on outside the structure. Like the automobile driver who doesn’t have to breathe the fumes he leaves behind outside his shell, I am unaware of how my building, or my activities inside it, are affecting the outside environment. This may seem the extreme case, but in most of our architecture almost all the information we perceive inside about what is going on outside is what we see through a few windows, hear because the “walls are too thin,” or feel because the “insulation isn’t good enough.” You see what I mean by isolation being highly valued—and worthy of cost. The loss of any sense of continuousness with the environment is just beginning to be valued differently.

This sounds like general theorizing, and I’m going to do more of it, but I’m also going to talk about our work at finding ways to create the innate technology of soft materials by real buildings. If we want to create pneumatic structures that will be valued by people, our work cannot be just in our heads. Words, drawings, and models are not sufficient when our object is something experiential. We must build our ideas. We must make our fantasies something we can feel. We must build them before we know if they will work. We have to interact with our prototypes by inhabiting them, evaluating that interaction, and re-designing to solve problems and explore further. Our prime design consideration is no longer material, structure, or product, but the total experiential process. Our language about it must either be a metalanguage or we must start doing it and talk about what we are doing.

The need for doing is reflected in the pneumatic structure project at Antioch College, in Columbia, Maryland. An environmental design program was started so that students and faculty can be primarily responsible for developing their own facility. After building several smaller prototypes, we are now constructing a 30,000 square-foot structure to house all the administrative and academic facilities required by the programs in Columbia, one of several units in the Antioch system.

The facilities problems faced by not only most of our schools, but by most social institutions caught in an environment of accelerating change, has been extreme at Antioch-Columbia. Anticipating that the campus might choose to relocate as the focus of academic programs shifted, heavy investment in a conventional campus facility was ruled out. The rental space available proved to be expensive, cramped, and inappropriately designed. This campus of the college has, in its three years of existence, been in a continual process of radical change in institutional structure, programs, and related facility needs. The inability to anticipate further inevitable and dramatic changes compounds this problem and exemplifies how it is becoming increasingly difficult to design conventional buildings which will be useful for any reasonable amount of time.[2] A new kind of capability—the ability to accommodate unanticipated requirements—is called for now, and will be required for the future if we want the changes of our personal and social growth to be encouraged.

A precondition to this responsiveness is a process of informal convergence upon an architectural embodiment which includes the user as a central figure in the design, building, and later modification of his architecture in a real-time and on-line mode.[3],[4] Recognizing the importance of this at Antioch, the Environmental Design Program has as its major focus the development of this kind of process. Many of the users have no prior training in design, but this has the advantage of easy acceptance of fresh approaches. Work on the building goes more slowly, and there is much frustration of well-intended effort, but there is also an incredible amount of education and a high level of involvement with their architecture in a process of mutual evolution. The process is learned by involvement rather than academic explaining. The ease of learning about, building, and studying pneumatic architecture makes it very appropriate for this process. Heavy construction equipment is avoidable.

The ability of pneumatic structures to enclose large open spaces provides a further level of adaptability. Lightweight materials and portability now become reasonable for interior architecture, since the need for load bearing and weather resistance has been eliminated. At Antioch we have begun developing an interior system of furnishing, spatial definition, and services which is compatible with the potential portability and softness of the surrounding architecture.[5] But there is still great research needed in this neglected area, particularly in the context of real use and evaluation. Many of the traditional problem solutions become ridiculous in this new context. A high level of invention is needed.

All this flexibility is trivial, however, until we shift our inventive energies from trying to make pneumatic structures like hard structures with objectives like longer-lasting membranes and single-state high stability, and instead begin to value developing the innate softness and potential responsiveness of this form.

We might begin to approach this by going from the notion of architecture as shelter to architecture as interface. Now we are talking about a system which recognizes that there is a relationship between the architecture and its surround as well as between the architecture and its inhabitant(s), and that these relationships are loops where short and long-term mutual adaptive change is taking place. The construction of this kind of sophisticated pneumatics takes us into the realm of living things and ecology.

If this is difficult to recognize, it may have to do with the magnitude of these changes and their unfamiliarity in this context. A stone building will crumble as a response to its environment. All of our structures and materials have perishability, just as all living things. Our problem with hard architecture is not that it cannot change, but that the changes are not usually of a scale which has adequate meaning for people. The changes are not of a human scale (human effort will little affect the rate of change), and this is like the two parties of a conversation speaking different languages. The information transfer which is necessary to the process of adaptive change has been lost; participation by the perceiver is made impossible or irrelevant. But not all the changes should be of this very particular time-grain. The architecture we are developing is in larger forms (city as architecture) and smaller forms (clothing as architecture), and the multiplicity of its time-grains of response should be as rich as that of its environment. In each situation, the magnitude and rate of change ought to be compatible with what is meaningful in that environment. For example, at Antioch this means that there should be some changes which correspond to the moment to moment changes of the individuals and their groupings, some of the rhythm of the weather, some which reflect the academic schedule and calendar, some of the scale of the sun and moon. The participants derive meaning from these rhythms because of correlated personal changes which never before had an external referent.

With this level of responsiveness, then, we can see that one of the things the architecture can do is to absorb or feed back the information or energy it comes into contact with as its environment. In the context of biological-like behavior, this is simple. As an interface, however, the architecture also has the ability to transmit. The two loops become related and there is a transaction (between architecture-surround and architecture-inhabitants) which is (controlled by) the architecture. In hard architecture, the control which is built in the architecture is a screening out, sometimes selectively, of environmental information and energy. Another possibility, however, is that the interface may act as a lens to enhance aspects of the relationship it separates. It is an active interface and its functions are necessarily selective. This architecture has a behavior of its own which draws on both sides of the interface. It is like the selectively permeable membrane of a simplistic micro-organism. The notion, then, becomes one of an architecture which has, as a result of its structure, a capacity to change its own basic units of self-organization. It is self-referent. It modifies itself in response to changes in its environment, both inside and out, and the interactions which take place in real time.

These words will seem far-fetched to some, but obvious to others who are familiar with the development of self-organizing, adaptive, and self-referent systems in the mechanical and electrical modes as a high technology. I believe if we value the task, we can reduce this technology to a less abstract and more usable form, particularly with our architecture.[6]

As we increase adaptability, we can change the present over-design of structures necessitated by the need for one state to respond to varying environmental conditions.

To develop this responsiveness in pneumatic structures requires the adequate application of control technology which, for the most part, has already been developed, particularly for aerospace applications. Pneumatic structures are ideally suited to this kind of development because of their control potential. This suitability is largely because instead of relying on so much mass to build our architecture, we are substituting energy.[7] Hard architecture requires great energy input to begin to approach the responsiveness we are looking for, but when energy becomes identical with the greater part of our architecture, control becomes relatively easy. Instead of brute-force muscle, we’re using Jujitsu. We have added to it muscles of a hard system—a beginning of artistry in control.

The parameters and mechanics of control in this architecture, since they operate in a number of time grains, are evolutionary.[8] The problem is the orchestration of a number of environmental qualities, each in its own time scale,[9] to create a courteous and playful environment.[10] The control system functions, like the energy it controls, as an integral part of the architecture. It explores parameters, modes, and rates of change to create relevant responsiveness. Ideally, such a control system is distributed in the material of the architecture, combining structural with other forms of computation.[11]

The functioning of such a control system might be as follows:

Sensors which are providing information on the rules of behavioral changes in the architecture (skin and cable tensions [tendon sensors], internal climate [body temperature], lighting characteristics, condition of mechanical equipment [pain when malfunction begins] acoustics [hearing], power supply, interior structures, etc...), act through a self-organizing controller which cross-couples non-linear effects by continuously experimenting for achievement of high-level intentions. The controller is modifying the conditions of the architecture within the parameters set by a meta-control function. The fragments of sensed behavior are not responded to by corrective changes. Rather the system continually explores its own behavioral effects. Input and output are looped rather than separated by a controller. The loops and network of loops are provided more capacity to explore as they stabilize in an unsatisfactory way. The exploration which results from the addition of carefully constrained noise (so it will experiment in a non-random region) provides for not only magnitude of change, but also change in the basic units of response-time, complexity of cross-coupling, influence of any one behavioral sensor, rates of change, rates of rates of change, etc. This is the primary meta-control function.

The control loop we have described here enables the architecture to have a behavior of its own in exploring various internal relationships in the architecture. A behavior it effects interacts with the inhabitants and the surrounding environment as they loop back with new behaviors, and the new condition of the architectural loop is then sensed and modified again by the self-organizing controller.[12] The importance of cascading shorter and longer control intervals, behaviors that change hour by hour being nested within controls that operate quite independently day by day, will be evident. This interactive exploration of environmental behaviors enables a playful relationship between the architecture, its inhabitants and the surrounding environment (either of which includes people, plants, animals, etc.). Information on how well the system is doing loops back to be modified at the local, structural computation level, or by more general regional control, or by meta-control causing reinforcement on the whole state as it tends to settle into a meta-stable condition. This might take the form of indicating to the meta-controller that the discrete set of behaviors should be labeled and remembered, increasing the long term frequency of its repetition and making it available for future recall and use in special situations.

Reinforcement of behavior also takes place within the control loop as some fine-grain behaviors are reinforced by those of larger grain.[13] These processes tend to insure the gradual self-tuning and acquisition of courteous behavior by the system. This tuning is towards the creation of the information level that insures the variety necessary for the survival of the total system.[14],[15]

All of this responsiveness is, thus far, within certain parameters which change only in an explorative, evolutionary manner as the system learns when it is doing better at being actively supportive of inside and outside healthy environmental functioning. There are, however, times when a rapid change of these parameters is appropriate.

There are times when the control system should move into an emergency mode. This is far simpler and closer to systems with which we are familiar than the dialogue mode described above, and these controls can easily be put into a facility such as the Antioch structure. The function here is that of a decision in the context of information from sensors in the surrounding environment and monitoring of the primary control loop network, that indicates the necessity of shifting to certain emergency procedures through overriding control of the effectors in the architecture. For example, strain gauges on the cables and wind speed information may indicate the need, within a certain response time, to vary combinations of blower use and speed, ventilation, and interior temperature to quickly change the inflation pressure and corresponding cable tension to maintain structural stability. For gross emergencies, this kind of overriding control could be achieved by manual control by inhabitants who recognize and want to deal with an emergency situation, but often can be more effectively provided for by setting limits on loop behaviors, extending our level of prediction as our control skill warrants. When this function is built in to the system, we have an analogue to the control exerted in a crisis situation by the reticular core of the central nervous system, where certain sensory information results in abrupt behavioral change which assumes priority over all other behavioral tendencies.[16]

Another need for abrupt change in the architectural environment comes when the function of the space it encloses (or its role in the surrounding environment) changes abruptly and requires a whole new set of environmental behaviors. Control of this decision might require direct indication to the meta-controller that a new mode of behavior should be called from memory or synthesized. However controlled, the decision of the system involves knowledge of learned modes of successful behavior.[17] This ability to acquire, recall, and explore when the system is not critical, new modes of successful operation means that the system becomes richer and more responsive as its experience increases over time.

It is important to note that I am not describing a general system of control here, but rather a particular evolution for a first structure which need not be duplicated. As soon as there is interaction and skill is acquired by the inhabitants and surround, the behavioral model has already changed. Loops might be added into the surround where the architecture would be nourishing a vegetable garden. The system should suggest possibilities we could never imagine.

This childhood of a control system is a poor beginning which calls for extraction, re-thinking, and modification based on full-scale building and experience. Some of this can happen with the Antioch project, where we are trying to implement parts of it now, but still within the rather major constraints of a simple structural design and available materials. We need to develop guidelines which allow for fast overall adjustment of form, shape, size and other controllable characteristics.[18] We need membranes of more ecologically sound chemical composition that are more desirable for people to interact with; membranes of controllable elasticity and controllable transmission; membranes which function as beautifully and complexly as human skin.

My repeated use of the biological analog is not merely illustrative. It is a talking about a wedding of biology and architecture in the context of ecology to create what Rudolf Doernach calls “biotecture.”[19] The use of pneumatic principles is found throughout nature and its use in this task is obvious. Our architecture, and the technology it is an extention of, must become as living organisms if we are to go on living.

This is still only descriptive. The task of doing is enormous. We are working with a whole new technology; a post-industrial technology. The hard mechanical technology of the industrial revolution is useless in this task, and the inability of this old technology to deal with our present problems is becoming increasingly critical. We need help.

It is time to move into the new space.


Brodey, W.M. and Lindgren, N., “Human Enhancement; Beyond the Machine Age,” IEEE Spectrum, Sept. 1967 and Feb. 1968.

Bateson, G., Steps to an Ecology of Mind, Chandler Press, 1971.

Bateson, G., “The Role of Somatic Change in Evolution,” Evolution, vol. 17, no. 4, Dec. 1963.

Iberall, A.S. and Cardon, S.Z., Analysis of the Dynamic Systems Response of some Internal Human Systems, NASA Contractor Report CR-11, NASA, Wash., D.C., Jan. 1965.

Negroponte, N., The Architecture Machine, MIT Press, 1971.

Ekstrom, R., Technical Report (on the feasibility of the Antioch structure), Antioch College Env. Des. Program, Columbia, Md., April, 1971.

Doernach, R., “Bauen auf dem Meer” (Maritime Structures), Kunststoffe im Bau, K18 Oct., 1968, Verlag Chemie, Heidelberg, 1968.


Rossman, Michael, “The Technology of Technology and Social Reconstruction”; part of On Learning and Social Change, unpublished manuscript.

Hamilton, B. and Whitaker, W., Bubble Priere, Antioch College Environmental Design Program, Columbia, Md., 1971.

Wellesley-Miller, S., “Real Time Plans and Self-Organized Environments,” Participatory Architecture (conference Proc.), Manchester, Sept. 1971.

Storr, H., “Clothism and Design,” Colorado Quarterly, 1, (3), Winter 1954.

Development of a system of low-cost, highly manipulable interior components for large space-enclosing structures has been a major focus of the work of the Research and Design Institute of Providence, R.I.

Pask, G., “My Prediction for 1964,” Prospect, Hutchinson of London, 1962—(A particularly clear and delightful exploration of how these principles might enter into our lives with such examples as a “self-organizing dam.”)

Goering, P.L., “Energy Structures,” Canadian Architect, 1971.

Brodey, W.M., “Soft Architecture, the Design of Intelligent Environments,” Landscape, Vol. 17, No. 1.

Brodey, W.M., “The Clock Handles,” Ann. New York Acad. Sci., Vol. 138, pp. 895-899, 1967.

Johnson, A.R., “The Three Little Pigs Revisited,” in Eleven Views, Student Publication of the School of Design, vo. 20-1, North Carolina State University, April 1972.

Johnson, A.R., “Dialogue and the Exploration of Context; Properties of Adequate Interchange,” Proc. ASC, October 1970.

Johnson, A.R., “Self-Organizing Control in Prosthetics,” in Advances in External Control of Human Extremities, Yugoslav Committee for Electronics and Automation, Belgrade, 1970.

Brodey, W.M., “Information Exchange in the Time Domain,” General Systems Theory and Psychiatry by Gray, Dahl, and Rizzo, Little, Brown and Co, 1969.

Brodey, W.M., “Biotopology 1972,” Radical Software, no. 8, Raindance Corporation, N.Y.

Herber, F., Dune, Ace Books, N.Y., 1965. (see Appendix I on ecology, diversity and survival).

McCulloch, W.S., and Kilmer, W.L., “Some Mechanisms for a Theory of Reticular Formation,” in Systems Theory and Biology, Springer-Verlag, New York 1968.

Johnson, A.R., “Organization, Perception and Control in Living Systems,” Industrial Management Review, Vol. 10, no. 2, Massachusetts Institute of Technology, Winter 1969.

This work is being pursued particularly by the Eventstructure Research Group in Amsterdam, the Chrysalis group in Los Angeles (Mike Davies, Ian Stanton, Chris Dawson, Simon Longman and Mark Pilford of Columbia, Maryland), at Arup in San Francisco, and probably several others as well. I am here at Ecology Tool and Toy in New Hampshire and with present projects of Sean Wellesley-Miller in Boston.

Doernach, Rud, “Biotecture,” Architectural Design, February 1966.