Misfits and architecture machines

A few days ago, I wrote about some basics of cybernetics, concluding with a snippet from Gordon Pask’s “The Architectural Relevance of Cybernetics. “Let us turn the design paradigm in upon itself,” he wrote, “let us apply it to the interaction between the designer and the system he designs, rather than the interaction between the system and the people who inhabit it.”[1]

This idea proved very attractive to two young architects: Christopher Alexander and Nicholas Negroponte. When architects started engaging with cybernetics, they saw new possibilities for how designers would work. Technology would surprise and challenge the designer, would break down design problems into smaller parts, would address issues of complexity.

Christopher Alexander applied cybernetics and AI (among other disciplines) to architecture in an attempt to address the growing complexity of design problems. He noted the difficulty of designing for intermeshing systems, even when the designed object itself (whether something as big as a village or as small as a teapot) seemed uncomplicated. “In spite of their superficial simplicity, even these problems have a background of needs and activities which is becoming too complex to grasp intuitively,” he wrote in Notes on the Synthesis of Form in 1964.[2] The design process he described in Notes required a computer to analyze complex sets of data to define “misfits”—design requirements—that the designer ameliorated by creating a form that solved the problem.

While Nicholas Negroponte is best known today as a technology guru and founder of the MIT Media Lab, I’m interested in his architectural background and the notion of “architecture machines”— evolving systems that worked in “symbiosis” with designer and resident that Negroponte thought would change the making of architecture. As director of the Architecture Machine Group at MIT, founded in 1968, he assembled a theory of how such systems would work in the 1970 book The Architecture Machine (dedicated “to the first machine that can appreciate the gesture” [3]) and the 1975 book Soft Architecture Machines, and a series of computer-aided design tools and programs throughout the 1970s.

An architecture machine, in Negroponte’s estimation, would turn the design process into a dialogue that would alter the traditional human-machine dynamic. He wrote, “The dialogue would be so intimate—even exclusive—that only mutual persuasion and compromise would bring about ideas, ideas unrealizable by either conversant alone. No doubt, in such a symbiosis it would not be solely the human designer who would decide when the machine is relevant.”[4] In order to achieve the design goals and close relationship with the user the machine would have to incorporate artificial intelligence, he wrote, “because any design procedure, set of rules, or truism is tenuous, if not subversive, when used out of context or regardless of context.”[5] Intelligence for Negroponte is thus not a passive quality but an active one, expressed through behavior, and improved over time.

However, building a successful architecture machine proved a much more difficult concept in practice because of the quality of interaction they achieved and their designer’s overall fascination with bells and whistles. The URBAN5 (1967) program was Negroponte’s first, major computer-aided design program that sought to use his ideas about conversation, dialogue and intelligence. In his own judgment, it failed because it could not adapt and its dialogue was too primitive. The shortcomings of URBAN5 led the Architecture Machine Group to develop “The Architecture Machine”—a time-sharing computer that in addition to typical peripherals, had a camera interface on wheels (GROPE), robot arm (SEEK), tablet-based sketching stations and “an assemblage of software.” Negroponte wrote, “The prognostications of hardware enumerated in wanton fantasy have been achieved and even superseded in the actual Architecture Machine of 1974. All too often we spend our time making better operating systems, fancier computer graphics, and more reliable hardware, yet begging the major issues of understanding either the making of architecture or the makings of intelligence.”[6] “The Architecture Machine” was perhaps a failure of its own success.

Today, computer-aided design systems proceed as we ask them to. They don’t jump in and do things for us, they don’t create new layouts, they don’t have conversations with us. Expert systems often fall short: they guess wrong, they get in the way. But have we thrown out baby with bathwater? In the surprises and the challenges of our systems, perhaps we would come up with things we never would have imagined.

[1] Gordon Pask, “The Architectural Relevance of Cybernetics,” Architectural Design 7, no. 6 (1969): 496.
[2] Christopher Alexander, Notes on the Synthesis of Form (Cambridge,: Harvard University Press, 1964), 3.
[3] Nicholas Negroponte, The Architecture Machine (Cambridge, Mass.,: M.I.T. Press, 1970), 11-12.
[4] Ibid., 11-12.
[5] Ibid., 1.
[6] Negroponte, Soft Architecture Machines, (Cambridge, Mass.,: M.I.T. Press, 1975), 157-71.

This is one of 50 posts about cyborgs, a project commemorating the use of the term.