Science for Designers: How Complex Adaptive Systems Work

Anatomical forms do not arise within one large undifferentiated collection; they develop as specific groupings of systems and sub-systems.

Today the world of design is in a position to benefit enormously from advances in sciences, mathematics and particularly, geometry—probably not in a way that many designers think. As humans we are remarkably good at conceiving the world as a collection of objects, their geometric attributes, and the ways they can be taken apart and re-assembled to do spectacular things (either perform marvelous tasks for us, or provide an aesthetic spectacle, or both).

This way of designing underlies much of our powerful technology—yet as modern science reminds us, it’s an incomplete way. Critical systemic effects have to be integrated into the process of design, without which we are likely to trigger operational failures and even disasters. Today we are experiencing just these kinds of failures in large-scale systems like ecology. As designers (of any kind) we must learn to manage environments not just as collections of objects, but also as connected fields with essential features of geometric organization, extending dynamically through time as well as space. This is a key lesson from the relatively recent understanding of the dynamics of “complex adaptive systems,” and from applications in fields like biology and ecology. At issue is not just avoiding failures. Though our designs can certainly be impressive, nature’s “designs” routinely put us humans to shame. No aircraft can maneuver as nimbly as an eagle (or a fruit fly, for that matter), and no supercomputer can do what an ordinary human brain does. The sophistication and power of these designs lies in their complex geometric structures, and more particularly, in the processes by which those structures are evolved and transformed within groupings or systems.

We can readily see that in the natural world forms arise as adaptive evolutions that solve specific kinds of problems—an eye gathers information about predators and prey, a wing or leg allows rapid movement, and so on. Anatomical forms do not arise within one large undifferentiated collection; they develop as specific groupings of systems and sub-systems. These systems in turn relate to and comprise other, larger systems. The structural dynamics of systems are consequences of interactions between parts and wholes. This is a new science built upon a previous generation of biologists recognizing the adaptive processes of form generation, and their characteristic geometries—what is now known as “morphogenesis”. Pioneers like D’Arcy Thompson saw that living structures had characteristic groupings that were intimately connected to the processes by which they grew. Crucially, these pioneers came to see that formal and aesthetic characteristics were not separate, but were systems-specific geometric attributes. Over evolutionary history, organisms had learned to identify such attributes, the better to respond effectively to their environments. Our own capacity to experience beauty is, from an evolutionary point of view, just such a biological recognition of what is most likely to promote our wellbeing.

M15-Fig2-Timothy Pilgrim

Soap bubbles form a complex pattern as a result of their mutual adaptation. It was not put in.
Image: Timothy Pilgrim, Wikimedia Commons

What does this mean for designers, in concrete terms? It means that all the parts have to be mutually adapted to each other to an adequate degree, through a process of some kind. So let’s consider a general procedure for adaptive design, one that uses these new insights from systems theory. First, we will need to decompose a design problem so that it actually represents fundamentally distinct yet overlapping subsystems. Second, we will employ several alternative decompositions of the system into more tractable subunits or components. As is known since the work of complexity theorist Herbert Simon, a hierarchical complex system has several inequivalent decompositions. Connectivity dictates how to perform each of the problem decompositions based upon one different aspect of the entire system: the designer has to discover and give equal weight to connective components as well as to the structural components. Relations among objects are just as important as the objects themselves, and system decomposition in terms of relations makes that clear.

M15-Fig3-Partitions

Six distinct ways (among an infinite number of possibilities) of partitioning a disk to implement radial sectors, or concentric rings, or linear strips, etc. In an analogous manner, we can decompose a system according to distinct conceptualizations, for example to emphasize the distribution of interior spaces, or the path structure, or exterior urban spaces, etc.
Drawing by Nikos A. Salingaros

For example, designing a building involves at least five distinct system decompositions. These could be concerned with:

  1. harmonizing the building’s exterior with its environment and avoidance of geometrical conflict, which of course includes adaptation to climate, orientation to the local solar and weather patterns, etc.
  2. connecting the site to the circulation present in its environment
  3. shaping public spaces, from a sidewalk to one or more open plazas
  4. planning interior paths
  5. identifying the interior spaces in relationship to each other

There could be other systems as well, based upon individual needs, conditions, and uses. Each of these problems requires a system decomposition that defines a distinct type of subsystem of the entire design. And each has to be addressed separately, at least initially. Of course, eventually everything will have to be recombined, and a professional with experience will in practice handle all of the subsystems simultaneously. But since this method is unusual for today’s designers, we offer this artificial separation to make the point of alternative decompositions. Our task as designers is to optimize the functions of each subsystem so that those functions support the whole system in which they are embedded, but do not impede any alternative system decompositions. We require adaptive selection criteria that guide the design to converge to an overall coherence (which we help along but do not dictate). The final configuration converges neither to an “approved” image, nor to some fixed initial abstraction, but rather towards an emergent quality of the system itself as it adapts to generate strong internal and external coherence. The operational secret for achieving a tight connection of a design to its environment is to make as many design decisions as possible on the site itself. In this initial conception, no overall form has yet been decided! The procedure described here was developed by Christopher Alexander, following a method used by humankind throughout the ages for vernacular building. Such a procedure simply cannot be performed in the office, because it is fundamentally contextual. The design method relies upon on-the-ground experience. Only after key decisions about the dimensions, positioning, and geometry of the various subsystems have been taken in the actual setting using one’s imagination aided by physical props, then, this information can be transferred to a scale model, sketch, and computer screen. Adaptive design’s principal aim is to facilitate the different components of a particular subsystem so they assemble themselves into a coherent subsystem. For example, the conditions and uses require specific internal paths, but there is freedom in connecting them into a network — this must be done in a way consistent with all the other system decompositions. Here is where the real novelty lies: we let each distinct subsystem develop according to rules for adaptation, and our role as designers is merely that of facilitator. Namely, we are not going to dictate its design using any preconceived ideas or images (a shocking suggestion for contemporary practitioners), only search for the possibilities that satisfy the constraints of use, site, environment, etc. In this way, the components we have to work with will, in a real sense, “assemble themselves”. This phenomenon is called self-organization — a very important topic that we discuss extensively in our essay “Frontiers of Design Science: Self-Organization”. The result should still have a degree of roughness, for reasons that will become clear later. This procedure is repeated for each distinct subsystem to give us several subsystems that are more-or-less coherent within themselves. In the end, we superimpose and combine all the different subsystems into a coherent whole. Crucially, the distinct subsystems will engage in a way that makes functional sense. Again, we don’t impose our will, but simply facilitate an intimate union of all the subsystems. In the case of a building as discussed above, there will be at least five subsystems, and these will need to merge together.

M15-Fig4-Biourbanism

Non-adaptive versus adaptive plans for a group of buildings: Left, the plan is only a formal geometrical idea; right, the plan reflects typical adaptations to several distinct systems of human needs, such as complex spatial volumes, movement, definition of usable urban space, connectivity on a human scale, etc.
Drawing by Nikos A. Salingaros

The final design will be a structural compromise among all the alternative system decompositions, which compete with each other in design space. It is important to accept and handle this “conflictual” component of design, which arises from the need to accommodate several distinct systems, each one of which has its own optimum, but which could very easily degrade another subsystem’s functionality. Thus, the intertwining of the distinct subsystems can only be achieved through each of the subsystems compromising to some extent. This is how the larger whole achieves an optimum configuration. This description might sound exotic—but something like this goes on all the time in natural systems. It’s the process by which the mitochondria adapt to the cell nucleus and vice versa, or the organisms within a reef’s ecology mutually adapt to one another. At our best, we do the same thing—or we let the natural processes around us do this for us. We “copy nature,” or we go through an “optimization cycle,” for example. But as we noted earlier, too often, we humans tend to treat the products around us as separated things of very limited function that we can choose to isolate or recombine at our whim, with little consequence. This is, functionally speaking, a mistake. According to a key principle from systems theory, we can only treat systems as closed up to a point. Ultimately we have to see the ways in which all systems are partly open and inter-connected. Biological and ecological systems—of which we humans are ultimately an inseparable part—are open systems. A key lesson for designers of all kinds follows: Product design can’t really be separated from environmental design. We are all, in some sense, environmental designers, working in the human environment. Since every system is only partially closed, we have to find ways to work on these systems as open systems — that is, as parts of larger, optimizing wholes. Routine failure to do so has led to our ecological misfortunes.

M15-Fig5-Place-Networks-Example

Human places are systems of room-like structures that span many scales — literal rooms indoors, and then more room-like outdoor spaces. These systems are made to adapt well to our activities and needs (especially our need for privacy) and to be adaptable by users — we can close doors and windows, draw curtains, etc. On the right, a composite example of a typical mixed-use London street.
Photos by Michael Mehaffy

This means we must come to see (and work on) these systems of spaces where we live as a fabric of connections between partially open sub-systems of spaces with geometric characteristics. As designers, our job is to weave together parts of this fabric into more life-supportive, continuous structures. We discuss the details of this structure elsewhere (in what is known as “place network theory”); but for now, we can think of this structure as a network of room-like structures, each with a membrane-like connection to the other spaces around it. (Think of rooms with doors and windows, gardens with gates and hedges, etc.) An important aspect of adaptive evolution is afforded to users in such environments. They give us the capacity to control the degree of stimulation and variety, to explore intricate and varying layers of space, to locate rich geometrical structures that users might find interesting and beautiful. We might elaborate on these structures as a way of clarifying them and making them more legible—or even more beautiful. It is the freedom to evolve our environment (in part), thereby vastly broadening its functionality, which is missing from the deterministic approach of most contemporary architecture. As we alluded to earlier, research in environmental psychology reveals that such aesthetic characteristics are essential attributes of human wellbeing—they are not separate from this cellular, systemic structure of the human environment. The boundaries of different spaces become identifiable borders. And the geometrical centers become identifiable points around which local temporal symmetries might regularly appear. We might see regular patterns of repetition or alternation, or other characteristic patterns of human use and movement that arise from the particular geometry. It seems we are hard-wired to find geometries that generate these patterns aesthetically interesting, and often very beautiful. (Elsewhere, we have discussed the fascinating and promising topic of biophilia in more detail.)

M15-Fig6-Place-Networks-Comparison

Two places in London, not far from one another, with opposite system characteristics: Left, a “place network” that is a well-articulated system of geometric spaces. Right, a place without a network — a jumble of poorly-articulated abstract parts, with little relation to human experience or need.

Photos by Michael Mehaffy

This, then, is a key role of environmental designers: to facilitate such adaptive evolutions in both short and long (more permanent) time scales. It is essential to understand and apply the geometric properties of human space, particularly its patterns of connections. We, as urban designers, or as architects—as designers of any kind—have to take this problem seriously. The art of our work lies in the way we elaborate and elucidate these deeper realities of life. Understanding geometric systems within environments gives us a remarkably coherent way of approaching the problems of the human environment. The question at stake is whether we can actually design, in the deepest spatial sense—that is, harness the organizational power of evolutionary systems, to generate richer, more connected, more adapted, more alive human environments. We must contrast this approach with today’s dominant “business as usual” approach—a holdover from an earlier pre-modern industrial mode of design (indeed, of science). Instead of creating and transforming mutually adapted systems, disconnected objects are created and assembled, and then aesthetic “packaging” is layered onto them. Someone creates the “guts” of the car, and then somebody else places a sleekly “styled” body on top. Or we create filing-cabinet-like buildings around prosaic “programs” and then we create razzle-dazzle aesthetic veneers, outside and perhaps inside—all package, no substance. Or we create filing-cabinet cities of superblocks and segregated zones, and then we “shrub them up” with various forms of landscaping and ecological gizmos. This last example often comes with a phony “sustainable” label. In the process, we leave a toxic planetary wreckage, the consequences of which, it is clear, we simply will not survive. This, too, is a necessary adaptation we must make—one that will challenge our orthodox thinking, about the very methods and aims of design.


Michael Mehaffy is an urbanist and critical thinker in complexity and the built environment. He is a practicing planner and builder, and is known for his many projects as well as his writings. He has been a close associate of the architect and software pioneer Christopher Alexander. Currently he is a Sir David Anderson Fellow at the University of Strathclyde in Glasgow, a Visiting Faculty Associate at Arizona State University; a Research Associate with the Center for Environmental Structure, Chris Alexander’s research center founded in 1967; and a strategic consultant on international projects, currently in Europe, North America and South America.

Nikos A. Salingaros is a mathematician and polymath known for his work on urban theory, architectural theory, complexity theory, and design philosophy. He has been a close collaborator of the architect and computer software pioneer Christopher Alexander. Salingaros published substantive research on Algebras, Mathematical Physics, Electromagnetic Fields, and Thermonuclear Fusion before turning his attention to Architecture and Urbanism. He still is Professor of Mathematics at the University of Texas at San Antonioand is also on the Architecture faculties of universities in ItalyMexico, and The Netherlands.

Recent Viewpoints