2.0 Background

A background of architecture with regard to systems science and complexity science is provided in this section.

Systems science and complexity science are used as a framework for this thesis because of their comprehensive
agenda. General systems theory (GST) is envisioned by Bunge (Bunge, 1977) as a scientific metaphysics and offers
an array of processes and tools for engaging large unwieldy problems which overlap with many disciplines
including architecture and design. Boulding writes, “General Systems Theory is a name which has come into use to
describe a level of theoretical model-building which lies somewhere between the highly generalized constructions of
pure mathematics and the specific theories of specialized disciplines” (Boulding, 1956). Klir offers a common sense
definition of a system as simply a set of things and the set of relations among them (Klir, 2013). S = (T, R).

Epistemologically, a system as defined by Lendaris is a defined focal unit or whole that consists of defined sub-units
or elements as well as a defined supra-system or larger context. These three aspects can be thought of as: the suprasystem perspective B, the unit perspective C and the sub-unit perspective D in the following diagram (Fig. 2.0.1) (Lendaris, 1986)

FIG 2.0.1. System diagram (Lendaris, 1986). Redrawn by author.
Essentially any focal perspective is nested within a higher and lower level. In an open sense, this pattern can expand
upwards and downwards indefinitely, in a closed sense, there is some limit to the expansion or contraction. GST is
concerned with pattern of relations and processes at multiple scales and over time. Complex systems are
characterized by feedback loops between many variables. As such, systems approaches are particularly suited to the
understanding and practice of architecture and design. Simon, in developing hierarchy theory, makes a distinction
between complexity and complicatedness. To Simon, complexity represents the behavior of a structure with many
levels and parts or sub-systems coordinated within an overall framework or hierarchy. The behavior of such a system
is often simple or well organized. Whereas, complicatedness represents a relatively simple structure with fewer or no
hierarchical levels but many parts often having very complicated behavior (Simon, 1996). An example of complex
behavior may be the traffic patterns in New York City, whereas, an example of complicatedness may be the traffic
patterns in Mexico City.

Architecture is fundamentally systems oriented in the sense that it has to do with many interrelated parts over a
range of scales. What is now called systems theory has been a topic at least since Aristotle’s Poetics (Aristotle,
Golden, 1968). In modern times artificial systems have been studied in the cybernetic project of the 1950s (Ashby, 1961), (Wiener, 1961) and Simon’s, Sciences of the Artificial (Simon, 1996) to current research in complex adaptive
systems (CAS) such as living systems and evolutionary programming. In today’s increasingly complex world, where
we face the motherlode of wicked problems, systems approaches have undergone a renaissance of sorts. Important
multi-disciplinary research into complex systems is happening at the Santa Fe Institute (SFI) and New England
Complex Systems Institute (NECSI) among others. This research agenda includes a variety of concepts such as:
edge of chaos, emergence, self-organization and open systems far from equilibrium within the general study of
artificial intelligence (AI) and artificial life (AL). Active research also includes new approaches to understanding
biological and urban scaling through the application of fractal geometry (West, 2005), (Bettencourt, 2013). Fractal
geometry is an important aspect of complex systems research and can in some ways be used as a precise
mathematical measure of complexity in the form of Hausdorff dimension which is also called fractal dimension
(FD) (Mitchell, 2009). Fractal geometry and fractal dimension is central to this thesis and will be discussed further
in section 2.2. The research and theory discussed above offers many tools for assessing as well as designing systems
and provides an appropriate theoretical and practical framework for investigating systems in architecture.

To borrow from the field of Artificial Life (Bedau, 2003), we may think of systems in architecture as three types:
hardware, software and wetware.

HARDWARE: In terms of hardware, architecture is concerned with many inter-related physical parts (brick and
mortar) and it would be easy to list examples of physical systems and subsystems which the architect must
coordinate within a design. Some generic categories of physical systems in architecture are: structural systems,
mechanical systems, electrical systems, plumbing systems, glazing systems, waterproofing systems etc. The physical
and ecological context for a building can be thought of as a supra-system in Lendaris’ terminology. Architects must
integrate all these various systems together into one relatively integrated whole if a building is to function properly.
Some architects have also developed meta-systems or systems of systems towards a more unified approach to
managing the sub-systems within a building. One such approach was developed in the 1960’s by Ezra Ehrenkrantz
called School Construction Systems Development (SCSD) (Boice, 1965).

SOFTWARE: Architecture is also concerned with software or computer based systems. These systems include
computer modeling, computer aided design (CAD) and building information modeling (BIM). These types of
systems are currently developing in a variety of ways and fall under the general heading of parametric design.
Parametric design is a manner of design where parameters and rules are encoded algorithmically and allow
designers to utilize various aspects of CAD and BIM to solve complex problems. These problems are increasingly
multi-variate and data driven ones that require mathematical analysis and computation based search heuristics that
were, until recent times, unavailable to the architect. Genetic algorithms (GA) and evolutionary programing are
examples of these types of computation based systems and have given the architect powerful new tools to invent and
study form. Galapagos is one such tool and represents an out of the box genetic solver plug-in for Rhino
Grasshopper tailored specifically for architects and urban designers (Rutten, 2010, 2013). Genetic algorithms and
evolutionary programming will be discussed further in section 2.1.

WETWARE: Architecture is also concerned with wetware or the socio-technological dimension. This includes the
behaviors and needs of people which has a pragmatic dimension in the sense of habitable and occupiable space as
well as the historic and cultural context which can be thought of as a supra-system. I also include cybernetics in this
category or the hybridization between living organisms and technology. Wetware also suggests a more ephemeral dimension in terms of: meaning, beauty, culture etc. This abstract / symbolic dimension is important to mention
because it gets at some of the ontological aspects of architecture and design that are more qualitative and metaphysical.
To borrow from linguistics, if the first two categories described above (hardware, software) are pragmatic
and syntactic, this third category (wetware) is semantic. As such, the problem architecture attempts to solve is very
difficult to define in concrete terms alone and must also include the intent of the designer which can be a highly
subjective and intuitive aspect. Some theorists have attempted to quantify meaning in architecture using information
theory and other mathematical models (Baird, 1969) (Alexander, 1964) as well as shape grammars and space syntax.

This approach has met with significant pushback, with critics pointing out that the term information and how it is
used in certain contexts does not imply semantic meaning (Arnheim, 1971). Arnheim is skeptical of advocating
quantitative approaches in assessing design and emphasizes the need for the “finger pointing” critic (Arnheim,
1977). Simon’s concept of partial decomposability suggests that systems with many parts and sub-systems often
reside within a range of order and disorder that relies on hierarchy and modularity as organizing principles (Simon,
1996) (I may add scale-free modularity). This idea is expanded on in complexity theory where complex adaptive
systems are theorized to move toward the edge of chaos (Kauffman, 1991). Langdon and Wolfram have studied a
range of cellular automata with complex behaviors capable of universal computation within certain parameters
which also suggests a zone where structure and flexibility co-exist (Langdon, 1990) (Wolfram, 2002). Mitchell and Crutchfield have demonstrated the computational potential of cellular automata to solve problems using genetic algorithms (Mitchell, Crutchfield et al., 1993). I discussed above several broad categories of systems in architecture and a framework for approaching the design process. Next I discuss the overlap between a systems theoretic framework and architecture today.

Design technology has in some ways paralleled early developments in cybernetics. Archer and others developed models of design processes in the 1960’s which reflected parallel developments in systems approaches such as the systems
morphology models introduced by Hall (1969). Alexander and his approach to architecture and urban planning as a
pattern language has been influential to a generation of designers and has also influenced the development of object
oriented programing. These concepts are directly related to building information modeling (BIM) and parametric
design. BIM utilizes parameterized objects that provide a quick and easy ways to update a three dimensional model
globally to solve specific design problems. Models may also be exported in a variety of ways, for instance as
traditional construction documents (CDs), 3D printed models, or directly to automated manufacturing systems and
so forth. Objects are also available from manufacturers as free downloadable libraries. BIM has introduced a
platform where architecture, manufacturing and construction have begun to be integrated in a way that never would
have been possible before the advent of digital technology.

Some architects and visionary designers have taken the potential of digital technology in far more exotic directions
creating virtual spaces that employ algorithms to connect architecture and context in “intelligent” ways or not
confined by physical constraints at all, such as the work of the firms Morphosis and Asymptote, and the work of
artist / architect Marcos Novak (Judelman, 2004). A pioneer in algorithmic design is Karl Sims who developed
artificial evolution in the 90s as a tool for designing virtual creatures that could respond to their environments and
adapt (Fig. 1.2.1) (Sims, 1994). This work follows from earlier research into genetic algorithms and evolutionary
programming which will be discussed next.