by Philip Belesky

Abstract
The philosophy of Giles Deleuze is the primary reference point for architectural theorists
when discussing computer-generated forms.
This paper examines this connection, by investigating the conditions inherent to the
philosophy of Deleuze, and to the process of designing architecture within a digital
environment. It identifies that a connection can be established between these two
fields because they both mandate an understanding of the creative process as based
upon virtual structures. Present within Deleuze’s ontology, and within the logic of
computation, these virtual structures are abstract processes that dictate the manner in
which form is produced.
This connection is important because it has not been the traditional method of correlating
these two fields. Moreover, by shifting the modes of discourse and practice to engage
with this connection, Deleuzian theory and computer-generated architecture can work
towards overcoming the clichés that they are associated with. This shifting mode of
practice and theory is demonstrated within ‘parametric architecture’ and ‘morphogenetic
design’ theories. These emerging fields demonstrate that an operative understanding of
virtual structures allows architects to harness the digital environment in a way that
disrupts the prior paradigms of computer-generation.

CONTENTS

• ABSTRACT
  
• INTRODUCTION
  
• DELEUZE’S ONTOLOGY
  
• PARAMETRIC ARCHITECTURE
  
• SYNTHESIS
  
• BIBLIOGRAPGHY

INTRODUCTION

Whereas the contemporaries of Giles Deleuze — primarily Michael Foucault and
Jacques Derrida — influenced philosophy and architecture concurrently, architects
did not start to reference Deleuzian theory until the ‘digital revolution’ of the 1990s.01
As a result of this revolution, architects began to utilise advanced computer software
to design and construct a novel formal language;02this prompted a similarly novel
theoretical discussion that was dominated by Deleuzian concepts such as The Fold and
Smooth Spaces. Because Deleuze’s work lay latent in architecture until this new area of
architectural discourse emerged, it can be inferred that his theories were especially wellsuited
to exploring the implications and possibilities of computer-generated architecture.

This referencing of Deleuze has been widely criticised by both philosophical scholars03
and computer-science theorists.04 The former suggest that architecture has ignored the
major tenets of Deleuze’s philosophy because of a near-exclusive focus on a small set of
relatively minor concepts. The latter suggests that the architectural reading of Deleuze
is crippled because it discusses his ideas only in relation to the phenomena and aesthetics
of buildings, and thus ignores the design environment of the computer. If the criticisms
from each of these two camps are to be addressed there is a need for architectural
discourse to move its investigation beyond the end results of both Deleuzian theory
and computer-generated architecture in order to reconcile these two fields in a more
intrinsic manner. Thus the aim of this research is to investigate the connections between
Deleuzian theory and computer-generated architecture by examining the fundamental
conditions that enable their compatibility.

These ‘fundamental conditions’ are the routines of praxis. In both architecture and
philosophy, the end obscures the means. Buildings and concepts only hint at the
processes that were used to generate them. To examine these underlying processes is to
uncover the intent of the philosopher or architect, and to understand the modus operandi
through which they create.

Throughout his work, Deleuze posited a model of the creative process that exists
across philosophy, the arts and the sciences.05 Within this model, creation is seen as an
interaction between three elements: the initial source information (Plane), an authorial
lens that filters and orders this initial reservoir (Persona), and that which is produced as
an end result. The nature of each of these three elements differs between disciplines. For
example philosophy creates Concepts as an end-result; science creates Functions and art
creates Affects.06 However, the interaction between each of these elements is constant
across disciplines, and thus can be used as a universal model of the creative process.
In relation to this model, my investigation is concerned with the second element, that of
the Persona. This element can be seen as the crux to understanding the creative process
of a particular author or ideology because it is the means by which source-material is
processed into an end result. It is particularly important when working across disciplines
because it is an abstract operation that can be applied to a variety of different contexts
while still enacting a consistent intent. A prominent example of this would be the Persona
of Derridian deconstruction, which can be considered as a routine whereby sourceinformation
is critically analysed to uncover abstruse relations and intents. In philosophy,
this was used to produce critiques that focused on the composition of language and
ideas within a text, whereas in architecture it was used to produce buildings that were
focused on subverting traditional compositions of form. Both of these examples adhere
to this basic methodology of deconstructivism, but then synthesize the Persona with the
specifics of their disciplinary context.

The fundamental conditions, or Persona, of Deleuzian philosophy are encapsulated
in the way in which Deleuze views the world, and the processes through which he
develops his conclusions. These conditions are investigated in Chapter 1. The Persona of
computer-generated architecture is encapsulated in the procedures that must be engaged
in order to produce architectural form within a digital environment. These conditions
are investigated in Chapter 2. The concluding chapter discusses the marriage of these
two subjects within architecture in relation to a mutual compatibility between Personas.

NOTES ON DELEUZE

Although Deleuze’s work is a vehement counter to post-modern philosophies, his
writing contains a meta-interest in the mediums of text and language that was typical
of his contemporaries.07 Each of his works08 espouses a consistent set of ideologies and
concepts, yet the terminology used to refer to them, and the argumentation used to
develop them, is very different in each of his major books. The point of this technique
was for Deleuze to regularly retell his thoughts with new narratives and neologisms, thus
creating new lines of argument and expanding the semantics of his concepts through the
use of semi-overlapping synonyms.

These linguistic games create problems when referencing and discussing Deleuzian
concepts. Within this paper, I have chosen to utilise the lexicon established by Manuel
De Landa in his discussion of Deleuze in Intensive Science and Virtual Philosophy. This
perspective was chosen because of De Landa’s emphasis on the analytic and scientific
portions of Deleuze’s analysis, and because of his favoured position09 as an expert in the
applications of Deleuzian theory within software, art, and architecture.

NOTES ON THE PARAMETRIC

The theorization of computer generated architecture has produced a variety of labels
and ideologies such as virtual architecture, computational architecture, generative
architecture, blobitecture, genetic architecture, topological architecture and nonlinear
architecture. Presently, these labels have yet to coalesce into a single prevailing term.

From Chapter 2, I commit to the term ‘parametric architecture’ rather than ‘computer
generated’ or ‘digital’ architecture. The reasons for this will be made clear in that chapter,
but it should be noted that these terms never signal a preference for a particular branch
of architectural theory or ideology, but instead always refer to the inherent processes of
computation.

DELEUZE’S ONTOLOGY

In this chapter, I discuss a series of ideas that comprise the crux of Deleuzian philosophy.
I selected these ideas because they constitute the Deleuzian Persona; they are the
principles of his ideology that define his world-view, and thus form the fundamental,
underpinning notions that all the later applications of Deleuzian thought filter through.

Deleuze expresses these ideas earliest and most definitively in Difference and Repetition,
wherein he constructs the basic tenets of his philosophy through multiple lines of
metaphor and reasoning. This chapter recreates this construction, but focuses exclusively
upon Deleuze’s use of mathematic, scientific and spatial references to articulate
and justify how his philosophy operates. By focusing solely on these arguments, an
understanding of Deleuze can be developed that is highly applicable to other disciplinary
contexts — especially fields which are comfortable with technical references.10

AN ONTOLOGY OF MORPHOGENESIS

Ontologies articulate the most fundamental assumptions that define reality. An
ontological system attempts to define the manner in which entities can be said to be
exist. Primarily, it is that which causes certain phenomena to be identified as discrete
objects, and how such discrete objects can be correlated to, or distinguished from, one
another.

Ontologies can be broadly classified according to their degree of anthropocentricism11.
The most human dependent ontologies define only the act of perception itself as real.
They posit that there is no reality outside of what we perceive, and thus the only things
which can be said to exist are the mental processes of perception: the linguistic and
conceptual frameworks that classify, explain, and thus are, the information that we
receive from the senses. In contrast, there are ontologies which completely disregard
perception, positing that the world exists completely independently of our experience
of it, and that all phenomena exist independently and equally, that they are all real
regardless of whether they are observable or unobservable by the human mind.

Before examining the ontology of Giles Deleuze, it is important to note that it belongs
to the latter category, that it is an ontology of the real — a metaphysic that is inalienably
material, and in which entities are always inherently composed of the concrete conditions
of objective reality.12 It is Deleuze’s utter rejection of the anthropocentric perspective that
placed him against the idea that language was the cornerstone of reality, and thus in direct
opposition to the prevailing school of thought that dominated philosophical discourse
in the 20th century.13 Because of this distinct difference, the ontology of Deleuze is
commonly considered to be his most important project.14 Operating simultaneously as
the subject of, and modus operandi for, all of his work, it was the consistent application of his
unique metaphysic to a great diversity of subjects that made his research so interesting.15

Ultimately, the grounds for this difference can be traced back to a single fact,16 that
Deleuze utterly rejected all forms of transcendental essences — the abstract, conceptual
conditions that are used to classify, and thus define, phenomena. To illustrate:

In observing a dog, it is generally held that there is an essential condition of
‘Dog-ness’ to which the observed creature belongs. This essential concept of Dog
is then used to identify and correlate distinct entities (the specimen, Spot) within
a whole (the species, Dog). In this way, there are two distinct and dichotomous
entities at play; a particular instance of a four-legged-domesticated-mammal,
and the eternal archetype of Dogs.

Questions of this object:essence dichotomy are found throughout the history of
philosophy, and it was the different conceptual models of how this dialectic operated
that defined one ontology from another.17 The key to Deleuze’s ideology was that he
sought to eradicate this dichotomy outright by positing a world-view that possessed a
single, univocal perspective that held every entity as an ontological equal. He achieved
this by creating an ontology that redefined each entity as being exclusively comprised
of the dynamic processes that led to its creation. Instead of properties being imposed
onto entities by our senses or minds, they are held to be inherent to the entity itself.
Because of this, there ceases to be a distinct ontological category between distinct
entities (Spot) and eternal archetypes (Dog), as both the individual instance and the
categorical property are held to exist in exactly the same manner: through the unfolding
of a generative process. To illustrate:

In observing a dog within a Deleuzian ontology, we would recognise that there
is an overlap of conditions, that there is an individual organism that is also an
instance of a species. What changes is that there is no conception of either of
these entities as eternal, or as defined by mental observation. Dogs are seen as
an evolutionary process defined by untold generations that have been, and will
continue to be, shaped by chance and environment, whereas Spot is an organism
created through embryogenesis and growth, which is also shaped by chance and
environment. The only difference between these two entities is the temporal
scale at which each operates — millions of years for the one, a dozen or so for
the other.

In removing the totalities of transcendental essences, Deleuzian thought stresses a
bottom-up perspective, in which all entities can only ever be seen as the sum of their parts,
and whereby universals must always be based upon the concrete conditions of reality.18
In doing so, identity is granted to entities through explanations, not descriptions. These
explanations of the processes of creation operate across all scales and contexts, such as
the scientific models that define properties like flight or conductivity, the psychological
models that define fear or Pavlovian response, or the biological models that define
evolution or embryogenesis. This concept of unfolding generative processes is broadly
referred to as morphogenesis.19

This line of Deleuzian thought this raised a question: how are ‘generative processes’
not just another meta-reality ontological category, another way of defining an essential
category? Was this not just a shifting of the notion of universal properties from the endresult
onto the conditions of creation? If that were true, the concept of transcendental,
non-real,properties would still exist, and the Deleuzian ontology would not be truly
realist. To answer this question, Deleuze developed the concept of the virtual multiplicity.
This concept is an explicit, operative conception of how these morphogenetic processes
function. It makes it clear that these processes are inherently real, and have no elements
of transcendence.

THE MULTIPLICITY

As with its etymology, the underpinnings of the multiplicity in Deleuze’s philosophy are
mathematical. Over the course of his major works it was defined in relation to numerous
contexts, but by focusing on its initial definition in Difference and Repetition, and Manuel
De Landa’s examination of it in Intensive Science and Virtual Philosophy, we can create a
narrative that selectively defines the concept of multiplicity purely through mathematic
and scientific phenomena.

INTENSIVE AND EXTENSIVE

The distinction between intensive and extensive phenomena is crucial to Deleuze, and to
his conception of the multiplicity. These terms originate in thermodynamics, where they
form a binary that classifies all scientific phenomena20 into two distinct types.

A property is said to be extensive if its magnitude is dependent on the size of the overall
system. This means that if the size of the system is reduced, there will be a corresponding
change in the value of the property. For example, an object’s volume is clearly extensive21
— if you slice an apple in two, the amount of volume it possesses is halved. Mass, length,
energy, electrical resistance and many other properties are also extensive.

In contrast, the magnitude of an intensive property is completely independent of the size
of the overall system. This means that if a portion of the system is removed, the property
will remain unchanged. For example, density is intensive: — if an apple is sliced in two,
the density of each portion is identical to that of the prior whole. Temperature, pressure,
velocity and viscosity are also examples of intensive properties.

This distinction is important to Deleuze, as his philosophy is focused on an examination
of phenomena that are inherently intensive, and how that intensity is perceived. Of
the entities that we observe in an everyday sense, Deleuze points out that the majority
are discrete phenomena, and are thus extensive. For example, countries and people are
observed as having geographic and physiological thresholds that are extensive because
they are discretely spatial. In contrast with this, a gas would occupy space in an intensive
manner because its borders can not be demarcated.

Deleuze maintains that, for all entities, extensive phenomena mask an underlying
genesis developed by intensive conditions. For example, the weather manifests itself as
distinct, tangible objects in the form of rain or snow, but it is the gradiented changes
in temperature and pressure across the local atmosphere that lead to the creation of
these discrete phenomena. The underlying intensive properties — temperature, pressure,
humidity — are all imperceptible, but nevertheless extremely real. Deleuze suggests
that it is intensive conditions that are the most productive in creating difference and
diversity, and they must therefore be understood in order to examine the origins of
morphogenetic processes.

MANIFOLDS AND PHASE SPACE

The notion of the manifold forms the basic structure of a multiplicity. It borrows heavily
from the mathematical field of differential geometry, a field of investigation into non-
Euclidean conceptions of space that can be defined through algebra and calculus.22
In traditional geometry, points are articulated within a metricised matrix through a
discrete set of co-ordinates In differential geometry, manifolds define curved spaces23
whereby co-ordinate points are articulated solely by their instantaneous rate of change,24
rather than through an instantaneous location. In this way, manifolds can be said to be
intensive, because they are never defined in the discrete terms of Euclidean ordinals.

This space-system accomplishes two important things. The first is that there is no fixeddatum,
no x=0, y=0,z=0 against which all things are measured. The second is that there
is no extrinsic means of co-ordination other than the space itself; that is to say there is no
higher dimension which the current plane can be measured against. Together, these two
conditions create a geometry which can accommodate an infinite (n) number of planes.
This collection of n-dimensional planes is referred to as a manifold25.

Deleuze then intertwines this notion of the manifold with another mathematical
concept: that of phase space. Phase space is a method of representation in which all the
possible states of a system are represented simultaneously. For example, in a physical
system the velocity and displacement of a single mass on a spring can be measured over
time, showing that, after any initial displacement, its acceleration and kinetic energy
eventually reach equilibrium at zero. A phase space diagram would visualise this event
by plotting these two variables on a graph, and tracing their values over time. In this case,
a spiral pattern would be created, according to the reciprocal but entropic oscillations
between acceleration, energy and gravity within the spring system.

Deleuze then equates this idea of phase space with that of the manifold, by expanding
upon the notion of what phase space can conceptually represent. According to Deleuze’s
conception, each property of an object can be conceptualised as a ‘plane of possibility’ in
which all the possible values for a property exist in a kind of conceptual map, or graph.
As earlier, these planes of possibility are comprised of intensive spectrums, rather than
discrete values. For example the brightness of a light bulb possesses a plane of possibility
that is a gradient spanning Off, Dim, Bright and Full. A light bulb also possesses a
plane of possibility for its temperature property, its voltage property, its colour property,
its physical form, and its current.

Deleuze equates each of these ‘planes of possibilities’ as analogous to the planes of a
manifold, so that each of an object’s properties, forms a single dimension within a
manifold space that then encapsulates all the possible values of the property. If we were
to imagine a light bulb as a system composed only of brightness and temperature, this
would create a manifold of 2 planes — a 2 dimensional space. In this space, the entirety
of an object’s states are represented by the inter-meshing of these two planes. Thus, every
possible combination of brightness and temperature are represented in this manifold.
Therefore, a Deleuzian manifold acts as an n-dimensional phase space, a ‘matrix of
potential’ that represents all of the possible properties, and the possible values of each
property that are inherent to an entity. These matrices range from relatively simple (the
spring system) to infinitely complex (an animal).

ATTRACTORS, SINGULARITIES AND BIFURCATIONS

To develop this idea of multiplicity and phase space further, Deleuze defines the
operations of his manifolds through several highly formal elements.26 These elements
are not external to the manifold, but merely a label for the tendencies that each possess.

Attractors are features that morph the shape of the state-space so that it is structurallybiased
towards certain outcomes. In this way, each plane of possibility has a trajectory
that manipulates its values towards a predisposed condition. For example, the spring
system has an attraction towards the equilibrium state of rest, thus its properties trend
towards a zero outcome when disturbed. This represents the spring’s dissipation and
absorption of any initial modification. In the case of the light bulb, an attractor would
morph the inter-meshing between temperature, current and brightness so that an
increase in current would also increase the latter values. These features the attractors
create are defined as singularities — sub-sets of phase space that define points of interest.

A more complex example of this phenomenon could be in the process of crystallisation, or
in the bubbles produced by soap. In both cases, a series of physical-chemical tendencies
result in specific formal outcomes, and thus can said to act as attractors that guide the
system towards singularities. In the case of crystallisation, the populations of molecules
collectively seek to minimise bonding energy by forming direct and rigid structural lattices;
such as the formation of salt crystals by way of cubic forms — however, in the case of
bubbles, the population of molecules seeks to minimise surface tension by minimising
surface area, thus leading to spherical forms.

Due to both of these attractors — minimal surface area, or minimal bonding energy
— the matter is guided towards singularities: bubbles or crystals.27 It is these innate
tendencies in the manifold-space of matter that make objects unique and thus create
identity. Alternatively, it is the shape of an object’s phase space that makes it unique
because it defines a specific matrix of potential. In this way, the ‘bubble-ness’ of bubbles
is seen to be an inherent and immanent property of the matter itself because this outcome
has always been present within its manifold. In this way the matter is morphogenetically
pregnant: able to create forms and identity without the need for any exterior observation
or interference, because it is a self-organising system driven by a completely endogenous
process.

Bifurcations are phase-transitions that represent shifts in phase space, whereby as an
object passes certain thresholds, the set of guiding attractors changes, thus changing
the shape of the phase space and therefore the behavioural patterns of the object.
For example, the flow patterns found in water will shift as the temperature increases,
changing from a steady pattern to a defined flow and finally to a chaotic turbulence as
the transition from conduction to convention occurs. Each of these pattern changes is
a bifurcation in hydrodynamic behaviour, whereby a new set of attractors defines a new
type of flow.

Particularly critical thresholds are termed ‘symmetry breaking events’. In these, clusters
of bifurcations act simultaneously to dramatically modify phase space by introducing
new sets of attractors and bifurcations, thus dramatically redefining behaviour patterns.
In water, for example, the transition points between gas-water-ice are symmetrybreaking
bifurcations that entirely shift water’s physical properties. The formation and
development of an embryo can likewise be understood as a complex cascade of symmetry
breaking events whereby anatomical elements are increasingly defined through cellular
division and specialisation.

THE VIRTUAL

To synthesise all of these concepts, a multiplicity is a system defined by the features of
a manifold.28 This manifold is comprised of intensive planes of potential, which, taken
together, embody all of the possible states of an entity. Moreover, these manifolds are
driven by a unique set of attractors and bifurcations that shape them towards specific
tendencies and outcomes. Multiplicities thus “give form to processes, not to products.”29

Going back to the ontological distinction discussed earlier, it is traditional to divide the
world into the ‘real’ and ‘transcendental’. Deleuze rewrites this, by viewing all entities as
created and identified, by a distinct morphogenetic process. To prevent his world view
being divided into the ‘real’ and the ‘possible,’ Deleuze conceives of these processes as
part of a higher concept, the multiplicity. This multiplicity contains the ‘real’ within its
matrix of potential, as well as the inherent tendencies of an object that give it identity.
However, unlike ‘essences’, multiplicities in no way resemble the objects they create; they
are abstract and intangible structures.

To rewrite the traditional essence-object distinction, Deleuze posits an actual-virtual
distinction. According to this distinction, all objects have a virtuality defined by
multiplicities, and an actuality that is their instantaneous state in reality as matter.
Significantly, Deleuze holds both the actual and the virtual as ‘real’. Moreover, the
virtual can be held to be more real because multiplicities define systems that contain
both the present, actual state of an entity as well as the morphogenetic recipe that could
generate all possible past and future actualisations. In this way, the ‘real’ exists inside the
virtual, and prior to its existence in the actual. The virtual shelters the genetic conditions
of the real, and therefore entities are always mere actualities of intangible — but everimmanent
— virtual structures.

CONCLUSION

“The key to the ontology I defend is the idea that the world is made out of individual
entities at different levels of scale, and that each entity is the contingent result of an
individuation process30”

In summary, Deleuze creates a unique world view through an unrelentingly realist
ontology that replaces the eternal transcendence of essences with the immanent virtuality
of multiplicities. Therefore, the adoption of the Deleuzian Persona mandates a conscious
and continuous engagement with the virtual.

PARAMETRIC ARCHITECTURE

In this chapter the conditions that comprise the design environment of computergenerated
architecture are discussed. These conditions construct the Persona of digital
architecture because they dictate the operatives that must be engaged when utilising
the computer. This Persona is particularly relevant because of its heavy influence on the
design process, and thus upon the architecture that is produced as a result.

This Persona is identified as a methodology defined by the abstract processes of
computation: the craft of creating computer code, and the means by which code operates.
The nature of this craft is discussed in relation to the history of CAD software because
a clear distinction can be made between prior paradigms, and the recent emergence of
‘parametric architecture.’

COMPUTERISATION

From the outset, a distinction needs to be made between computation and computerisation:

“While computation is the procedure of calculating, i.e. determining something
by mathematical or logical methods, computerization is the act of entering,
processing, or storing information in a computer or a computer system.
Computerization is about automation, mechanization, digitization, and
conversion. Generally, it involves the digitization of entities or processes that are
preconceived, predetermined, and well defined. In contrast, computation is about
the exploration of indeterminate, vague, unclear, and often ill-defined processes;
because of its exploratory nature, computation aims at emulating or extending
the human intellect.”31

The first CAD programs operated very differently to those of the present. The
forerunning examples of CAD acted as frameworks for the storage and visualisation of
geometric data. The geometric data itself had to be produced through the use computer
programming languages, where code was developed that would produce geometries that
could then be translated into co-ordinates.32

As CAD softwares entered into mainstream practice over the span of the 1990s their
nature rapidly changed, and they became increasingly used for computerisation, rather
than computation. This was because they embraced the use of graphical user interfaces
in order to make the programs accessible to a wider audience.33 Although this made the
software accessible to those without a knowledge of programming languages, it did so
by reducing common programmatic procedures into predefined ‘tools’. Moreover, these
tools themselves could now be used to directly manipulate visualisations of geometric
data, thus producing an interactive feedback loop between these tools and architectural
form. This ability to directly manipulate a graphic model through a graphic interface
made CAD accessible to designers, and thus this mode rapidly eclipsed the previous
code-driven approach, and became the dominant mode of utilising CAD software.

The accessibility provided by these graphical interface tools came at a cost. Previously, a
set of programming routines would be written especially for the forms that each specific
building required — however, the use of standardised tools severely limited the range of
operations available within CAD programs, and thus the possibilities of form-making.34
Furthermore, because these tools were designed to mimic conventional pen and paper
draughting concepts, as well as the traditional elements of architectural geometry, they
embodied highly entrenched architectural forms and methods.35 As a result of this, the
computer lost its ability to generate specialised forms on a project-by-project basis, and
thus became predominantly used as tool to represent preconceived forms, rather than a
means in which to design highly specialised forms.

This reliance upon standardised tools was evidenced in the emerging presence of complex
curvatures within the architecture of the 1990s. The popularity of such forms can be
directly traced back to CAD tools that allowed for the manipulation of spline curves and
surfaces . However, even though these tools could produce highly complex results, they
were still predominantly used as methods of computerisation in that they were a means
to represent or optimise a pre-imagined design.36 This was prominently demonstrated in
the design process of Frank Gehry, where his buildings were conceived through the use
of traditional modes — such as sketches and mock-ups — and the computer was used
only to test structural viability and produce technical documentation.

Another common use of CAD was to employ the software as a kind of improvised
sculpting environment to manipulate form according to the successive musings of the
designer. This remains, however, as a primarily computerised mode of operation because
it mandates tweaking a depiction, and is thus only a digitisation of a traditional modelmaking
process, such as shaping clay.

Faced with the rapidly-established formal clichés of blobitecture and the curvilinear,
many designers have recently returned to the initial mode of utilising CAD systems:
that of computation. By once more creating forms through code, they could escape
the conforming influence of codified procedures. In doing so they could create their
own tools and thus enter a ‘higher form of craftsmanship’ wherein the influence of the
software can be overcome.

ALGORITHMS AND ASSOCIATIVE DESIGN

At the most basic level, everything that a computer does is an interaction between
operations and information; between code and data. For example, when creating a 3
dimensional model within a computer program, what is being created as an end-result is
a list of data that contains x,y,z co-ordinates. In order to create or manipulate this data,
procedures need to be written in code that define geometries that can then be encoded as
co-ordinates. These logical procedures are known as algorithms, and are abstract logics
that have been translated into computer code and thus address a problem through a finite
number of steps. For example, in order to create a perspective view of a list of geometric
data, an algorithm is run whereby each of the Euclidian co-ordinates is ordered and
arranged to face the viewport, and then deformed to simulate the diminishing vanishing
point of perspective projection.

Although this process seems obvious and natural to anyone who has created perspective
projections on a draughting board, it is no more advanced or ‘natural’ to a computer than
many other methods, even those that would seem radically alien, such as perspective
projection where the vanishing lines curve toward an end point. This illustrates that,
within a computer, there is a distinct dichotomy between an internal representation of
forms as defined by algorithmically generated geometries, and an external representation
of this abstract geometric data into pictorial norms such as perspectives, elevations and
sections.

The key to ‘parametric architecture’ is that it is procedure- driven. It operates by engaging
the computer’s internal production of form rather than an external representation.
Because it translates the design intent of the architect directly into algorithms, it is not
reduced to working through typical CAD tools and procedures.

Because these algorithms are created on a project-by-project basis, they can be much more
readily adapted to the specific needs of each building. A key part of this is the ability for
algorithms to associatively link data sets by defining reciprocal relationships. For example,
the height of a column could be made to depend on the height of the ceiling, and
vice-versa. Thus, if the project required that the space was to become double-height, the
column would automatically increase to match. This is an basic example of an extremely
simple algorithm, and much more powerful uses can be produced when algorithms
integrate more complex networks of data inter-relationships and responsive environment
simulations. For example, algorithms could produce a facade pattern that optimises the
length and tilt of louvres in order to produce a more even spread of sunlight over the
course of specified time period. This algorithm could also then be linked into a series
of other algorithms that determine relationships between passive solar gain, privacy,
visibility, and insulation needs.

Through the use of associative relationships that are generated by algorithms, architecture
can be designed through the creation of a dynamic ecosystem of components that are
defined by reciprocal relationships that link across scales and contexts. The nature of
designing architecture through such associative relationships leads to a holistic design
understanding, and so the production of form within the computer is more concerned
with creating a project, rather than merely an object. This has drastically affected the
implications of designing within a computer environment. For example, through the use
of associative algorithms, criteria such as structural efficiency, or budget estimates can
be engaged with while the design process is still occurring, rather than after a design
intent has been fully crystallised.

In producing these associative algorithms there is a need to rationally formalise the
architectural project into the abstract logics of code. It requires, also, that design
problems and solutions are explicitly expressed, and thus are always apparent — albeit
encoded within the abstractions of code. As a result, design problems are articulated
across all scales, and thus produce a deeply nested set of rationalisations where all higher
level outcomes are consequences of lower-level properties. This process is radically
different to the traditional modes of design, as it is a process which demands that the
architecture is always defined from first principles, creating a “method of constituting
the architectural project in a long sequence of relationships from the first conceptual
hypotheses to the driving of the machines that prefabricate the components that will be
assembled on site.”37

GENETIC ALGORITHMS AND EMERGENCE

Because the traditionally dominant mode of discussing creativity is through intuitive
play and artistic sensitivity,38 the mechanistic determinism of an algorithm is seen as
a distant and anonymous process that is thus nonhuman — however, this perspective
needs to be rethought in the face of algorithms that escape the confines of performing
tedious procedures in which the criteria for success is predetermined.

When problems emerge in which the required solutions are unpredictable, unimaginable
or unconceivable, these more radical and sophisticated algorithms can be employed to
‘search out’ a solution. As with other algorithms, they rely on a programmatic definition
of environmental conditions and procedures. Where they differ is that they introduce
an element of randomness into these conditions, and assess a population of outcomes
in parallel — rather than merely following a singular procedure through to a singular
outcome. These ‘genetic algorithms’ are so called because they proceed through a series
of successive adaptations, whereby a variety of different options self-optimise and selfselect
towards a goal, and thus instrumentalise Darwinian natural selection.39 At their
most complex, genetic algorithms can generate vast sets of possible solutions that are
then selected for ‘fitness’ against predetermined success criteria. To further optimise,
many successive generations can be developed through further randomisation, and
through combining the results that were found to be most ‘fit’. Because the computer is
able to run these algorithms so easily, they can be used as a very efficient means to solve
design problems, in which large scale tests would not be possible in reality. For example,
given the material conditions of aluminium, and the process through which turbulence
operates, an optimised form of an aeroplane wing could be generated as a result from a
large variety of tests.

Although these algorithms are used to generate efficiency, the stochasticity of the
search process means that a solution can be found that was previously unconsidered.
For example, the most optimal outcome of the wing’s optimisation for performance in
turbulence could be a wing configuration that is swept forward, instead of swept back.
In this case, such a form may not have been preconceived as a possible solution, but
because of the element of controlled and directed chance, it can be produced through
procedures that are nonetheless mechanistic. These solutions are often said to be
‘emergent’ solutions in that they are higher-level properties that have been generated as
a consequence of an interaction between lower-level properties.40 This echoes the ‘blind
watchmaker’ approach of evolutionary biology whereby ‘dumb’ processes can often
generate solutions that are incredibly well-devised purely through the large time scales
within which evolution operates.

This structural behaviour resembles techniques such as the use of random words to
produce Dadaist poetry. Unlike such completely unpredictable processes, though, genetic
algorithms generate chaos within the confines of a guiding system which “produces
effects that, although unpredictable, are intrinsically connected through the rules that
govern that system.” Thus, they merge the extreme randomness of stochastic algorithms
with the extreme rationality of associative algorithms, and as a result, offer a unique
process of solving design solutions.

Genetic algorithms have a limitation, however, in that they can only optimise against
predetermined criteria for success (for example, criteria such as the estimated rental
income, the area of the rooms or the amount of sun coverage). Because each of these
criteria must be encoded into algorithms, there is a difficulty in crafting criteria to assess
subjective values such as aesthetics.

However, there is the possibility of adding human selectivity into the procedures of
genetic algorithms to address this shortcoming. For example, 50 variations of a facade
striping pattern can be produced, from which the designer selects outcomes that are
judged to be attractive. From this, alternative variations can be generated, or the selected
outcomes optimised further. This approach still allows for ‘emergent’ solutions, but also
involves a human element that can assess subjective criteria41. This produces a hybrid
gestalt that synthesises the best instances of human and machinic thinking. Moreover,
increasing computational power and incorporating more accessible interfaces into these
procedures offer the potential for this introduction of human intuition to be present
along the parametric-driven road to creation, thus creating a highly optimised synthesis
of the best, unique, qualities of each mode of thought.

SYNTHESIS

“The phase of transmodernity that we are now in is perhaps best characterised by the
use of computation still operating under the vestiges of the old paradigm. In other
words, architecture has still yet to incorporate the architecture of computation into the
computation of architecture.”42

In the first chapter, I established that Deleuze’s philosophy is defined by the operations
and conditions of the virtual, and that, therefore, an adoption of the Deleuzian Persona
must consciously engage this ontological system. By adopting these fundamental
conditions — and thus assuming the Deleuzian Persona — Deleuze’s creative process
can then be used to reconstruct his concepts, and in doing so understand how and why
these concepts were created. Moreover, the ability to assume this Persona enables nonphilosophers
to employ Deleuzian thought within their native disciplinary context.

Until recently, the discussion of Deleuze in architecture ignored the fundamental
conditions of his ontology, and instead focused on spatial models such as Smooth Spaces
and The Fold. Because of this, it did not fully adopted the Deleuzian Persona. Due to
this failure, it lost out on two fronts. Firstly, it lacked the ability to fully understand the
origins of Deleuze’s spatial models, and thus ruled out the possibility of creating a novel
architectural understanding of these concepts. Secondly, this shortsightedness meant
that the discussion of architecture became focused on aesthetic outcomes, rather than
the processes of creation. As such, Deleuzian theory in architecture merely served as a
guide and justification for the design of clichéd forms.

An analogous situation occurred in architecture’s utilisation of the computer. Because the
tools used to create form in software programs were limited to predefined process that
merely imitated traditional design modes, CAD was used to reproduce and represent
a preconceived design. In this way, the use of the computers in architecture became a
limited means to a limited end: a method of tweaking and optimised buildings that had
already been designed by traditional methods. Again, a final outcome was prioritised
over the underlying processes of creation, and so computer-generated architecture also
became defined by formal clichés. Because the formal clichés of computer-generated
architecture matched those of Deleuzian theory, these two fields became closely
intertwined.

When architects began to ignore the codified tools of CAD programs — and instead
began to use code to create their own tools — they adopted the Persona of computation.
This dramatically changed their design process. By encoding information and design
intent within crafted algorithmic procedures, the computer was moved beyond being
just a machine for the generation of form, and instead became a semi-independent
mode of design thought. In this way, the computer was no longer a means to implement
predefined concepts; but rather a partner in the design process because it possessed the
ability to be generate original outcomes.

Significantly, this shift in the Persona of computer-generated architecture triggered a
change in the paradigms of architectural theory. Deleuzian philosophy continued to be
referenced as the main source of inspiration, but the focus within Deleuzian philosophy
shifted. Concepts that had been used to discuss formal properties were passed over, and
new topics such as Manuel De Landa’s discussion of morphogenesis and Neil Spiller’s
discussion on machinic phylum rapidly become the dominant concerns. Because these
new concerns engage with the conditions of Deleuze’s ontology, they require that
Deleuze’s Persona be adopted, and thus represent a more fundamental re-engagement
of Deleuze within architectural theory.

This re-engagement reveals the fundamental conditions of compatibility between
Deleuze and computer-generated architecture: both are concerned with a reification of
the virtual. The virtual in philosophy and the virtual in computer systems are defined by
their encapsulation of the rules by which objects are realised. Neither the multiplicity, nor
the parametric, represent form, but rather represent the intrinsic instructions that generate
form.43 The procedural operations of a computer can thus be considered analogous to the
phase space of the virtual multiplicity because each defines abstract features — either
through the logic of algorithms, or through the shape of manifolds — that guide
actualisation.44

An understanding of this fundamental compatibility allows for a mode of design that
synthesises the Parametric and Deleuzian Personas. This synthesis, evident in the
work of many emerging architects, allows for an understanding and production of
architecture that maximises the possibilities of designing within virtual environments.
Given Moore’s Law, and the maturation of CAD, CAM and BIM systems, virtual
environments are increasingly seen45 as incredibly powerful and valuable. Because these
virtual environments offer a fountainhead of computation possibility, the primary factor
constraining their use is not technological, but the failure of architects to integrate this
computational power into their design process46. To address this shortcoming, architects
must fully engage with the design of virtual environments. Simultaneously adopting the
Deleuzian and parametric Personas enables architects to do this.

CONCLUSIONS

“…What is needed is a radicalisation of the prevailing paradigm of architecture, beyond
retroactive manifestos, by developing a new concept of architecture that is adequate to
the demands imposed by computation…”47

Deleuze has been the dominant mode of theorising computer-driven architecture.
Historically, this relationship between Deleuze and architecture has been flawed, as
evidenced by the fact that — before the dot-com boom — the modes of theory and
practice were both trapped in within high-order paradigms based upon the fetishisation
of forms. Deleuzian theory could only critique architecture in relation to preconceived
concepts, and architects could only design towards a preconceived design intent.

The recent second wave of digital design and Deleuzian theory has surpassed this
restriction by focusing on process over result, and thus has revealed a much more
fundamental compatibility between Deleuze and computer-generated architecture. This
correlation of the Deleuzian ontology to parametricism can be explained through the
shared and unprecedented reification of the virtual that is inherent to each Persona. In
Deleuze, this Persona is demonstrated in his analysis of reality as an actualisation of
immanent multiplicities. In architecture, this Persona is demonstrated in the creation of
buildings as an actualisation of designed algorithms.

The Deleuzian ontology allows for a rich understanding of reality as defined through
virtual systems, which is the very reality that architects are engaging with when designing
parametrically. Therefore it provides a theoretical impetus and means for architecture to
more consciously and productively engage with computation. An architectural synthesis
of the Deleuzian ontology with the parametric design process would best enable
architects to utilise the computer in a way that is fundamentally different from prior
paradigms.

BIBLIOGRAPGHY

• Andrasek, Alisa. “CONTINUUM: A Self-Engineering Creature-Culture.” Architectural Design 76, no.
  5 (October 2006): 18-25.
• Aranda, Benjamin, and Chris Lasch. Pamphlet Architecture 27: Tooling. 1st ed. Princeton Architectural
  Press, 2005.
• Ballantyne, Andrew. Deleuze and Guattari for Architects. London ;;New York: Routledge, 2007.
• Bessa, Maria. “Algorithmic Design.” Architectural Design 79, no. 1 (February 2009): 120-123.
• Bishop, Richard. Geometry of manifolds. Providence R.I.: AMS Chelsea Pub., 2001.
• Badiou, Alain. Deleuze: The Clamor of Being. University of Minnesota Press, 1999.
• Buchanan, Ian. Deleuze and space. Toronto: University of Toronto Press, 2005.
• Cache, Bernard. Earth Moves: The Furnishing of Territories. The MIT Press, 1995.
• Cache, Bernard, and Patrik Beauce. Objectile: Fast-wood: a brouillon project. New York: Springer, 2007.
• Cache, Bernard, Manuel De Landa, Sandra Knapp, Sanford Kwinter, Detlef Mertins, and Mark Wigley.
  Phylogenesis foa’s ark: foreign office architects. Illustrated edition. Actar, 2004.
• Chu, Karl. “Metaphysics of genetic architecture and computation.” Architectural Design 76, no. 4 (August
  2006): 38-45.
• Costanzo, Michele. “Twenty Years After (Deconstructivism): An Interview with Bernard Tschumi.”
  Architectural Design 79, no. 1 (February 2009): 24-29.
• Manuel, DeLanda, and Idhe Don. “1000 Years of War: CTHEORY Interview with Manuel De Landa,”
  January 5, 2003. http://ctheory.net/articles.aspx?id=383.
• De Landa, Manuel. A New Philosophy of Society: Assemblage Theory and Social Complexity. Annotated
  edition. Continuum International Publishing Group Ltd., 2006.
• De Landa, Manuel. A Thousand Years of Nonlinear History. Zone Books, 2000.
• De Landa, Manuel. “Diagrams and the genesis of form.” ANY – Architecture New York 1, no. 23 (1998).
• De Landa, Manuel. “Deleuze and the Genesis of Form,” Concordia University, Montreal, Canada, 1997.
  http://www.cddc.vt.edu/host/De Landa/pages/genesis.htm.
• De Landa, Manuel. “Deleuze and the Use of the Genetic Algorithm in Architecture,” Between
  Bladerunner and Mickey Mouse: New Architecture in Los Angeles Exhibition, Madrid, Spain,
  April 22, 2001.
• De Landa, Manuel. “Immanence & Transcendence in the Genesis of Form.” A Deleuzian Century? 96,
  no. 3 (Summer 1997): 499-514.
• De Landa, Manuel. Intensive Science & Virtual Philosophy. Continuum, 2005.
• De Landa, Manuel. “Material Expressivity.” Domus, no. 893 (June 2006): 122-23.
• De Landa, Manuel. “Materialism, Experience and Philosophy,” European Graduate School, Saas-Fee,
  Switzerland, September 6, 2009. http://www.youtube.com/view_play_list?p=38A848FA0C7479C3.
• De Landa, Manuel. “Smart Materials.” Domus, no. 894 (July 2006): 122-23.
• De Landa, Manuel. “The Emergence of Synthetic Reason,” October 20, 2009. http://www.t0.or.at/De
  Landa/De Landa.htm.
• De Landa, Manuel. “The Philosophy of Giles Deleuze,” European Graduate School, Saas-Fee,
  Switzerland, June 3, 2006. http://www.youtube.com/view_play_list?p=D649C765D91C1120.
• De Landa, Manuel, John Protevi, and Torkild Thanem. “Deleuzian Interrogations: A Conversation
  with Manuel De Landa, John Protevi and Torkild Thanem,” 2005. www.dif-ferance.org/De Landa-
  Protevi.pdf.
• Deleuze, Gilles. Difference and Repetition (Continuum Impacts). New Ed. Continuum International
  Publishing Group Ltd., 2004.
• Deleuze, Gilles. Pure Immanence: Essays on a Life. Zone Books, 2005.
• Deleuze, Gilles, and Felix Guattari. A Thousand Plateaus: Capitalism and Schizophrenia. University of
  Minnesota Press, 1987.
• Deleuze, Gilles. Anti-Oedipus. New Ed. Continuum International Publishing Group Ltd., 2004.
• Deleuze, Gilles. What is Philosophy? Verso Books, 1994.
• Eisenman, Peter, and Eisenman Architects.;Albert Speer & Partner.;Hanna/Olin Ltd.;Aedes Galerie
  für Architektur und Raum. Unfolding Frankfurt. Berlin: Ernst & Sohn, 1991.
• Eugene, Thacker. “Language, Life, Code.” Architectural Design, October 9, 2005.
• Frazer, John. An Evolutionary Architecture. Architectural Association Publications, 1995. http://www.
  amazon.com/Evolutionary-Architecture-Themes-John-Frazer/dp/1870890477.
• Fuller, Matthew. Media ecologies. MIT Press, 2005.
• Hensel, Michael. “Computing self-organisation: environmentally sensitive growth modelling.”
  Architectural Design 76, no. 2 (March 2006): 12-17.
• Kerslake, Christian. “The Vertigo of Philosophy: Deleuze and the Problem of Immanence,” October 20,
  2009. http://www.generation-online.org/p/fpdeleuze8.htm.
• Kwinter, Sanford. Architectures of time : toward a theory of the event in modernist culture. 1st ed. Cambridge
  Mass.: MIT Press, 2002.
• Kwinter, Sanford. Far from Equilibrium: Essays on Technology and Design Culture. Actar, 2008.
• Leach, Neil. “Digital Morphogenesis.” Architectural Design 79, no. 1 (February 2009): 32-37.
• Leach, Neil. “The Limits of Urban Simulation: An Interview with Manuel De Landa.” Architectural
  Design 79, no. 4 (August 2009): 50-56.
• Lynn, Greg. Folding in architecture. Rev. ed. Chichester West Sussex ;;Hoboken NJ: Wiley-Academy,
  2004.
• Lynn, Greg, and Mark Rappolt. Greg Lynn Form. Rizzoli, 2008.
• Manaugh, Geoff. The BLDGBLOG Book. Chronicle Books, 2009.
• McCullough, Malcolm. “20 years of scripted space.” Architectural Design 76, no. 4 (August 2006): 12-15.
• Parr, Adrian. The Deleuze Dictionary. Edinburgh University Press, 2005.
• Rajchman, John. Constructions. The MIT Press, 1998.
• Rajchman, John. The Deleuze Connections. The MIT Press, 2000.
• Rea, Ceb. “Process/drawing.” Architectural Design 76, no. 4 (August 2006): 26-33.
• Reiser, Jesse. Atlas of Novel Tectonics. 1st ed. Princeton Architectural Press, 2006.
• Rocker, Ingeborg. “Calculus-based form: an interview with Greg Lynn.” Architectural Design 76, no. 4
  (August 2006): 88-95.
• Schumacher, Patrik. “Parametricism: A New Global Style for Architecture and Urban Design.”
  Architectural Design 79, no. 4 (August 2009): 14-23.
• Schumacher, Patrik. “The future is parametric,” September 19, 2008. http://www.bdonline.co.uk/story.
  asp?storycode=3122853.
• Silver, Mike. Programming Cultures: Architecture, Art and Science in the Age of Software Development.
  Academy Press, 2006. http://www.amazon.com/Programming-Cultures-Architecture-Development-
  Architectural/dp/0470025859/ref=sr_1_1?ie=UTF8&s=books&qid=1253763410&sr=1-1.
• Silver, Mike. “Towards a programming culture in the design arts.” Architectural Design 76, no. 4 (August
  2006): 5-11.
• Terzidis, Kostas. Algorithmic architecture. 1st ed. Oxford ;;Burlington MA: Architectural Press, 2006.
• Terzidis, Kostas. Expressive form : a conceptual approach to computational design. London ;;New York: Spon
  Press, 2003.
• Trummer, Peter. “Morphogenetic Urbanism.” Architectural Design 79, no. 4 (August 2009): 64-67.
• Tschumi, Bernard, and Irene Cheng. The State of Architecture at the Beginning of the 21st Century. Monacelli,
  2004.
• Ulysses, Ronquillo. “On De Landa’s Intensive Science and Virtual Philosophy.” http://mediaecologiesresonate.
  wordpress.com/2008/11/24/on-De Landas-intensive-science-and-virtual-philosophy/.
• Weinstock, Michael. “Advanced simulation in design.” Architectural Design 76, no. 2 (March 2006):
  54-59.
• Weinstock, Michael. “Self-organisation and material constructions.” Architectural Design 76, no. 2
  (March 2006): 34-41.
• Wiscombe, Tom. “Emergent Processes,” October 20, 2009. emergentarchitecture.com/pdfs/OZJournal.
  pdf.
• Zizek, Slavoj. Organs without Bodies: On Deleuze and Consequences. 1st ed. Routledge, 2003.