harding (2008) architectural systems
TRANSCRIPT
architectural SYSTEMS
An Essay based on the initial text “Cybernetics and the Mangle” by Andrew Pickering
John Harding
MSc. Architecture, Computing & Design, CECA, May 2008
2
“.... Cybernetics, and especially the work of the English cyberneticians – is all about this shift from epistemology
to ontology, from representation to performativity, agency and emergence, not in the analysis of science but
the within the body of science itself.”1
“"Understandably, mathematical difficulties become prohibitive in the case of three- or multi-compartment
systems.”2
Introduction
This essay looks at two things. First, a short history of cybernetic thinking, in particular English cybernetics
with respect to Andrew Pickering’s 2002 paper ‘Cybernetics and the Mangle.’ Secondly, a look at some of
the architectural applications of this approach to science & design, with particular attention to the work of
Cedric Price.
The ‘Mangle’
In his 1995 book, The Mangle of Practice: Time Agency and Science., Pickering pointed out that the ‘classical’
way of doing 20th century science was purely representational- i.e. finding mathematical ways of
classifying things that already existed. This approach to science is closely related to epistemology, the
accumulation of more and more collective human knowledge of things, a representational idiom.. Some
examples he sighted were the recent progress in particle physics, and the traditional view of biology being
a long list of taxonomies and classifications. The limitations of this approach to science are also exposed in
the book, for example the inability to find a solution to Poincaré’s 3-body problem... multi-agent complex
systems just cannot not be solved analytically.
In his view however, machines, facts, theories, mathematical structures and humans are in constantly
shifting relationships with one another—"mangled" together in unforeseeable ways that are shaped by
the forces of culture, time, and place – this should concern scientific practice just as much as anything else.
1 Pickering, A. The Mangle of Practice: Time, Agency and Science, 1995 2 Bertalanffy, Ludwig von., General Systems Theory, 1968, P.19. The classic example is the 3 (or n>3) Body Problem in Physics.
Diagram of a classic cybernetic system -
Ashby’s homeostat showing feedback
between subsystems A & B.
(source: Ashby, 1957)
3 bodies exerting gravity on each other
cannot be ‘solved’ in a classical sense, even
though the system is deterministic.
3
Science, Pickering argues, should be thought of as being alive; a system of constant progression of
interaction between human and environment, a dynamic interplay of constant change. One can draw
parallels with Zuse’s ‘Calculating Space’ thesis of 19693 with regards to an cellular automaton theory of the
universe – a theory performance, of doing.... an ontological standpoint as opposed to a strictly
epistemological one.
It was therefore in his 2002 paper Cybernetics and the Mangle that Pickering writes about his self-discovery
that this philosophy of Science had been approached 50 years ago by a group of English cyberneticians,
William Ross Ashby, Stafford Beer & Gordon Pask – and he goes on to say how their approach can inspire a
science not of representation, but one of performance.
Cybernetics
Cybernetics is a theory of control systems based on communication (information transfer) between
system & environment , and control (using feedback) of the system’s function in regard to environment.4
Most of these ‘systems’ are interesting to cyberneticians when there is a display of stability, not necessarily
in the individual components, but in the system organisation as a whole.
The field should also be seen as antidisciplinary in the sense that many cybernetic models can be applied
in various scientific fields – indeed, its origins are from the fact that same models were cropping up in
disparate areas of study. The system is therefore abstracted from a physical assembly and studied in its
own right. 5 With regards to the human ‘brain’ for example, the system can be ‘seen’ as a brain, but actual
physical things like neurons are meaningless to the cybernetician.
The first known example of a cybernetic system was the so called ‘governor’ by James Watt, analysed as a
system by James Clerk Maxwell in 1865. This device utilised the principle of negative feedback to exert
control on a steam supply to power an engine at a constant rate, using information from the environment
(in this case the current speed of the engine).
An analogous system to this is the common thermostat, which uses information from a given room in
order to control the output of heat energy into it. In this case the decision rule as it is called is the required
3 Zuse, K. Rechnender Raum, 1969.
4 Bertalanffy, Ludwig von., General Systems Theory, 1968, P.19
5 Pask, G. An Approach to Cybernetics, 1961, P.15
diagram showing controller ‘A’ subsytem with
input & output relationship with the
environment.
Zuse saw the universe as a vast closed system,
an interconnected lattice of cells with inputs
and outputs - deterministic but infinitely
complex.
(source: Zuse, 1969)
4
temperature, and this requires external input (by a human in the case of the thermostat). By removing the
feedback from the environment, the controller is said to dominate the environment.
The name ‘Cybernetics’ was given by Wiener (1948) in his book of the same name, it’s literal translation
being ‘steersman.’ Others such as Von Neumann, Walter, McCulloch, Pitts and Shannon for example were
crucial to the foundations of cybernetics, however Pickering concentrates more on the English
cyberneticians due to their relationship to ontological thinking being more relevant to his ‘Mangle.’ Of
particular relevance is a device called the homeostat, conceived in the late 1940s by Mr. William Ross
Ashby.
Ashby & Ultrastability
Ross Ashby was an English pschyciatrist & cyberneticist, mostly known for his ‘Law of Requisite Variety,’ a
important cybernetic concept which I shall not go into detail here, but roughly states that in order for a
system to be stable, the number of control states must be greater than or equal to the number of states
being controlled.6
Ashby’s other famous cybernetic work was the Homeostat; an assemblage of self-similar machines with
electrical inputs and outputs and a random number source, the inspiration for which drawn from his
interest in the human brain and it’s self-regulation.
On their own the units are useless, but linked together they displayed behaviour that although was
initially chaotic and impossible to predict in advance, settled into a stable system over the course of time.
The system would randomly reconfigure itself without any external observer telling it to do so. It could
also take a knock from the outside world, and still settle into a stable state. It was this resistance to not
being anything other than stable that Ashby termed Ultrastability. This is the opposite of an unstable
system – a good example being trying to balance a pencil on its end.
In Pickering’s text, he goes as far as agreeing with Wiener’s description of the homeostat being ‘one of the
great philosophical contributions of the present day.’7 High praise indeed!
6 For a full description see Ashby, R. 1957. (p.134)
7 Pickering, A. Cybernetics & the Mangle, 2002. (p.5)
Ashby’s sketch showing the setup of a
single homeostat device
(source: Ross Ashby Digital Archive)
2 variable system in phase space:
(i) stable sink (system falls into
equilibrium point if inside dotted region).
(ii) unstable source – system diverges
from current state.
(source: Pask, 1961)
5
The homeostat can also be seen as one of the first machine like analogues of the neural network, first
studied by Walter Pitts & Warren McCulloch in the 1950s, where the output of a neuron is dependant on
the inputs of other neurons. The feedback is cyclic in the sense that output information from a single
homeostat does have an effect on the input to the same device. In effect, the tripping of a relay in the
homeostat is similar to the firing of a neuron in a brain.
This ultrastability is related to the study of homeostasis8, where changes in for example body temperature,
are regulated over long time scales. It is this ultrastability too that I think has broad appeal in its
applicability to architecture - in the face of changing environments and human users with ephemeral
goals, the long-term stability of an architectural system (between humans and their built environment)
can be seen as a measure of good design.
Social Systems
Pickering also refers to Stafford Beer9, another English cyberneticist, in particular his investigations
regarding the application of cybernetics to social studies. In one of Beer’s early projects for example, he
looked at children’s decision making in the light of feedback, and how they began to learn which
selections led to pleasure and which to pain. The children were adapting or self-organising their brains to
secure rewards in the long term, much like the homeostat was reconfiguring itself to a stable state, but
with no pre-programmed objective to do so.
Of other notable interest is his Viable System Model and it’s near application to the peoples of Chile in 1971.
Beer’s idea was that organisations needed homeostatic controllers in order to regulate unpredictable
inputs, much like the homeostat. Beer saw society as a giant brain, an exceedingly complex system and
hence the Viable System Model was modelled on the human nervous system with all kinds of filtering
devices, redundancy & feedback loops.
His work was interesting, because it can be seen as one of the first applications of cybernetics in social
control, and hence of interest to architects and controllers of the built environment – but it still treated
people as participants only in the sense that they were components of companies, for example individual
human emotions and the suchlike were not considered. It was the finance and productivity of companies
where the control was to be regulated based on dynamic information feedback.
8 As described by Cannon, W.B. Wisdom of the Body, London, 1932.
9 See Pickering, A. Cybernetics and the Mangle, 2002 (p.12)
A single homeostat unit amongst its
conversational colleagues.
(source: Ashby, 1957)
6
New Cybernetics In the 1970s, Second Order Cybernetics, or New cybernetics as Gordon Pask called it grew in popularity.
This definition represented a shift from Wiener’s machine & physics based approach to a more biological
one – how in the face of increasing entropy, can organisms arrive at a stable organisation? New
Cybernetics also included the observer into the mix... the observer no longer just observes, but partakes
and exerts influence on the system and the system exerts influence on it– a participant observer10. The
observer was therefore a sub-system within his or her own right with inputs and outputs, parameters and
variables, and humans were thought of as being Black boxes, ready to be explored cybernetically.11
This version of cybernetics was explored from a theoretical standpoint in biology by two Chilean
biologists, Humberto Maturana and Francisco J. Varela, culminating in their theory of Autopoesis, in
particular its application in work concerning the animal cell. This work concerns the dynamics of living
systems (such as humans). In autopoesis, the environment itself is part of a closed system that is coupled
with the organism (itself a subsystem). There are no goals or objectives set from the outside, living systems
just live because they are stable survivors co-evolving with their environment and whose only products
are themselves.
This inter-relationship between a living system & its environment, co-evolving with each other over time is
called ‘Structural Coupling.’ Maturana & Varela see this relationship as a definition of ‘cognition’ in living
systems – be that machine, man, or anything else displaying the same behaivours.
It is this Second order cybernetics draws parallels to architecture. Our subject matter is exactly this, man’s
interaction with the built environment he conceives can thus be seen as a kind of structural coupling.
Autonomous systems are not individual humans, but a posthuman coupling between people and their
environment.
10
Pask, G. An Approach to Cybernetics, 1961. (p.35) 11
Black Box in the sense that they themselves could not be ‘observed’ themselves as a system. For example, at present at least I can ever know
exactly what is going on in your head!
Diagram showing how structural
coupling between a “living system” and
it’s environment modifies through time
via communication channels.
(www.cs.ucl.ac.uk/staff/t.quick/autopoies
is.html)
Abstraction showing the inner processes
of a cell. Note the communication
channels crossing the boundary.
(source: Maturana & Varela, 1973)
7
Architectural Systems
“The role of the architect here, I think, is not so much to design a building or city as to catalyse them; to act that
they may evolve. That is the secret of the great architect”12
The antidisciplinary approach that cybernetics offers finds its natural partner in architecture– at traditional
mix of arts and science, and by and large from modernism onwards, a field open to cross-pollination from
many different fields. This combination along with an attitude of cultural optimism was most reflected in
architectural design and the arts during the 1960s. The traces of modernism & direct social control were
fading away in the face of post-war Europe. The machines for living in became dynamic machines to
interact with.
This attitude towards a more dynamic architecture was reflected in the attitude of certain design
movements of the time: Archigram, for example with the Sin Centre project by Michael Webb; the
Situationists with the New Babylon project for a city by Constant Nieuwenhuys,13 & the philosophies of
design pursued by Buckminster Fuller to name a few.
But perhaps the best example to talk about here is the unrealised project by Cedric Price, first conceived in
the early 1960s called the Fun Palace – A project that if realised, would have been one of the greatest
architectural experiments of its time.
Fun Palace
The Fun Palace was a fully flexible space, designed without any specific brief as such other than to house,
inspire and respond to any creative activity the users of the space saw fit. The project was conceived at the
beginning of the 1960s as a joint collaboration between Price and Joan Littlewood, a member of the
Theatre Union with a joint ambition to change the way architecture could combine dynamically with the
creative arts it traditionally housed.
12 Frazer, J. An Evolutionary Architecture, 1995, (p.7. Gordon Pask’s forward) 13 A lecture by Nieuwenhuys on the New Babylon project was attended by Price, and no doubt influenced the Fun Palace to a degree.
Isometric of Fun Palace, 1964.
(source: google images)
Detail of the New Babylon Project,
Constant Nieuwenhuys,1962.
(source: google images)
8
Thus, the concept behind the Fun Palace was the antithesis of the static building, designed merely as
shelter. In Price’s opinion, the architecture should be part of the theatre it housed – part of the ongoing
dance between man and his environment. As he said, it was his first true ‘antibuilding’ in that there was no
way even the architect would be able to predict how the system would precisely function,only have a
good enough idea that system could function.
The Fun Palace space was to be surrounded by a spaceframe for supporting the movable units. That the
spaceframe was architecturally sparse was the key- it’s completely homogenous nature was what kept the
Fun Palace so adaptive to change and fully flexible. In effect, the whiter the canvas, the more possibilities
on offer.
Although Price was the main architect on the job, the project was so collaborative it is hard to say who
designed what in the end – the concept of the author architect was forgotten, indeed Price referred to
himself as an anti-architect. Price drew in consultants from all creative disciplines, including cybernetics.
The fact that the architecture was heavily system based in its dynamic nature lent itself nicely to
cybernetic input. The Fun Palace would have to self-regulate, and its physical configuration and
operations would need to anticipate and respond to probable patterns of use.14 Price realised managing
this kind of complexity was the real challenge and duly convinced Gordon Pask to become involved.
Cybernetics & The Palace
Born in Derby, 1928, Gordon Pask’s first involvement in cybernetics came while at Cambridge in a theatre
group. With his friend Robin McKinnonwood, Pask’s interest in technology combined with his own
theatrical connections resulted in a succession of interesting machines – for example a Musical typewriter,
a self-adapting metronome, and the Musicolour machine (see later). This background meant there was
always a continuing theatrical theme in his work.15
This theatrical background no doubt helped when Price approached Pask in 1963 to offer him the
poisiton: ‘cybernetic consultant’ on the Fun Palace Project. Pask got to work, setting up the Fun Palace
cybernetics sub-committee. This committee began to develop psychological & sociological models,
14
Matthews, S. From Agit Pop to Free Space: The Architecture of Cedric Price,2007 (p.73). 15
Pask’s ‘Colloquy of Mobiles’ for example as described in Pickering (2002) which involves an ongoing theatrical dance between male and female
balloon-shaped objects.
Pask’s diagram of the cybernetic control
system of Fun Palace.
(source: Matthews, 2007)
Fun Palace early plan detail, 1962
showing movable partitions.
(source: Matthes, 2007)
9
thinking about the possible activities that might occur as to have some idea at least of possible
connections and associations that could be formed. These activities ranged from archery to ‘restoration of
vintage cars,’ from drama tutoring to finger painting. The Palace had come a long way from its theatre
origins.
The Fun Palace was seen as a vast social control system as can be seen on Pask’s system network diagram
(see page 8 images). This attitude of abstracting people as modified or unmodified ‘systems’ at first glance
seems like it should be treated with caution, Pasks language however merely referred to whether the user
had modified the system previously, and hence in a kind of architectural conversation, had been modified
by it.
Here one can see a comparison to Beer’s Viable System Model as mentioned previously, as a model of
social control but with underspecified goals, with the architecture working in tandem with it’s users. The
difference here was that the users themselves were systems to be altered – and Pask refers to the Fun
Palace as an ongoing conversation between the building and its users - an assemblage of interactive
systems of interaction.16
Generator & Musicolour
Another of Price’s works of cybernetic relevance was the so called Generator Project from 1976. With the
help of computational input from John Frazer, the Generator was a series of relocatable components , a kit
of parts that could create enclosed spaces, corridors, screens, in different configurations depending on the
requirements of a specific time. Much like the Fun Palace, there would be a mobile crane to move the
parts around when these changes were implemented... the users of the space became the controllers of it.
What was interesting was that instead of an abstract diagram acting from outside like the Fun Palace, the
components became the hardware itself, with processing devices in each. In effect, the architecture was a
massively parallel computer- and there was no need for any centralised control. As Price mentioned,
architecture should have little to do with problem solving – rather it should create desirable conditions
and opportunities hitherto thought impossible.
One key element that Frazer introduced to the project was boredom. Afraid that the users wouldn’t
reconfigure the system as desired, the ‘building’ would subsequently get bored and suggest
16 Pask, G., The Architectural Relevance of Cybernetics, Architectural Design, September 1969. (p.494)
Sketch by Price showing possible
configuration of units. (source: google images)
John Frazer’s generator realisation in
hardware. (source: Frazer, 1995)
10
reconfigurations itself having learnt previously from the users possible layouts. It also learnt from the
success of its own layouts. It was a truly intelligent building, however the system was somewhat closed in
the sense that no site specific input, the original site was in Florida, was allocated for. I guess this was the
point, that the people would shape the architecture to work with the local condition.
It is probably no coincidence that this element of boredom was also prevalent in one of Pask’s early works,
the ‘Musicolour Machine’ from the 1950s. This machine was to be used in tandem with a human musician
by using a microphone input for voice and an electrical input from an instrument. A musicolour
performance centred on a feedback loop running from the human performer through a musical
instrument and the machine itself into the environment (lights displayed to the performer), and then back
to the performer.17
Again, there is no direct ‘goal’ as such, no problem to solve other than to avoid repetition that perhaps the
human on his own would not realise. The music created could be seen as not being possible by the human
alone, rather a conversation between human and machine resulting in a unique ‘gestalt’ performance.
Summary
To conclude, I would like to link some of our work this year to this short introduction to cybernetics, in
particular, the computing of parallel systems within the study of cellular automata.
This approach of distributed computing and architecture was recently explored in our urban coding work.
Plots of land were given certain rules that society was to obey. The long-term behaviour of the system
could not be known before hand – no amount of scientific knowledge could predict what was to happen.
The CA just had to be run on the computer in order to ascertain the long-term stability of the system –
whether it survives in a stable condition or whether it dies out altogether. Factors such as the initial
conditions of the system could also be investigated. Real-time perturbations and their resultant behaviour
was explored, much like with the homeostat. This was architectural system design – a cybernetic approach
of doing rather than representing.
In my opinion, this example nicely shows that cybernetic models that have already been studied in their
own right (such as CAs) have a lot to offer architects, especially when there is a wealth of the scientific
research already completed. However, it should always be remembered that it is their careful application
17
Pickering, A. Cybernetics and the Mangle, 2002. (p.16)
Cellular Automaton of an urban condition.
Plots are given a ‘wealth’ and rules governing
the distribution of wealth to others are
specified.
Th e long term behaviour of the system can
only be known by actually running it.
Pask attempts to get a sound out of the
Musicolour machine. (source: google images)
11
that requires skill, not simply a direct mimicking of scientific systems. This shift in architecture from
problem solving a fixed brief, to creating a dynamic systems that adapt themselves to a fast changing
world would definitely find favour with Pickering for sure.
Bibliography Ashby, R.W. An Introduction to Cybernetics, Chapman & Hall, 1957.
Bertalanffy, L. von. General System Theory, Penguin University Books, 1968.
Frazer, John. An Evolutionary Architecture, 1995.
Haque, U. Gordon Pask & Architecture, 2007.
Matthews, Stanley. From Agit-Prop to Free Space: The Architecture of Cedric Price, Black Dog Publishing, 2007.
Maturana, R.H. & Varela, F.J. Autopoiesis, The Organisation of the Living, 1973.
Maturana, R.H. Biology of Language: The Epistemology of Reality, 1978.
Pask, G. An Approach to Cybernetics, Hutchinson & Co, London, 1961.
Pask, G. The Architectural Relevance of Cybernetics, Architectural Design, September 1969.
Pickering, Andrew. Cybernetics and the Mangle: Ashby, Beer & Pask (Social Studies of Science, Issue 32), 2002.
Pickering, Andrew. The Mangle of Practice: Time, Agency & Science, University of Chicago Press, 1995.
Price, Cedric. Works II, Architectural Association, 1984.
Wiener, N. Cybernetics: or Control and Communication in the Animal and the Machine, MIT Press, 1948.
Zuse, K. Rechnender Raum (Calculating Space), 1969.
The Ross Ashby Digital Archive: “http://www.rossashby.info” (front cover image from google image search)
Pask in conversation. (source: google)