The Architecture

of Information
Despite its potential to break the mould, digital information has been character-
ised by its reliance on metaphors from a pre-digital era. Architectural ideas have
pervaded discussions of digital information, from the urbanisation of cyberspace
in science fiction through to the adoption of spatial visualisations in the design
of graphical user interfaces. Are these the equivalent of the car’s ‘horseless car-
riage’ phase or are they a pointer to a more fundamental relationship between
human beings and their representations of information?
Architectural, philosophical, psychological and historical knowledge
are united in this book to develop an understanding of the relationship between
information and its representation in a post-digital era. From the development
of Memory Palaces to the modern library, buildings have acted as classification
devices by associating the arrangement of ideas with the organisation of physical
objects. This tradition – knowledge made manifest through the articulation of
architectural space – has been challenged by the development of digital technolo-
gies which separate information from its material representation.
This book tackles:

• the historical importance of physical places to the organisation and expres-

sion of knowledge
• the limitations of using the organisation of objects as the basis for systems
of categorisation and taxonomy
• the emergence of digital technologies and the new conceptual understand-
ings of knowledge and its organisation
• the concept of disconnecting the storage of information objects from their
presentation and retrieval
• ideas surrounding ‘semantic space’
• the potential of hypertexts versus the realities of the types of user interface
which now dominate modern computing.

Posing the question ‘what sort of space is information space?’, the book exam-
ines the motivations behind the perceived need to disguise the complexity of
digitally encoded information with metaphors of physical spaces and architec-
ture. The conclusion of the book looks for a better understanding of information
architecture, defining a new design domain for the practice of architecture as it
relates to the complexity of digitally organised information.

Martyn Dade-Robertson is Lecturer in Architecture and Communication in the

School of Architecture, Planning and Landscape at Newcastle University. He
originally did a degree in Architecture at Newcastle University before embarking
on an MPhil and PhD at Cambridge University (Darwin College) on the topic of
Information Architecture.
The Architecture
of Information
Architecture, interaction design
and the patterning of digital
information

Martyn Dade-Robertson
First published 2011
by Routledge
2 Park Square, Milton Park, Abingdon, Oxon OX14 4RN

Simultaneously published in the USA and Canada

by Routledge
711 Third Avenue, New York, NY 10017

Routledge is an imprint of the Taylor & Francis Group, an informa business

© 2011 Martyn Dade-Robertson

The right of Martyn Dade-Robertson to be identified as author of this work

has been asserted by him in accordance with sections 77 and 78 of the
Copyright, Designs and Patents Act 1988.

All rights reserved. No part of this book may be reprinted or reproduced or

utilised in any form or by any electronic, mechanical, or other means, now
known or hereafter invented, including photocopying and recording, or in any
information storage or retrieval system, without permission in writing from
the publishers.

Trademark notice: Product or corporate names may be trademarks or

registered trademarks, and are used only for identification and explanation
without intent to infringe.

British Library Cataloguing in Publication Data

A catalogue record for this book is available from the British Library

Library of Congress Cataloging-in-Publication Data

Dade-Robertson, Martyn, 1979-
The architecture of information : architecture, interaction design, and the
patterning of digital information / Martyn Dade-Robertson.
p. cm.
Includes bibliographical references and index.
1. Communication in architecture. 2. Space (Architecture) 3. Information
organization. I. Title. II. Title: Architecture, interaction design, and the
patterning of digital information. NA2584.D33 2011
720.1—dc22 2010042922

ISBN13: 978–0-415–56183–9 (hbk)

ISBN13: 978–0-415–56184–6 (pbk)
ISBN13: 978–0-203–82838–0 (ebk)

Typeset in Univers by
Pindar NZ, Auckland, New Zealand
To my dad
Contents

Preface ix

Acknowledgements xi

Illustration credits xiii

Introduction 1

Part 1: The architecture of everything 1

Part 2: Changing patterns of architecture and information 7
Part 3: Scope and focus 16
Part 4: The place of architectural theory 21
Part 5: The multi-disciplinary context 29
Part 6: Structure and content 34

1 The architectonic system 35

Part 1: Architecture and meaning 35
Part 2: Architectonic systems and the emergence of
categorisation 38
Part 3: Conceptual spaces: categories in the mind 46
Part 4: Topic and topos 57
Part 5: Conclusion 63

2 Between city lights receding and the non-space of the mind 67

Part 1: Loosing the ballast of materiality 67
Part 2: Spaces between logic and ontology 71
Part 3: Display becomes reality 79
Part 4: The web is not a tree 82
Part 5: Remediating space 87
Part 6: Conclusion 89

3 The spaces of information 93

Part 1: Spaces of information 93
Part 2: The failure of architectural metaphors 96
Part 3: Measuring – semantic space 101
Part 4: Mapping – screen space 107
Part 5: Exploring – interaction space 110
Part 6: Navigating – architectonic space 115
Part 7: Conclusion 120

4 Reality becomes display 123

Part 1: After information architecture 123
Contents

Part 2: Non-discursive formations 131

Part 3: Place–action 134
Part 4: Space agency 138
Part 5: Invisible architectures 141
Part 6: Conclusion 143

Conclusion 147

Glossary 153

Notes 161

Bibliography 165

Index 173

viii
Preface

A thought experiment, which I have started using with my postgraduate stu-

dents, is to ask them to imagine a parallel universe containing a society at a
similar stage in its technological development to our own but with one key
difference: that the idea of architecture hasn’t been invented yet. Buildings still
exist, along with a plethora of other designed artefacts, but there is no distinc-
tion between design disciplines. A bright design theorist decides that this lack of
distinction between the design of different artefacts is a problem and endeavours
to categorize design into coherent disciplines based on criteria he must define.
I then ask my students to imagine, free from their knowledge of how the world
actually is, what categories he or she might come up with.
Would it be possible to define design in relation to the materials
from which various artefacts are created? Wooditecture, steelytecture, plas-
ticitecture …? Or through geometric similarity between designed artifacts?
triangletecture, rectangletecture …? It’s a silly game but allows my students to
do three things.

1 They stop thinking about the design of the built environment as something
independent from all other design disciplines since many of their fictitious
categories define groups of artefacts which include, but are not limited to,
buildings. Wooditecture for example includes chairs, pencils and log cabins.
2 By thinking in terms of categories which are invented by them and are out-
side their everyday experience, it reveals how even logical categorisations
can be alien when viewed from a different perspective. The notion of a
wooditecture seems intuitively wrong but is not ridiculous given that there
are common methods for working and constructing with wood.
3 When the students return to our universe, they are able to look critically at
our own categorization of the design of artefacts and to assess where titles
such as ‘architecture’ seem logical and where they seem arbitrary.

The division of design disciplines is useful but it can also be limiting. I graduated
with a degree in architecture in 2000 with a nagging doubt. I had enjoyed my
three years and was about to embark on my year in practice to be followed by a
further two years of study and another year out before I would be able to receive
full RIBA accreditation. Despite the extensive nature of an architect’s education,
however, I still felt that I was missing something. During the late nineties I had,
as an architecture student, seen territory that I felt belonged to me as an architec-
tural designer, captured by the World Wide Web. I felt instinctively that there was
Preface

something architectural about buying books online, navigating a news website or

searching for files and folders on my computer’s desktop graphical user interface.
Furthermore, terms and ideas were being stolen from my chosen profession as
a new breed of ‘information architects’ arrived, citing cannons of architectural
theory such as Kevin Lynch and Christopher Alexander. Despite this, we were
not being taught web programming in our technology lectures or discussing the
history of computing as part of our ‘history of the designed landscape’ courses.
This intuition, that there was more to architecture than the built environment,
continued through my Masters and PhD. Studies and the results of my search
for a broader definition of architecture in the digital age is charted in this text.
This book is narrated through classifications. In the first instance it
is about the classification of architecture itself as something which has been
applied, and has the potential to be applied much further, outside the design of
the built environment. I don’t hold with, what seems to me to be, the parochial
distinction between architecture and building as being exemplified by the dif-
ference between a cathedral and a bicycle shed. Rather, I see architecture as a
design practice defined by the creation of objects of a certain scale in relation to
the human body. Any artefact that surrounds us and through which we move has
the potential to be architectural (independent of whether an architect has had a
hand in its design). The structuring of our environment into patterns which are
perceived as we move through them is the central role of architectural design
and this, in an age of digital environments, is extended to include environments
that are virtual rather than physical.
In the second instance it is about classification as a subject in its
own right. In other words how we structure our world in relation to groups of
objects or concepts in real or conceptual spaces. In particular I will focus on the
role of architecture in articulating categories through the creation of patterns in
our environment. By understanding classification as a patterning activity, light
can be shed on a whole range of digital artefacts which might not seem to have
much to do with architecture at all, but are central to how we communicate.
It will be up to the reader to decide whether I have made a case for
an ‘architecture of information’. In fact, while I have used the idea of architecture
to analyze a range of digital artefacts, we are, with digital technologies, living in
a similar place to my parallel universe. The lens of architecture allows me to cut
across a range of ideas and designed artefacts and to understand a commonality
that has nothing to do with the way they are categorized by their professional and
disciplinary boundaries, at a time when a new generation of designed objects is
emerging that have yet to find a framing discipline of their own.

Martyn Dade-Robertson
Newcastle, September 2010

x
Acknowledgements

For musicians there is the difficult second album and for academics there is the
difficult first book (although having only written a first book I can only assume
that the second book is going to be easier). The ideas and research for this book
started in the year 2000 as I completed my architecture degree and faced the
challenge of what to do next. I had the gut feeling that I had missed something in
my education and the research project outlined here started with some notional
ideas I had in the year following my degree and which took form through my
Masters degree, PhD and, ultimately, the start of my academic career.
This book was started, conceptually if not materially, at Newcastle
University while I was doing the BA in Architectural Studies and it seems fitting
that the book should also have been finished at Newcastle University, albeit now
as a member of staff in the department where I did my first degree. Suffice it to
say that I am grateful to the staff and students I have encountered during both
my spells at Newcastle, with a special mention to Stephen Kite and Di Leach,
who both guided me on my current academic path and encouraged the more
experimental aspects of my design practice.
Most of this work took shape at Cambridge University while I was a
student on the MPhil in Architecture and the Moving Image and ultimately as a
PhD student in what was called the Cambridge University Moving Image Studio
(CUMIS) at 1 Bene’t Place. The group has now lost its building and its name,
but its alumni have spread far and wide and my time there has left an indelible
impression on this book. Special mention should go to my supervisors and the
directors of CUMIS, Francois Penz and Maureen Thomas, who created some-
thing special in Cambridge, the effects of which are still to be fully realized. In
addition, I’d like to thank Alan Blackwell, who acted as an unofficial co-supervisor
and provided the much needed link to computing and Human–computer interac-
tion. I also want to thank the staff and students of Darwin College who became
my Cambridge family for the best part of five years and to thank CRASSH (Centre
for Research into Social Sciences and Humanities) and Ludmilla Jordinova and
the interdisciplinary reading group for opening my eyes.
While at Cambridge I undertook an internship at Microsoft Research
and this opportunity has turned out to be critical in the development of some
of the ideas in this book and my future career trajectory. The work I did there
still awaits proper publication, but I want to make special mention of Ralph
Sommerer for his patience and the development work he did on our attempts
at new information visualizations, and Ken Wood for giving me the opportunity
to work at Microsoft in the first place.
Acknowledgements

Upon completing the PhD, my thesis was examined by Mauri

Kaipainen. I will forever be grateful to him for his support during my PhD and for
his part in an extraordinary PhD exam, which did what a good PhD examination
should do and opened up research horizons beyond the thesis. My pursuit of
cognitive psychology and embodiment in particular owes much to Mauri’s influ-
ence. I also want to make special mention of my other PhD examiner, William
J. Mitchell, who sadly passed away in June 2010. I owe him, not only for his
extraordinary generosity in wading through my thesis while on a long-haul flight
at a time when his illness had already been diagnosed, but also because his
influence on my work has been profound. As I occupy that strange territory of
design that has been fuzzied by computer technologies, I am sometimes inclined
to explain what I do to others with the phrase ‘I’m a bit like Bill Mitchell’ – if only
that were true.
Returning again to Newcastle, my first post-doc position was in the
newly created Culture Lab. While Cambridge is undoubtedly an extraordinary
place to study, I have always found, and continue to find, a creative energy and
enthusiasm in Newcastle which is unique. These characteristics are epitomized
by my former boss, Sally Jane Norman, who gave me my first proper job and
launched my career. If this book were an organism, its DNA would be from
CUMIS and its sustenance would be from Culture Lab. With this in mind, I would
also like to thank Patrick Olivier for maintaining my connection to computer sci-
ence, driving my ambition and opening up research opportunities which aren’t
usually options for those studying in the arts and humanities.
Behind all these physical and intellectual moves, there is an infrastruc-
ture of funds and supports that made all the research possible. To this end, I
must acknowledge the Arts and Humanities Research Council, who have funded
my Masters degree and PhD. While at Culture Lab, I was also funded by the
EU 6th Framework Program. Libraries are an important part of this book, both
as a theme and as a resource and, to this end, I would like to thank Cambridge
University Library (not least for finding my keys, mobile phone, wallet and any
other possession I lost while training to become an absent-minded academic)
and the Robinson Library in Newcastle.
And finally, I would like to acknowledge my parents and wife Meng,
whose support, encouragement, editing, belief and love I do not have enough
words to describe.

xii
Illustration credits

The authors and publishers gratefully acknowledge the following for permission
to reproduce material in the book. Every effort has been made to contact and
acknowledge copyright owners. The publishers would be grateful to hear from
any copyright holder who is not acknowledged here and will undertake to rectify
any errors or omissions in future printings or editions of the book.

Figure I.1 © Neringa Stonyte

Figure I.2 © James Britton
Figure 1.8 © Ruth Dickie
Figure 3.3 © Steve Benford
Figure 4.2 © Pattie Maes and Pranav Mistry
It is patterns which connect.
(Bateson 1988: 11)

In Ersilia, to establish the relationships that sustain the city’s life, the
inhabitants stretch strings from the corners of houses […] When the
strings become so numerous that you can no longer pass among
them, the inhabitants leave and the houses are dismantled; only the
strings and their supports remain […] Thus when travelling in the
territory of Ersilia you come upon the ruins of the abandoned cities,
without walls which do not last, without the bones of the dead which
the wind rolls away: spider webs of intricate relationships seeking
a form.
(Calvino 1997: 76)
Introduction

Part 1: The architecture of everything

The Theatre of Memory
It is sixteenth-century Italy and a young scholar stands, with trepidation, on the
stage of a massive amphitheatre. The scholar’s nervousness is not caused by
stage fright as his only audience is a distant, solitary figure bent down over a
scroll. Rather, our young scholar is overwhelmed by the task that lies ahead of
him. Contained within the drawers and shelves that make up the theatre’s seven
graded stalls are ‘all things men can conceive’ (Viglius quoted in Yates 2001: 137)
and the phenomenal challenge ahead of the young scholar is to memorize the
entire content of this massive archive and to emerge, after many months and
years have passed, with a truly encyclopaedic knowledge.
The scholar’s task will be aided by the fact that the knowledge con-
tained within this extraordinary building is ordered through a master classification
system which is both physically and conceptually held aloft by Solomon’s Seven
Pillars of Wisdom and organized by the seven known planets. This represents
nothing less than a perfect organization of objects and the ideas they articulate.
The scholar is right to feel apprehensive but he may also be comforted by the
knowledge that this building has been constructed to channel an almost mystical
power so that, as surely as he will be able to navigate the stalls of the theatre,
by the end of his diligent study, he will have a perfect mental representation of
the theatre and an ability to navigate through any discourse and comfortably
inhabit any subject.

The Library of Babel

At an undisclosed time and location, an old man calling himself ‘a librarian’ sets
down his pen for the final time. He sits in a dimly lit, hexagonally shaped room
containing row upon row of identically bound books. This is one amongst a
seemingly infinite pattern of rooms stretching horizontally and vertically in all
directions.
It is possible, perhaps likely, muses the librarian, that this vast struc-
ture contains all books that have been or might ever be written. To this end, the
library contains every utterance that can be made and thus every piece of knowl-
edge that can be articulated using the 25 symbols that make up the librarian’s
language. There is, however, a catch. The combinatorial completeness of the
library means that, alongside the coherent texts, there are many more containing
random configurations of letters. There are, in fact, so many of these incoherent
I.1
A re-imagining of ‘The Memory Theatre’ by Neringa Stonyte

I.2
Illustration of the Borges story ‘The Library of Babel’ by James Britton
 Introduction

books that the likelihood of the librarian finding a book with a single meaningful
sentence, within his lifetime, is negligible.
The librarian’s despair at this personal revelation is reinforced by the
monotony of his surroundings. The repetition of identically bound books and the
invariability of each library cell is a sure indicator that this terrible, vast building is
not governed by a master organizing system but is rather a cathedral to random-
ness, chance and improbability.

The idea of the universal archive

Both the scholar and the librarian are protagonists in stories of impossible worlds.
The Theatre of Memory, a building proposed by the sixteenth-century polymath
Giulio Camillo Delmino in his book L’idea del theatro and highlighted by Francis
Yates in The Art of Memory (2001), was never built. The theatre’s mnemonic
origins, and the strategy of externalizing memory which were the theatre’s
genesis, have been relegated to the status of an ‘intellectual fossil’ (Rossi 2000:
xxi). The explosion of knowledge and our appreciation of the complexity of its
representation and organization has advanced to such an extent that we can
be certain that no such master organizing system exists, let alone that it could
be contained and articulated in a single building. We might also observe that
Camillo’s idea of a universal archive, perfect in its organization and completely
comprehensive, could never be realized.
The tragic figure of the librarian was first described in the twentieth
century in one of Jorge Luis Borges’s most enduring short stories, ‘The Library
of Babel’ (2000). The Library of Babel is, of course, an allegory in keeping with
Borges’s other works, which deal with problems of knowledge and society
through often fantastical and enigmatic fables.
The two stories have clearly different origins and aspirations but are
bound by common themes. They both speak of the idea of a universal archive.
An idea that Michael Foucault articulates in his discussion of the heteratopia.
An idea that exists

… of accumulating everything, of creating a sort of universal archive,

the desire to enclose all times, all eras, forms and styles within a
single place and yet a place that is outside time, inaccessible to the
wear and tear of the years, according to a plan of almost perpetual
and unlimited accumulation within an irremovable place.
(Foucault 1997: 355)

Both stories also depict the interaction between a solitary individual and a vast
store of information that is embodied and articulated by a building and the organi-
zation of physical objects within it. This interaction involves both physical and
mental tasks for the protagonists, epitomized by the purposeful navigation of
the scholar and the hopeless wanderings of the librarian. These different types
of bodily engagement are further articulated by the buildings themselves and
the way they configure space from the bounded panoptic order of the theatre
to the unbounded maze of the library.

3
Introduction

An articulate architecture
‘All buildings organize something’ suggests Thomas Markus in Buildings and
Power (Markus 1993: 4). If we look beyond the pragmatic purpose of providing
shelter and the more indulgent paraphernalia of architectural decoration and
style, we are still left with articulate buildings. We are left with buildings which
structure the organization of space, generating relationships between people
and objects by shaping our ‘concrete reality’ (ibid.). We are left with buildings
that impose order on chaos. In doing these things, buildings can be inspirational
or tyrannical, creating spaces of beauty and harmony or stamping on complexity
with an imposed simplistic order. Buildings are, therefore, great communicators,
informing on the political, social and moral ideals of those who built them and
affording possibilities and restrictions on the communities they serve. Buildings
define our institutional world and restrict what we can do in it. Our ability to par-
ticipate in society is shaped, however, not only by our use of buildings, but also
by our ability to read them, to know the difference between a shop and a prison,
a school, a bank, a library and a church. Buildings, therefore, define the world
through typologies described through the bounding and configuration of space.
Three particular building typologies are of interest here. Perhaps more
than any other types of building, the library, archive and museum, stand out as
clear illustrations of architecture’s organizational and communicative modes.
Although, by modern definitions, these building types serve different purposes,
their origins, are, in fact, hybrid forms of the great temples for the muses such
as the Library of Alexandria and the memory palaces. Their aims are the same,
to store and organize material objects and structure them in such a way that their

… high and incomparable placing not only performs the office of

conserving for us the things, words, and arts we confide to it, so
that we may find at once whenever we need them, but also gives us
true wisdom from those founts so that we come to the knowledge
of things from their causes and not from their effects.
(Camillo quoted in Yates 2001: 147)

In other words, more than a storehouse of knowledge, these buildings, through

the organization of the spaces in which material objects are held, form a dis-
course that is separate, although not independent, from their contents.
The vision, articulated by Camillo through the Theatre of Memory, is
recognisably the scourge of the modern library. Bounded by the constraints of
space and the inevitably linear arrangements of books on shelves, the physical
organization of objects ignores ‘the multi-dimensional relations among books,
and forces a librarian to choose one amongst many possible relations’ (Markus
1993: 174). But, although limited, these arrangements provide a useful purpose.
In Michael Foucault’s terms, these physical organizations represent ‘discursive
formations’, a concept which Gary Radford articulates in his description of an
encounter with the shelves of a library (Figure I.3):

Just by looking at the titles on the spines, you can see how the books
cluster together … you can identify those books that seem to form

4
 Introduction

I.3
The books are a territory.
Imagining the organization of
books on a library shelf as a
discursive formation

the heart of the discursive formation and those books that reside on
the margins. Moving along the shelves, you see those books that
tend to bleed over into other classifications and that straddle multiple
discursive formations. You can physically and sensually experience …
those points that feel like state borders or national boundaries, those
points where one subject ends and another begins, or those magical
places where one subject has morphed into another.
(Radford 2003: 3)

By being ‘discursive’, these physical organizations of objects also reveal their

weaknesses. It is impossible to look upon Camillo’s plan for the Theatre of
Memory, for example, without noting that the seven-tiered organization, articu-
lated by the seven grades of the theatre and relating to the seven known planets,
is rendered obsolete given our contemporary knowledge of at least nine planets
orbiting the sun. David Weinberger gives an account of the problematically static
nature of classification systems based on physical organization in his discussion
of the Dewey system of library classification:

Dewey’s arrangement of the top-level categories has gotten less

appropriate over time. The speakers of ‘ural-altaic, paleosiberian, and
dravidian’ get their own whole number category [494] but the 1.2 bil-
lion who speak Chinese do not. And there’s still a special category
for the ‘education of women’ dating back to when educated women
were a special case ….
(Weinberger 2007: 48)

Even if we don’t focus on the major omissions and misrepresentations in its

structure, Dewey, like any other categorization system, is riddled with inconsist-
encies when measured against our evolving representations of a topic domain.
Indeed, Weinberger goes further:

5
Introduction

If we want to see how the physical world has silently shaped how
we put together our ideas about the world – and why any traditional
classification scheme is bound to embarrass somebody – there is no
better example than the Dewey Decimal system.
(ibid.)

Whereas in the past the discourse on the evolving nature of knowledge and its
classification might have led to an amendment of Dewey or, occasionally, the
proposition of a new system, Weinberger is able to propose a real alternative by
separating information from its material manifestation entirely.

Part 2: Changing patterns of architecture and information

Weinberger was not the first to discuss the limitations of physical organization
and classification and nor was he the first to suggest an alternative. If libraries, for
example, survive the twenty-first century, they will be fundamentally reshaped.
Indeed, we may be witnessing a re-convergence of the typologies of library,
museum and archive. Even the most conservative onlookers cannot expect the
library, in its current form, to remain at the centre of information and its organiza-
tion and to continue to be the Mecca for those wanting to worship the products
of knowledge. As Mitchell states in his description of the reading room at the
British Library where Karl Marx is said to have written Das Kapital: ‘It will not be
possible to tell tourists where some Marx of the next millennium sat. All that is
solid melts in the air’ (Mitchell 2000: 56–7). What is true of the library may well
be true of other types of architecture as well.
It is a well-rehearsed truism that the processes and products of
digital technology have revolutionized the way we live and work and the way
information is distributed and interfaced. The ‘information revolution’ (although
the era has various names) has not been caused by one technology or platform
but by many: notably the World Wide Web (WWW), personal computers (PCs)
and mobile telecommunications. However, these technologies are bound by a
common digital and computational genesis, which has involved the separation
of information from its material means. Digital technologies are not material-less
(anyone who has seen the scars left on the pavements and roads that have been
dug up to lay fibre-optic cable will bear testament to this) but the fidelity of cop-
ies, the ability of information to proliferate and the negligible impact of distance
on the speed of its distribution has meant that it is possible to talk of informa-
tion as a concept without referring to its material inscriptions. This process
was described by one architectural theorist as ‘losing the ballast of materiality’
(Benedikt 1991: 4), a statement that reflects a more general consensus on the
virtuous nature of information’s escape from its material prison. This idea goes
well beyond the museum and library to the fundamental idea of architecture as
a material practice and its role in the patterning of information.

Cyberspace
How has the architectural profession handled the dematerialization of information
on its practice? The revolution in the use of Computer-Aided Design (CAD) in mod-
ern architectural practice is well known and discussed but the relationship between

6
 Introduction

buildings as information objects and architecture as a pattern-making activity in a

world of dematerialized information has remained at the fringes of the architectural
discourse. Notably however, in the 1990s, a group of architectural thinkers and
designers tackled the problem of architecture and information in the digital age
head-on. The mid-nineties became the era of ‘cyberspace’ and this period is marked
by a great revolutionary enthusiasm (not least by the investors in a new, but as it
turned out fragile, e-economy). The work on architecture and cyberspace looks
dated now and is more likely to be referenced as a period piece rather than as work
with contemporary relevance, but, at the heart of the discussion of cyberspace is
a dichotomy that does have contemporary relevance and for this reason it is worth
recalling the cyberspace discourse with the benefit of hindsight.
The term ‘cyberspace’ is a product of the influential science-fiction
author William Gibson, who coined the term in his 1982 novelette Burning
Chrome in Omni magazine (later published in 1986 in the collection also called
Burning Chrome) and made famous by his novel Neuromancer (Gibson 1995),
first published in 1984.
The term cyberspace is often considered to be part of an unwelcome
culture of meaningless buzzwords, just another lame metaphor like ‘informa-
tion superhighway’ and ‘surfing the web’ (McCullough 2004: 10). Cyberspace
has been defined vaguely as ‘the totality of the world’s networked computers,
which form a huge virtual space inside which people can communicate and locate
stored information’ (Pountain 2003: 101), or ‘where you are when you’re talking
on the phone’ (Rucker et al. 1993: 78). But, although it eludes a clear textual
definition, the term cyberspace has been in widespread use from the early 1990s
and is often conceptualized, using a passage from Neuromancer, as:

A consensual hallucination experienced daily by billions of legitimate

operators, in every nation, by children being taught mathematical
concepts … a graphical representation of data abstracted from the
banks of every computer in the human system. Unthinkable complex-
ity. Lines of light ranged in the nonspace of the mind, clusters and
constellations of data. Like city lights receding.
(Gibson 1995: 67)

Cyberspace has, since Neuromancer, been adopted as a placeholder in a wide

range of contexts, acting as a stand-in for terms that would require more complex
definitions without it. Cyberspace can be used equally to refer to the infrastruc-
tures and information of the post-digital revolution and to a way of manifesting
those processes and products through, so-called ‘virtual reality’ technology.
Gibsonian cyberspace seems to be more grounded in the latter, through a meta-
phorical relationship to real spaces: ‘like city lights receding’ (Gibson 1995: 67)
and hence has been used almost interchangeably with the technologies of 3D
virtual environments.
Literary and cinematic1 descendents of Gibson have focussed on
cyberspace as an immersive world in which abstract digital information is made
manifest through metaphors of real spaces. Neil Stephenson’s Snow Crash (1993),
which, along with Neromancer, has become a recognized classic of the cyberpunk

7
Introduction

I.4
Imagining Neil Stephenson’s
‘Metaverse’ based on a
SketchUp model of Las
Vegas by Koen Jespers

genre, extends Gibson’s vision by developing the Metaverse (Figure I.4), which,
like Gibson’s cyberspace, consists of an immersive, purely computer-generated
environment but which, even more than Gibson’s cyberspace, is explicitly meta-
phorical, resembling the buildings and streets of a city:

The sky and the ground are black, like a computer screen that hasn’t
had anything drawn on it yet; it is always nighttime in the Metaverse
and the street is always garish and brilliant, like Las Vegas freed from
the constraints of physics and finance …. Downtown is a dozen
Manhattans embroidered with neon and stacked on top of each other.
(Stephenson 1993: 26)

8
 Introduction

Gibson’s influence went well beyond the cyberpunk genre of science fiction
which he and others had defined, and the notion of an information space made
manifest through an immersive 3D environment gained extensive currency
(Dodge and Kitchin et al. 2001b: 230). With the publication of John Walker’s heav-
ily Gibson-inspired white paper from 1988 for the software developer Autodesk,
entitled Through the Looking Glass: Beyond User Interfaces later published in
an abridged form in The Art of Computer Human Interface Design (1990), it
became clear that technology developers were taking Gibsonian cyberspace
seriously. It was natural to conflate the metaphor with the real thing and it fol-
lowed, therefore, that a new set of designers would be needed to design and
articulate this new space and, since Gibsonian cyberspace was already framed
by the metaphor of the built environment, that architects should have a stake in
this new design domain.

Architects and cyberspace

The first significant architectural text on the subject of cyberspace emerged in
1991 with Michael Benedikt’s edited book Cyberspace First Steps, which, from
its first reference, took its cue directly from William Gibson (Benedikt 1991: 1).
Even in the introduction to this edited text, however, Benedikt faced two prob-
lems. In 1991, neither an appropriate definition nor the technological means of
realizing Gibsonian cyberspace existed. Benedikt instead takes on the challenge
of finding an understanding that would ground cyberspace, not in a technologi-
cal reality but in a more ancient cultural ‘mental geography’ (Benedikt 1991: 4),
aligning himself with a view of cyberspace as a ‘nonspace of the mind’:

Like Shangri-la, like mathematics, like every story ever told or sung,
a mental geography of sorts has existed in the living mind of every
culture, a collective memory or hallucination, an agreed-upon territory
of mythical figures, symbols, rules and truths, owned and traversable
by all who learned its ways, and yet free from the bounds of physical
space and time.
(Benedikt 1991: 3)

As for the problem of a technological justification, Benedikt takes an even more

speculative position. Again following Gibson, he looks toward immersive virtual
environments for cyberspace’s technological realization through the use of head-
mounted displays (HMDs) and, ultimately, ‘direct neural connections’ (Benedikt
1991: 11–12).
It seems not to have occurred to Benedikt to question the specula-
tive technological basis for his claims for cyberspace.2 For Benedikt, Gibsonian
cyberspace was more than a science-fiction speculation; it was a reality and one
that the architectural profession needed to embrace:

The door to cyberspace is open, and I believe that poetically and

scientifically minded architects can and will step through it in sig-
nificant numbers. For cyberspace will require constant planning and
organization. The structures proliferating within it will require design,

9
Introduction

and the people who design these structures will be called cyberspace
architects.
(Benedikt 1991: 18)

Despite this speculation, only two ‘cyberspace architects’ contributed to

Cyberspace First Steps, and the volume was composed of an interdisciplinary
mix of technologists and social scientists, each with a unique perspective on
the future of cyberspace technology. Apart from Benedikt, the other cyberspace
architect, Markos Novak, made the most controversial and, in the architectural
domain, most influential contribution with his article, ‘Liquid Architectures’.
Following Benedikt’s work closely, Novak is no more precise in his definition of
cyberspace:

Cyberspace is a habitat of the imagination. Cyberspace is the place

where conscious dreaming meets subconscious dreaming, a land-
scape of rational magic, of mystical reason, the locus and triumph of
poetry over poverty, of ‘it-can-be-so’ over ‘it-could-be-so’.
(Novak 1991: 226)

Novak’s work is divided between experimental artwork and philosophical mus-

ings and, although sometimes unfathomable, his writings on cyberspace became
temporarily influential. Novak, however, though undoubtedly an original thinker,
is dogged by the inevitable contradiction between his poetic writing and his prac-
tice. Where Novak has remained loyal to Gibson’s ‘nonspace of the mind’, stating
that ‘Cyberspace offers the opportunity of maximizing the benefits of separating
data, information and form ….’ (ibid.), inherent in Gibsonian cyberspace is a con-
tradiction. Where Gibson as a novelist is able to get away with verbal descriptions
of cyberspace as both conceptual and embodied; as a ‘trans-architect’, Novak
had been inspired to realize his ‘liquid architectures’ through visual representa-
tions. Almost as an afterthought, Novak’s images are introduced in a few lines
at the end of his essay (Novak 1991: 252), revealing a critical contradiction. Far
from ‘disconnecting information and form’ (Novak 1991: 225), Novak’s images
of cyberspace, though fantastical and unencumbered by Newtonian physics, are
clearly defined by another set of rules. We might reasonably ask the question
of where form comes from when it is separated from information? His liquid
architectures are, in fact, a search for ‘cyberspace’ through the vehicle of tradi-
tional image making:

Despite Novak’s conviction that the computer can transform archi-

tecture through time, … [his images reveal] … a dilemma. The formal
composition of his virtual architecture, not unlike De Stijl explorations,
is strictly controlled by 3D (perspectival) space.
(Perez-Gomez and Pelletier 2000: 379)

Novak’s liquid architectures are far from being ‘nonspaces’. The images Novak
presents carry with them assumptions about how space should be constituted
and represented in the emerging cyberspaces, but this is not elaborated on in

10
 Introduction

either his or Benedikt’s texts. These ‘transarchitects’ are, as Benedikt suggests

in his opening chapter of Cyberspace First Steps, ‘loosing the ballast of material-
ity’ (Benedikt 1991: 4), but this has not freed them from geometric constraints.
Despite the inconclusiveness of Benedikt’s and Novak’s work, the
concept of a cyber architect did have some influence and, throughout the nine-
ties, there were articles in mainstream architecture journals on the ‘Architecture
of Cyberspace’ (see in particular Ostler 1994 and Anders 1994). By the time,
in 1995, that Architectural Design Profile 118 emerged, it had become clear
that the architectural profession was taking cyberspace seriously. Entitled
Architects in Cyberspace, it again contained a contribution from Marcos Novak;
‘Transmitting Architecture’ (1995). Whereas Novak’s contribution continued to
be highly speculative, others were offering a more grounded approach to the
cyberspace question, reacting not to futuristic visions but the reality of emerging
technologies. For example, in William Mitchell’s ‘Soft Cities’, (1995) and Dunne
and Raby’s ‘Fields and Thresholds’ (1995), the nature of cyberspace reflected
concepts that were recognisably architectural and there are direct references
to physical contexts and to realized technologies and applications. In Mitchell’s
article, which can be seen as a precursor to his City of Bits trilogy,3 references
are made to computer games and MUDs (multi-user domains) and even to the
physical context of ATMs (automatic teller machines), all technologies that were
in widespread use. Similarly, Dunne and Raby’s article seemed to challenge
the Gibsonian dematerialized cyberspace by attempting to make the invisible
electromagnetic climates of digital technology tangible in real physical spaces,
reversing the idea that cyberspace was something ‘out there’ or part of another
world and reminding the reader of their embodied existence:
For us, the communication aspects of telecomputing are less about
ways of ‘inhabiting’ abstract digital ‘space’ and more about the exploration of
new situations arising in physical space (Dunne and Raby 1995: 61).
By the time Architects in Cyberspace II (Spiller 1998a) was published,
the key cyber-architects, or trans-architects who were still holding out for the
emergence of a cyber-architecture, had seemingly become blinded by a revolu-
tionary zeal and were drawn into making increasingly radical pronouncements,
dismembering our bodies and space time itself:

The vision cone that has for so long been used to describe the
boundaries of our visual fields is being postponed and cut out of rec-
ognition. Our new visual geometries are complex and multifocused.
We are creating new possibilities for architectural vision and site in
the light of human perception beyond fleshy limits.
(Spiller 1998a: 63)

Architects in Cyberspace II is also dominated by, now familiar, Novakian imagery.

Although more elaborate, highly rendered and composed of more complex geom-
etries, Novak’s data-driven forms exhibited the aesthetic started in Cyberspace:
First Steps. The influence of Marcos Novak had been felt, not so much through
his writings, but through the creation of a visual rhetoric and a digital aesthetic
achieved through an artful use of CAD, 3D modelling and rendering software,

11
Introduction

which, by the time Architects in Cyberspace II was published, had become the
de facto standard for the new brand of digital architectural graphics.

The failure of architectural cyberspace

There was to be no Architects in Cyberspace III, although subsequent jour-
nal issues, for example Reflexive Architecture (Spiller 2002) and 4D Space:
Interactive Architecture (Bullivant 2005), would continue the enthusiasm for
digitally enabled architecture. Other collections and manifestos were to follow,
including notable contributions in John Beckmann’s edited book, The Virtual
Dimension (1998) and Neil Spiller’s manifesto, Digital Dreams (1998b). The word
‘cyberspace’, however, had been lost from both the titles and the content, and
it became clear that digital architecture was not now seen as something that
existed in an illusory cyberspace, but as something affecting and residing in real
physical situations. The potential of the computer as a medium of communica-
tion and the growing concern with the real physical spaces of computing have
shifted the emphasis of architectural design away from notions of a purely digital
cyberspace toward discussion of the impact of digital technology on the design
of the real ‘grounded’ contexts in which architecture is more at home:

Digital networks are no longer separate from architecture. Unlike

cyberspace, which was conceived as a tabula rasa, pervasive comput-
ing has to be inscribed into the social and environmental complexity
of the existing physical environment.
(McCullough 2004: 1)

It is tempting to dismiss architecture’s short-lived fascination with ‘cyberspace’

as naive, a product of an architectural discourse coming to terms with the huge
impact of the computer on professional practice, or as merely an opportunity to
realize Modernism’s most extreme fantasies of transparency through a remak-
ing of the world ‘according to its own rational abstractions’ (McCullough 2004:
11), but these works need to be seen as period pieces, as insights into a time.
Awareness of the speed of technological advancement is not, however, neces-
sarily matched by knowledge of its direction, leading to a form of technological
determinism which is fed by an undetermined technology. This problem does
not render these endeavours into cyberspace irrelevant, however. Cyberspace,
McCullough states, is more than a metaphor and ‘at times seemed more like a
society enacted myth’ (McCullough 2004: 9).
Although the term ‘cyberspace’ can only be viewed as a placeholder,
the processes and products to which it refers remain and are still challenging
to the architectural designer. Far from redefining space, or moving beyond user
interfaces, cyberspace and its representations reflect a will to find a traditional
spatiality, or at least the metaphor of one, in the context of digital representations
of information. Digital information has been literally ‘objectified’, conceptualized
as objects held within 3D spaces.
The concepts so effortlessly conflated in Gibson’s description of
cyberspace as simultaneously ‘like city lights receding’ and ‘like the nonspace
of the mind’ are not so easy to reconcile outside the world of the science fiction

12
 Introduction

novel. As architects and architectural-theorists rapidly realized, architecture as

traditionally practiced is not about ‘nonspace’ but requires spatial design, expres-
sion and communication through visual means.
The problem of defining an architecture of cyberspace is a useful one
to address because it acts as a microcosm of the broader problem of how to
conceptualize digital technological systems, which are seemingly without form.
The history of cyberspace is, however, also a warning of what happens when
representation and metaphor are confused with reality and space:

A major problem with the discourse on cyberspace as a medium for

architecture seems to be the pervasive and radical gnosticism that
forgets ‘prior’ experience. Our reality is neither exclusively ‘natural’
or ‘cultural’ … mimesis is our true nature, a capacity that allows
humans to transform inherited culture into second nature and thus
understand our purpose.
(Pérez-Gómez and Pelletier 2000: 380)

Information architecture
The end of the nineties brought with it a new enactment of the cyberspace
dichotomy. This time the impetus was not from architecture or science fiction
but from the pragmatics of developing websites. The reality of everyday con-
tact with information on the web revealed a disconnection between the way in
which web developers were structuring the content of websites and the work
of graphic designers whose job it was to make individual pages attractive and
above all, usable. This disconnection manifested itself in the ‘lost in hyperspace
problem’, with the now familiar sense of disorientation that occurs when one
is attempting to find information on a particularly complex website. To solve
this problem, a new profession emerged called ‘information architecture’ (IA),
populated by ‘information architects’.
The practice of IA has grown substantially and Rosenfeld and
Morville’s book, called Information Architecture for the World Wide Web (2002)
is one among many which mark it out as an emerging field (see for example Van
Dijck 2003, Wodtke 2003). Connected to the growth of the WWW, IA is now
viewed as a field of expertise in its own right, attracting people from many profes-
sions, including traditional architectural design (Rosenfeld and Morville 2002: 20).
Although the term ‘Information architecture’ has entered into
widespread use, it has, so far, remained largely outside academic and theo-
retical discourse. At the time of writing (2011), searching online for the term
‘Information architecture’ reveals a host of professional bodies and online tutori-
als but few research papers or academic conferences. The term was coined by
Richard Saul Wurman (himself a trained architect) in his 1996 text Information
Architects and follows the highly influential Xerox PARC research centre’s vision
of creating ‘the architecture of information’ (Pake 1985). But, even by the 2003
edition of Information Architecture for the World Wide Web, which is now widely
recognized as one of IA’s key texts, no single definition of what constitutes IA
had emerged.

13
Introduction

As multidisciplinary as the traditional profession of architecture, IA

has been variously described in the ways listed below:

1 The combination of organization, labelling and navigation

schemes within an information system.
2 The structural design of an information space to facilitate task
completion and interactive access to content.
3 The art and science of structuring and classifying web sites and
intranets to help people manage information.
4 An emerging discipline and community of practice focused
on bringing principles of design and architecture to the digital
landscape.
(Rosenfeld and Morville 2002: 4)

These four definitions refer to different facets of the same underlying design
challenges but I will start by focusing on the fourth here.
The development of IA, which is seemingly analogous to architecture
as the design of the built environment, is not only a potentially new design field in
which architects can work, but it is also a challenge to the profession. Reflecting
on the notion of the virtual library for example, the architectural theorist, William
J. Mitchell, reflects on the impact on architecture as it is traditionally conceived
once digital edifices have replaced physical ones:

The task facing the designers of [the] soft library is a transforma-

tion (with some invariants but many radical changes) of what faced
the Smirke brothers and the librarian Panizzi as they evolved the
design for the British Museum and Library. The façade is not to be
constructed of stone and located on a street in Bloomsbury, but
of pixels on thousands of screens scattered throughout the world.
Organizing book stacks and providing access to them turns into a task
of structuring databases and providing search and retrieval routines.
Reading tables become display windows on screens …. The huge
stacks shrink to almost negligible size, the seats and carrels disperse
and there is nothing left to put a grand façade on.
(Mitchell 2000: 56–7)

In Mitchell’s description, architecture has clearly been reconfigured and perhaps

replaced by a new IA. Although he places emphasis on the radical changes that are
being brought about by digital technology, particularly in terms of the storage and
retrieval of information, he notes that there are certain ‘invariant’ characteristics.
From the very notion of an information architect, through to the building analogy used
by Rosenfeld and Morville (2002), it is clear that traditional notions of information
and of its representation in physical space provide a conceptual framework, linking
information, its organization, its display and its navigation in new digital contexts.
Where, if anywhere, is the architecture in information architecture?
By this I don’t mean to refer only to the discipline of web design but, more
generally, to the macro organization of information that has, historically, been

14
 Introduction

the domain of buildings. It is tempting to succumb to the revolutionary zeal

associated with digital technologies and the dematerialization of information but,
as I have illustrated here, the path to a digital nirvana has been a difficult one.
Whatever the potentials of a new medium, as it develops, it rarely takes on truly
revolutionary structures but instead goes through what Bolter and Grusin (1999)
refer to as ‘remediation’, by retaining some aspects of its original form and gradu-
ally acquiring new ones. This process of remediation has been well documented
for a range of media forms including theatre (Murray 1997), cinema (Manovich
2002) and literature (Bolter 2001) but not yet for architecture.
Architectural designers are very aware of the impact that computation
is having on their profession and new CAD systems, visualization, simulation and
advanced building methods have become part of modern architectural discourse.
There is also a growing acknowledgement of the importance of architectural
design in understanding embedded and ubiquitous computer systems as they
are integrated into our physical environment. However, aside from the short-lived
discussions of cyberspace, architectural designers have remained relatively silent
as, in a world of ‘information architecture’, a new domain claims part of their
discipline and borrows their concepts and terminology.

Part 3: Scope and focus

It is against the confusing backdrop created by the quest for an architecture of
information and the vision of an architecture for cyberspace that I ask the question:

Where is the architecture in information architecture?

The question provides a deceptively simple starting point but is associated with
a slightly less simple hypothesis:

The spatial patterns that we make in our environment are a primary

means of human communication. We use space to organize people
and objects and those physical organizations are the basis of much of
our conceptual thought. Consequently, despite the potential created
by digital technology to transcend the material organization of infor-
mation, patterns that relate to the organization of architectural space
continue to have a role in the design of digital information systems.

Testing this hypothesis requires us to broaden our understanding of what archi-

tecture is, while simultaneously focusing our discussion on a particular structural
understanding of architectural space. In doing this I will look beyond the surface
appearance of digital information systems to help to develop an understanding
of how human beings seek to pattern their conceptual world, whether that world
is mediated by digital technologies or by the physical placement and organiza-
tion of objects. This book thus unites historical, philosophical, psychological
and architectural knowledge to develop a framework for understanding the
relationship between information and its representation in a post-digital era. The
analysis suggests that, despite the opportunity to separate information from its
material means, metaphors based on objects and physical spaces, which are

15
Introduction

commonly associated with the storage of digital information, are characteristics

of a deeper relationship between the world of physical objects and the world
of ideas. This proposal stands in contrast to accounts of digital technologies
which propose material-less disembodied interactions with computer-based
information.

Technical focus
In defining the focus for this book I have taken three decisions which, in many
ways, go against the prevailing trends in architecture and the digital revolution,
but which provide, in my view, a useful basis for the consideration of topics that
have been addressed very little as part of a discourse on architecture.
The first decision has been to focus much of my initial attention,
in terms of digital technology, on screen-based user interfaces and, what now
might be considered passé (in research terms), the topics of graphical user inter-
faces for personal computers (GUIs) and hypertexts on the WWW.
There has been a proliferation of devices that have supported inputs
to and outputs from computers and, although the currently favoured input
devices are the desktop computer’s keyboard and mouse, as the computational
potential of a range of devices from mobile phones to fridges is developed, the
input mechanisms for computation are also set to change, resulting in much dis-
cussion around the rise of ubiquitous computing. The screen still dominates most
modern computer interaction systems, however, and, far from being replaced, it
has proliferated within computational devices, with the main changes being to
resolution and size. There are currently no realistic alternatives to the technology
of the screen, so it seems likely that screen technology is here for some time
to come. From the early history of virtual reality, where HMDs were discussed
as the future of visual computer interaction, to a modern interest in tangible
computing, the screen has often been considered as a neutral ‘window on the
world’ rather than a material presence in its own right, and as a mediating device
that frames the objects it displays and sets up a direct and specific relationship
with the person who is viewing it.
Leading-edge thinking, both in terms of architectural theory and
studies in human–computer interaction, now emphasize the technical paradigm
of ubiquitous and pervasive computing and architectural theory seems more
adept at handling the implications of a physical environment augmented with
computers than at dealing with computational environments in their own right.
By making GUIs my main focus, I do not intend to suggest that screen-based
interaction with information is the necessary or inevitable future of computer
information. Rather, by focusing particularly on the much neglected (from an
architectural point of view) paradigms in the history and development of comput-
ing, I want to show how a core understanding of spatial organization pervades
our understanding of computational technology. To this end, the examples used
in Chapters 2 and 3 will be historical and limited to past or current and everyday
technologies, with Chapter 4 used for a more speculative discourse on future
technologies.
The second decision involves the omission of virtual reality technolo-
gies, virtual environments, computer games set in real-time 3D virtual worlds and

16
 Introduction

other types of computer user interface that might be seen as direct simulations
of architectural spaces. In contrast, I will focus on a more abstract understand-
ing of architectural space and look at information environments where there is
no requirement for simulation, but which can and often have been framed by
metaphors of architectural space. Architecture as metaphor implies that archi-
tecture is not itself the focus of representation, but is used as a stand-in for
something else. Examples of architecture used as a metaphor can be found in
a range of web applications and in discussions on future ‘intuitive’ information
systems and it is argued by some that the use of virtual architectures provides a
familiar context for the user by simulating aspects of their everyday experience
in the representation of abstract systems. The focus of this study, however, will
be broader than this, dealing with a variety of digital information systems that
conceptualize information objects, organized to be both comprehensible and
structured in meaningful ways. Sometimes, to achieve this comprehensibility,
visualizations will be used which overtly reference architectural space but such
approaches are, it will be shown, the (not always logical) conclusion of a much
less overt conception of structured spaces as the basis for the organization of
digital information.
The third decision is to focus on single-user interfaces (SUI). These
are interfaces which can only be interacted with by one user at a time and
which do not allow for interaction between several users simultaneously, either
remotely or locally. For example, a digital library catalogue system allows for
a dialogue between a single individual and the computer and, although others
may be accessing the system from different locations, they are unaware of each
other’s presence. This decision has been taken because there is already a well-
defined discourse on the use of the computer in collaborative work as well as
on issues of telepresence and also because the single-user interface remains an
‘unsolved problem’ (Raskin 2000: xix).
This book will seek to move beyond the use of the architectural meta-
phor, and in particular the limited definition of ‘information space’ which results
from it, suggesting that the architectural metaphor is a façade which hides the
real potential of an architectural understanding of the design of information sys-
tems. It will deal in general terms with the graphical user interface component
of computer systems but, since this could refer to a wide range of applications,
it focuses on two types in Chapters 2 and 3.

1 Personal file management systems, with a particular emphasis on the

personal computer and the operating system’s desktop-based GUI with its
documents, files and folders.
2 Websites, with particular reference to those where hypertext documents
are connected and navigated using a browser, most often a web browser
such as Explorer or Firefox. It is noted, however, through the ‘internet of
things’, that the notion of hyper-linking has, as will be shown in Chapter 4,
extended outside the virtual world of the WWW.

These two application types represent underlying structures of information and

connect information objects, allowing these connections to be navigated or

17
Introduction

explored. However, each system is authored in a different way and represents

an alternative method of presenting underlying information structures.

Data, knowledge and information objects

Key distinctions that are made in this text are those between data, information
and knowledge in the context of computational representations. In simple terms
the difference between data, information and knowledge is determined by the
degree of connectivity between the data (single facts or values) that form struc-
tures of association (information), which can inform action (knowledge).
Definitions of knowledge, however, vary and, for example, those
involved in artificial intelligence research have attempted to systemize our
understanding of knowledge in order to create machines that can extend beyond
their original programming and learn. Such machines would be able to develop
an understanding of the world based upon the input of their own senses (or
sensors). This is not a matter of information accumulation alone, the system
must know what to do with the information, categorizing it and using it in new
but similar situations, in other words learning by experience. A simple example
of this would be a robot with a light sensor (equated to an eye), wheels, a turn-
ing mechanism and a motor programmed to move forward continuously as far
as possible without stopping but without the requisite programming to link its
sensory input to the action of avoiding a barrier. The robot is placed on a surface
with walls on all four sides. At first the robot will move forward and bump into the
wall; it will then engage its systems in different ways. First it may try reversing
and moving forward again but will still hit the wall, then it will engage its turning
motor, turning around 360 degrees before moving forward, but again it hits the
wall. It tries again and, having ruled out reversing and 360 degree turns, it tries a
340-degree turn. The robot hits the wall again but this time it can move for longer.
If this process continues, the robot will eventually learn an optimum turning circle
and will never hit the wall. To achieve this, the robot must remember a category
for wall that is instrumental i.e. a ‘wall’ equates to an inability to move forward
and with an input pattern in its mounted sensor. The robot now has knowledge
of the wall composed of a structure of memories and a corresponding set of
actions (Figure I.5).
Low-level descriptions of knowledge such as these are common
in AI and other areas of knowledge representation which deal with the way
the mind or the machine categorizes and structures incoming information in
a meaningful way. This knowledge, therefore, must be ‘represented’ to the
machine both through software and in relation to the physical structure of the
hardware. One of the key concerns for cognitive scientists is to discern exactly
what these knowledge representations are. In other words, how is knowledge
structured in a mind?
Defining human knowledge relies on a more subtle understanding of
concepts and actions but the same basic rules can be applied in distinguishing
between data information and knowledge. Saeema Ahmed suggests that ‘knowl-
edge’ is a ‘relative term’ and can be defined in relation to data and information.
Ahmed uses this analogy to help arrive at a definition:

18
 Introduction

I.5
Diagrams to show a robot
going through a process of
learning to associate a wall
with an obstruction

[An] … example of data is a signal from an electrocardiogram (ECG)

tracing the heart’s beat. The signal is data to someone who is una-
ware that this is an ECG. If the signal is observed by the patient,
who is aware that this signal is from the heart, the data becomes
information. The signal is potential knowledge to a patient who is also
a medical student and able to interpret the ECG. When the medical

19
Introduction

student interprets the ECG, he or she has gained knowledge of the

heart’s condition.
(Ahmed et al. 1999: 8)

This relativistic approach allows for a broader definition of knowledge based

upon context. So, for example, the sort of knowledge which is used to prevent
a machine from hitting a wall is much less complex than the knowledge used
by a medical student to interpret an ECG machine. This book will examine the
concept of human knowledge rather than artificial intelligence (AI) and will take
the standpoint that human knowledge is something that can be communicated
through computers but not created by them.
Throughout this book I will use the concept of the ‘information object’
to represent a discreet piece of data that can be distinguished from other data
in a system. How the information object is defined can depend on context. In a
library, an information object is a book; in a museum it is an artefact; in a hyper-
text it might be an individual page; and in an information-retrieval system it may
be a single word. The importance of viewing an information object in context is
that, by defining discreet objects, it allows me to look at their relationships with
other objects as part of a system so that, in the case of a library for example, I am
less interested in the content of individual book pages but rather the relationship
of book objects to other book objects as part of a discursive system represented
by their arrangement on the library shelves. Returning to the differentiation
between data, information and knowledge; an information object on its own
consists of data, a collection of information objects arranged with relation to a
meaningful organization strategy constitutes information and knowledge is the
interpretation and use of that organization by human minds.

Part 4: The place of architectural theory

In the post-cyberspace era of architectural theory, there has been a new focus
on the architectural discussion of digital technologies, dominated by three fig-
ures William J. Mitchell, Malcolm McCullough and Richard Coyne. Mitchell’s
writings predominantly reflect on the impact of information communication tech-
nologies (ICTs) on urban environments through discourse on digital urbanism.
McCullough’s work has contributed to the merging of architectural design with
interaction design and Coyne has found philosophical bridges between comput-
ing and architectural theory. Each of these writers has registered a voice which
is distinctive but distinctly architectural to the point of, in the case of Coyne in
particular, adding poetry to a broader culture of ICT, which is often technical and
dry. These studies are also notable because of their breadth. Each author has
attempted to define a broad understanding of a new domain and a territory for
future researchers.4 For those of us who follow this work, it is necessary to find
a way of deepening our discourse and to discover ways of researching that are
informed by architecture but are not necessarily architectural. In other words, to
find ways, not of crossing disciplines, but of occupying the spaces created by
the crossings that have already been made.
Over my short career to date, I have had the opportunity to work
in a range of contexts as a member of teams dawn from a wide range of

20
 Introduction

different disciplines, notably working with computer scientists involved in

human–computer interaction and interaction design. To practice design in these
interdisciplinary teams involves a negotiation of expertise and a need to demon-
strate what an architect can bring to the table. To articulate this, I have sought to
find theories from my discipline which are relevant and, above all, comprehen-
sible to a wider audience. This is not an easy task. Of the architectural theories
that are expounded in journals and books every year, many are highly specialized
and involve vocabulary guaranteed to alienate an outsider. In the postmodern era
of discourse in the arts and humanities, there is an emphasis on complexifica-
tion involving accounts of academic work that revel in the rich complexities of
life and, to coin a postmodern phrase, to know its unknowableness. This way of
looking at the world is in sharp contrast to work in the sciences, where enquiry
is only possible through the process of reductionism. In other words science
reduces the systems or processes of the natural world into components that
can be studied independently before being reassembled. To the designer, these
‘Two Cultures’5 are territories that must be bridged, particularity in the context
of design problems which emerge from technologies that themselves are the
result of scientific method.
This book is, therefore, narrow in scope (compared to the predeces-
sor works mentioned above) and is not only influenced by architectural thinking in
some areas but also driven by a search for patterns and ideas from across many
other fields that are equally applicable in the context of design. These ideas are
chosen because they are coherent (both internally and with one another), articu-
lated clearly and in ways that make them more open to scrutiny and because they
split the problem domain into manageable parts. The architectural approaches
to the topic are also chosen using this criteria and I have thus selected a group
of theories that can be described as deriving from a structural tradition of archi-
tectural analysis and its representations of spatial experience.

Form, function and space

To frame the idea of ‘architectural experience’, this book will focus on space as
a medium of communication and on how architectural space encodes meaning.
One of the most convincing contemporary accounts of architecture and its mean-
ing comes from Thomas Markus’s Buildings and Power, particularly in his elegant
classification of his own method, what he describes as his ‘tools of architectural
analysis’ (Markus 1993: 11–21). These tools he describes as form, function and
space, and it is worth summarizing them here.
A building is a multifaceted object and can be ‘read’ in many different
ways. Imagine, for example, walking into a bank. By necessity the bank building
must convey meanings associated with the institution to which it belongs and
the functional possibilities held within. The naming of the building as a bank car-
ries with it certain expectation and its function may be, literally, inscribed in text
on the façade of the building. Other articulations of the façade may give clues to
the bank’s status and ambitions, for example an austere classical frontage may
imbue the bank with institutional gravitas. Entering the bank reveals a configura-
tion of desks, counters and queuing areas, separating public and private realms
and affording the possibility for certain types of interaction and exchange. These

21
Introduction

meanings, encoded in the bank building, are not solely due to its architecture, but
also result from the relationship between cultural expectations, spatial behaviour
and material structure and order.

Form
Form is the most obvious area of architectural analysis and refers to the physi-
cal shape and organization of architectural components (walls, roofs, windows,
doors, etc.). Differences in the shape, scale and configuration of these compo-
nents are often classified as they relate to particular ‘languages’ of architectural
style. The form of buildings, as Markus points out, quoting Frankl (1969), articu-
lates and composes the geometry of spaces, as well as the mass, surface, light,
colour and ‘other optical phenomena’ (Markus 1993: 11).
It is in the domain of form that much architectural history and theory
is written and where substantial and self-evident contrasts between architectural
types are to be found. While building form is an obvious way of distinguishing
between styles of architecture, the relationship between building, form and the
communication of meaning is not so clear cut. Is architectural meaning encoded
in material forms and, if so, is that meaning universal or culturally specific? To
illustrate this conundrum, Markus distinguishes between Frankl’s history of
Renaissance and Baroque architecture where he ‘reads’ a universal meaning
in composition and style and the linguistically inspired writings on architecture,
notably Umberto Eco’s seminal Function and Sign: Semiotics of Architecture
(1997), where he suggests the ‘association of forms with words’ (Markus 1993:
11) and, thus, separates the sign from the signifier, believing that architectural
forms are only arbitrarily related to the meanings they signify and, therefore,
that these meanings are only communicated if they are received by individuals
with a shared system of knowledge. In other words, Eco is treating architectural
form as a language.
An example which highlights the difference between these two
approaches would be if we were to examine a building constructed follow-
ing the rules of a classical Greek order. The mathematical ratios that give the
building its scale and proportion and relate to those ratios we find in the human
body, may lead us to conclude that this architectural form communicates a uni-
versal beauty and that anyone comprehending the structure would consciously
or subconsciously understand these forms and our corporeal relationship to
them. Alternatively, we might conclude that the classical orders are symbolic,
representing the birth of civilization and democracy. To read this meaning in the
building requires cultural (in this case historical) knowledge. In both cases the
information content consists of an association between the building form and
a concept. However, where the first interpretation is pre-cultural – i.e. we can
assume a baseline of knowledge which makes the interpretation possible – the
second is only a viable interpretation in certain conditions relative to a cultural
context. There must be a cultural agreement between the forms as signifiers
and what they are signifying.6

22
 Introduction

Function
Function is an expression of a building’s use, where the form of a building
supports a particular set of activities and where these activities are expressed
through a building’s typology. A building’s function is not only defined by action
and form, however, but also by institutions and naming conventions and it is
possible to subvert a building’s function by simply changing its name (Markus
1993: 12). New functional types are forming constantly, often as a result of
particular economic or social developments. For example, the technological
developments of the industrial revolution gave rise to the building typologies
of the cotton mill and railway station (ibid.) among many others, where unique
building types supported revolutionary forms of transportation and industry.
The functional programme of a building can also be considered strong or
weak depending on the neutrality of the building’s spaces. A hall for example,
in terms of form, may constitute a large unobstructed space and thus serve
multiple functions from meetings to indoor sports. In contrast, a cathedral has
a deeply engrained connection to symbolic forms which dictate particular cer-
emonial and ritual functions and are articulated by a pattern of fixed architectural
elements.

Space
Markus’s final analytical category, space, is a more elusive concept. Space
represents more than the gaps between built forms and, although space is
articulated by the materials of the built form, it is possible to discuss space
independently from form and function. The problem with the concept of space
is that it cannot be easily seen, or at least space is often encountered through
its effects rather than as an objectifiable phenomenon. To understand and to
visualize the effect of space on those who experience it, Markus calls upon a set
of theories derived from Space Syntax to map spatial relationships as topological
networks. Markus’s discussion of the use of these structural theories of space
suggests that, using his analytical classification, space is essentially defined
by the configuration of spatial cells and their relationships to other spatial cells
through a pattern of connections. Furthermore, these representations of space
encode information on the way space is perceived on the ground, through visible
associations between discreet spatial units which govern what, and importantly
who, can be seen from where.

Space and social meaning

Space is a somewhat slippery concept, nevertheless a growing awareness of
‘space’ as a component of architectural meaning has pervaded discourses both
inside and outside architecture, most notably through Lefebvre’s spatial philoso-
phy, which is exemplified in his book The Production of Space (1991). A common
thread running throughout this literature is the relationship between space and
social structure, where space is seen as constraining or generating the possibility
of human social interaction.
Architectural understandings of space have, perhaps not surprisingly,
been focused on the relationship between spaces and the material structures
that define them, but I will concentrate here on theories that instead seek to

23
Introduction

represent space in its own right, specifically theories related to Space Syntax,
which describe space as interconnected places. By focusing on these theories,
I do not wish to propose that they are the only theories of space and place to
emerge from architecture but rather that, through their clear articulation of space
as a structuring phenomenon, they provide a way of understanding and repre-
senting many of the ideas which follow.

A spatial syntax
Space Syntax emerges as one among a number of recent (post-1960s) spatial
theories of architecture that attempt to comprehend and to represent space in
order to analyze the morphology (the shape) of the built environment. Many of
these theories originate in Christopher Alexander’s work on design methods and
pattern languages. In a series of influential books and papers Alexander proposed
an understanding of design through a system of diagrammatic representation
which he describes in terms of ‘an abstract pattern of physical relationships
which resolves a small system of interacting and conflicting forces’ (Alexander
1964: v). In the context of the design of urban form these systems work at various
temporal scales from the dynamics of everyday movements (e.g. flows of people
and traffic) to the fixed or very slow-changing parts of the system (e.g. build-
ings and other ‘receptacles’) which form a system’s ‘invariant parts’ (Alexander
1965: 59).
In ‘A city is not a tree’ (1965), Alexander provides his clearest illustra-
tion of his theory of environmental structure by synthesizing models of social
interaction with structures of planned urban spaces. He argues that, despite
many apparent differences in their geometric appearance, many proposed
structures for planned urban space adopt the same hierarchical organization,
consisting of a hub and tree-like branches which form peripheries of decreasing
density and closed groups of spatial sets. Alexander proposes that these com-
mon spatial morphologies over-simplify the nature of the social structure of a
given society. Instead, when environments are not designed systematically but
grow organically, their structure follows a more natural pattern, reflecting the
semi lattice of social relationships between the individuals who are part of the
community. Alexander considers the physical patterns of streets to be essential
to the preservation of social networks in an environment and, furthermore,
suggests that street patterns can be an emergent manifestation of such social
structures.
Structuralist approaches to architectural theory have been devel-
oped extensively since Alexander’s early contributions, becoming a scientific
and mathematical discourse through figures such as Lionel March and Philip
Steadman.7 As a result, topological representations of the sort used, as illustra-
tion, by Alexander, have proven to be useful in mapping patterns of spaces and
social structure.
From Hillier and Leaman’s paper ‘How is design possible?’ (1974),
topological descriptions of space have been used as the basis for the develop-
ment of a fundamental ‘Syntax’ of space, which can account for aspects of
architectural experience. Space Syntax has been developed from combinato-
rial theories which describe space through the formal mathematical system

24
 Introduction

I.6
Diagrams to show the
development of a justified
graph from a building plan
where each room constitutes
a node in a network of
connected places

of topology and has developed from earlier work on graph representations of

building plans.8
Hillier and colleagues published three books elaborating the theory of
Space Syntax: The Social Logic of Space (Hillier and Hanson 1984), Space is the
Machine (Hillier 1996) and Decoding Houses and Homes (Hanson 1998). Peponis
describes Space Syntax, following Hillier and Hanson, as ‘a methodology, or a set
of techniques for the representation, quantification, and interpretation of spatial
configuration in buildings and settlements’ (Peponis 1997: 34.2).
Examples of these topological representations include, ‘access
graphs’, which emerged from the earliest forms of Space Syntax analysis and
relate to situations where a building can be divided into clearly defined rooms.
Such graphs label individual rooms as nodes, which are connected through
graph edges where access points exist between rooms. Such descriptions of a

25
Introduction

I.7
Diagram to show how
isovists are generated using
five different reference
points (here represented by
people) and the unobstructed
views they have of their
environment

building plan capture the connectivity of different rooms and allow analysis to be
performed on the plans of existing and proposed buildings. Specifically, justified
graphs or j-graphs are used where the entrance point to a building is considered
to be a root node of the building and, by counting the number of edges which
need to be traversed to get to a particular room in the building plan, its ‘depth’
within the plan can be measured (Figure I.6).
Such representations are built by referring specifically to the way
space is perceived through isovists, which are ‘polygons that capture spatial
properties by describing the visible area from a given observation point’ (Franz
et al. 2005: 33) (Figure I.7). In large-scale open environments, where there is no
clear delineation between different spatial cells, isovists are used to generate
axial line maps, where each axis represents a clear unobstructed view and cross-
ings between axes are links between these discreet spatial cells.
Such topological descriptions of an environment can be analyzed by
examining the relationships of individual spaces to the whole system of spatial
units and inferring certain properties of spaces and their interconnectedness

26
 Introduction

based on this analysis. In individual buildings, measurements of depth, i.e. the

number of thresholds one must cross from a given justification point (usually
taken from the front door), provide information on public and private realms and
on security and control within a sequence of spaces.
Space syntax has now become a sub-field of architectural theory,
although it is not without its critics.9 However, aside from the methodological
problems of making wide inferences based upon such limited descriptions of
architectural space, Space Syntax and topological methods more generally pro-
vide compelling representations of space. These diagrammatic representations
emphasize the perception of space, on the ground, through the visual experience
of moving through it and describe how, by dividing the world into spatial cells,
architectural structures control this visual experience and physical interaction in
space. In other words, these representations illustrate the effect that the fixed
elements of architectural form have on the dynamic patterns of spatial life and
how space encodes information on social structure.

A language of space?
It is tempting to view Markus’s three categories in the context of Space Syntax
as aspects of an architectural language where form is akin to grammar, function
is akin to semantics and space is given an explicit syntax. Markus does not, him-
self, evoke the language analogy but I want to take this a little further. Syntax, as
a spatial phenomena, shares properties with syntax as a linguistic phenomena,
i.e. as a rule set for the logical placement and structuring of units (e.g. words or
spaces). A correct syntax in language is not a guarantee of meaning, it is possible
to have a syntactically correct sentence that has no meaning, however, without
good syntax, language looses a sense of order and the meaning being conveyed
becomes more difficult to extract. Meaning itself is contained in the words and
their appropriate grammatical organization. Syntax defines the combinatorial pat-
tern of words so that they appropriately express meaning. Using this analogy,
space does not encode information as such but structures the way in which
information is experienced.

Semantics in architecture
This book will not follow the Space Syntax methodology explicitly but rather
seeks, through an appreciation of architecture’s topological structure, an under-
standing of how other types of topology can be and often are considered as
architectural. To this end, it is also worth drawing the reader’s attention to
the recent work of Psarra (2009) who uses Space Syntax in an explicit way to
understanding the relationship between architecture and narrative. Psarra uses
examples drawn from many different architectural types and eras, including
architectural fictions such as Borges ‘Library of Babel’. Whist Psarra’s focus on
the real environment allows her a richer analysis than I will attempt here, the
central tenet of both her work and this book is the same; namely, that meaning
is structured in architecture through topological relationships between places as
experienced when we move through space. Meanings are culturally specific but
the syntax of space as a structuring phenomena is universal.

27
Introduction

Part 5: The multi-disciplinary context

I have shown how architectural theories may frame the questions raised in
this book but, alone, they are not enough to tackle it. This topic is necessarily
trans-disciplinary, with all the dangers that this entails. During the course of the
book, I will borrow ideas from anthropology, philosophy, linguistics, the history
of science and technology as well as a number of the core disciplines of human–
computer interaction, information visualisation, cognitive psychology and studies
of digital media. I will elaborate on these core disciplines here.

Human–computer interaction (HCI)

Human–computer interaction (HCI) has grown from its origin as an offshoot of
computer science and deals with the ‘human factors’ in computing. HCI also devel-
oped from pre-computation research into human–machine interaction (particularly
vehicles and various military machines) and evolved as a discipline to encompass
both design principles and research into human cognition. HCI has now grown to
become a major topic of research and is central to computer science.
The work of Lucy Suchman, exemplified in Plans and Situated Actions
(1987), has had a significant impact on HCI, revealing the contextual nature of
interaction and highlighting the importance of the dialogue between human
and machine and the problems of trying to understand humans as a predict-
able component of a closed mechanized system. In response, an alternative
view of HCI has grown to include many other disciplines including psychology,
anthropology, sociology and ethnography and has more recently given birth to
another discipline, interaction design, which deals with the design of digital and
computational artefacts as well as their usability and evaluation. There are a
large number of publications devoted to HCI and interaction design, including
text books,10 journals and, most notably, the proceedings of ACM’s Computer
Human Interaction Conference, CHI, the content of which reflects the changing
trends in the HCI discipline.
Another dominant theme in the field of HCI has been the search for
more intuitive interfaces. It is thought that the ideal computer interface is one
that does not require users to learn the controls and procedures in advance but
which enables them to ‘naturally’ interact with the computer. The use of the
term ‘intuition’ in relation to interface design, it has been argued, is something
of a misnomer since ‘intuition’ is knowledge acquired without prior experience
or rational thought (Raskin 2000: 150). No computer interface can claim to allow
thought-transgressing interaction; instead it is necessary to try to relate the
actions on the screen to already learned actions in real life.

Embodied computing
A recent trend in HCI has been in the development of ‘embodied computing’.
Embodied computing attempts to claim back the world of bits and re-materialize
it as atoms through tangible interfaces, moving the site of interaction into physi-
cal rather than virtual environments:

When the Association for Computing Machinery (ACM), the

world’s largest membership organization of information technology

28
 Introduction

researchers, launched a general-readership publication named

Ubiquity, and called its plenary conference ‘After Cyberspace’, the
paradigm-shift had become more or less official.
(McCullough 2004: 5)

As well as a literal, technological challenge, however, this book addresses the

importance of the paradigm of embodied computing in graphical environments
which have previously been considered disembodied and without physical
manifestation.

Information visualisation (InfoVis)

The development of information visualisation (InfoVis) is concerned with the
representation of information in onscreen computer displays and particularly the
visualisation of the non-visual, that is, data structures, which are not naturally
mappable in that they don’t have geographical coordinates or values that can be
plotted against axes. InfoVis has now grown into an international research topic
and encompasses elements of HCI, information retrieval (IR), user interface
design, graphic design and the study of diagrams and covers, in particular:

• visualization of data structures

• visualizing databases
• statistical and numerical data
• information workspaces.

InfoVis has grown to include consideration of a large range of information types

and interface systems, including the display of geographic information in GIS
systems, the viewing of statistical data and the querying of databases, and it
involves both 3D and 2D applications. Collected key papers can be found in Card
et al. (1999) and Bederson and Shneiderman (2002) and more general text books
include Spence (2001), Ware (2004) and Chen (2004). InfoVis is also the subject
of two regular international conferences held in Europe and the US and a Journal
of Information Visualisation published quarterly since 2004.
This book (particularly Chapter 3) deals specifically with systems that
Wise et al. (1995) describe as ‘visualising the non-visual’ (p. 51) which are, in
turn, ‘concerned primarily with the task of gaining insight into information which
exists in the form of text’ (Spence 2001: 175) and the use of spatial mappings
for such representations:

[I]t has been recognized that the process of spatialization – where

a spatial, map-like structure is applied to data where no inherent or
obvious one exists – can provide an interpretable structure to other
types of data. In essence, maps and spatialization exploit the mind’s
ability to more readily see complex relationships in images.
(Wise et al. 1995: 51)

In the development of modes of information representation in virtual envi-

ronments, the concept of spatialization has been commonly used. In the

29
Introduction

first comprehensive attempt at cataloguing these visualizations, The Atlas of

Cyberspace (2001a), the authors, Dodge and Kitchin, break these visualizations
into four categories: infrastructure and traffic; the web; conversation and com-
munity; and the imagination.
In addition to abstract data visualizations, more metaphor-laden virtual
environments have often been used, particularly in the context of web-based
information:

Web-based information is dispersed around the planet in geographi-

cal space, along certain vectors. This information is experienced, or
at least accessed via a ‘browser’, a type of graphical user interface,
a flaneuresque shopping metaphor predicated on two-dimensional
pages which allude to the architectural space of shopping malls,
arcades, libraries, galleries and the like.
(Dallo quoted in Munt 2001: 58)

This book will focus specifically on the visualization of information and on the
process of interacting with such systems through spatial representations using
two concepts: navigation and screen space.

Navigation
This is not navigation in terms of physical movement but rather in terms of the
visual traversal of information space. Ware (2004) details the cognitive features
of what he describes as ‘browsing’ a data set, suggesting that understanding and
fully taking in a visual scene is a cycle of navigation, which involves interpreting
the image, formulating a browsing/navigating strategy based on this interpre-
tation and then interacting appropriately. This must be facilitated by the user
having a visual image that supports the formation of a cognitive map (an internal
representation of the external stimuli) and by onscreen actions that articulate,
often through animation, changes to the onscreen representation.
A central concern of many InfoVis designers has been to get large
amounts of information onto a very small screen:

Our present systems of navigation, never satisfactory in the first

place, are completely inadequate in the face of terabytes of informa-
tion we have to scan. But people and animals have been navigating
through complex environments for a millennia and have some useful
techniques for doing so.
(Raskin 2000: 141)

In on-screen environments, navigation means the traversal of the screen space

from one location to another and can involve acts as simple as vertically scroll-
ing a window in a document or navigating in real-time 3D virtual environments.
Navigating in an on-screen environment is required because not all of the envi-
ronment can be seen at once so users must be able to relate their view to their
position within the entire scene. This can be achieved simply, in the case of a
document in Microsoft Word for example, by using the scroll bar to locate the

30
 Introduction

view in the scrolled window. For more complex environments, however, the
issue of navigation becomes more difficult.

Screen space
Separate from traversal methods, screen space can also be used to define the
type of representation. For example, Card draws distinctions between what
he terms 1D, 2D and 3D displays, but also separates the dimensionality of the
display’s content from the representation system itself. He makes reference to
the Document Lens (Robertson et al. 1993):

which uses a focus+context technique to distort the 2D surface that

holds an array of pages from a document … This visualisation involves
1D data (a sequence of document pages), a 2D visual structure (the
surface) and a 3D spatial substrate (for distortion).
(Card et al. 1999)

The interplay between visual images and interactional spaces will become a
central concern of this book.

Cognitive psychology
The term cognitive psychology was first used in Ulrik Neisser’s (1967) book
entitled Cognitive Psychology and refers to ‘the processes by which the sensory
input is transformed, reduced, elaborated, stored, recovered’ (Neisser quoted in
Friedenberg and Silverman 2005: 95). Cognitive psychology is central to research
in both InfoVis and HCI, and in this book it will be used to bind these disciplines
together. In particular, the two concepts below (embodied mind and cognitive
mapping) will be considered in some detail.

Embodied mind
Extending the notion of embodiment as it relates to human computer interaction,
this book follows the work of George Lakoff and Mark Johnson, who state that:

The mind is inherently embodied.

Thought is mostly unconscious.
Abstract concepts are largely metaphorical.
(Lakoff and Johnson 1999: 3)

Lakoff and Johnson suggest that thought is governed by certain mental frame-
works or ‘schemas’, which, to some extent, precondition thought processes. At
a fundamental level, parts of the brain which control our motor senses are also
responsible for higher thinking and abstraction. Lakoff and Johnson suggest,
therefore, that ‘there is no absolute perceptual/conceptual distinction’ (1999: 39).
Evidence for this lies in our use of spatial metaphors to conceptualize abstract
concepts and it will be suggested here that the arrangements of objects and
their spatial relationships carry metaphorical significance based on movement
and space-based conceptions:

31
Introduction

Spatial representations have a vectoral form in the nervous system

itself. The nervous system’s commitment to representing stimu-
lus properties in vector spaces of modest dimensionality extends
beyond the representation of spatial properties to encompass other
stimulus.
(Gallistel quoted in Gardenfors 2000: 53)

Although Lakoff and Johnson are the focus of this study, Varela et al. introduce
many of the key concepts used to describe the embodied mind. These are set
out in their book Embodied Mind: Cognitive Science and Human Experience
(1993), and this framework is useful in the context of this text because it defines
a relationship between abstract reasoning and physical action and experience
and, therefore, helps to account for many of the spatial metaphors which are
inherent in information visualisations and graphical displays.

Cognitive mapping
The cognitive map, which was introduced by Tolman in his article ‘The cognitive
map in rats and men’ (1948), refers to the encoding of large-scale environments
into memory and the use of such memories to aid navigation. Since Tolman, the
study of the processes involved in cognitive mapping has flourished. Of particular
influence has been Kevin Lynch’s The Image of the City (1960), which centred on
the environmental component of spatial memory, introducing the term ‘imaga-
bility’ and the discussion of spatial cognition into the discourse of architects
and urban planners. More recently, cognitive mapping has been challenged by
the notion of the ‘cognitive collage’ (Tversky 2001: 12.1), which recognizes the
multi-sensory nature of spatial memory and the pre-perceptual structures that
lead to memory creation.
This book will address the problem of cognitive mapping in screen-
based environments where the screen is not large enough or of a high enough
resolution to support a view of the environment as a whole, acknowledging the
difference between real-world spatial cognition and the cognition of abstract
virtual environments.

New media studies

Studies of ‘new media’ originate from research into more traditional media and
communications. Much work on new media stems from the influential works
of Marshall McLuhan, in particular his book Understanding Media (2001), first
published in 1964, which introduced his famous statement ‘The medium is the
message’. The rise of computing has seen the redefinition of digital technology
from a tool to a medium and, in particular, it has been characterized by the pro-
cess of remediation (Bolter and Gromala 2000), particularly in the acquisition of
textual media by digital technologies.
New media is characterized by Lister (2003: 13) through:

• its propensity for mimesis, simulating aspects of other media through

‘virtual’ representations

32
 Introduction

• its distributed nature, allowing information to be transferred across great

distances instantaneously
• its interactive potential, allowing users to effect or change the system based
on their inputs
• its so-called hypertextuality, allowing for the nonlinear arrangement and free
navigation of textual objects.

Jay Bolter, an influential new media theorist, defines a key feature of digital media
by stating that ‘Like the printed book, film, and television before it, the computer
is not a neutral space for conveying information’ (Bolter and Gromala 2003: 77).
The process of remediation and the recognition of digitality as a
medium with mimetic and interactive potential are central to this book and,
although the approach taken here is not entirely technocentric, the characteris-
tics of the digital medium are, it will be suggested, central to how meaning is
conveyed. However, where Bolter and others have concentrated on language
and the remediation of text, it will be suggested here that there is an equally
important process associated with the remediation of space and architecture.

Part 6: Structure and content

After this Introduction, the work is split into four further chapters and a conclu-
sion. Each chapter is thematically distinct but builds on the theoretical concepts
developed in the previous chapter. The chapters also represent a (dotted) time-
line starting with the pre-digital era of information architecture (Chapter 1) and
travelling through the early history of computing and hypermedia (Chapter 2),
current screen-based information systems (Chapter 3) and ubiquitous and per-
vasive computing technology (Chapter 4). This timeline structure should not be
mistaken for a history of digital media and I don’t offer a comprehensive historical
analysis. Instead, the timeline allows me to approach the issue of the increasing
complexity of computational information systems from their basics.
Chapter 1 starts with a foundational account of what I will describe as
the ‘architectonic system’. Combining observations on the origins of architecture,
the emergence of categorization as central to human communication and evidence
of the conceptual basis for human thought, the chapter will examine how our
spatial environment is able to shape us and how we shape our environment as
a product of conceptual thought and memory. The chapter will cover topics as
diverse as the anthropological basis for systems of categorization, the cognitive
basis for knowledge representation and the practice of mnemotechnics and the
method of loci. The architectonic system, it will be shown, contrasts to the linguistic
system as a means of communicating by structuring spaces rather than sounds.
Chapter 2 takes the place-based ‘architectonic system’ described
in Chapter 1 and charts a number of attempts to move beyond the restrictions
of place by introducing the possibility for mechanical change in the organiza-
tion of information. The chapter introduces the concepts of permutation and
combination systems as a basis for the expression of meaningful relationships
through a discussion of the logic machines invented by Ramon Lull and looks
at the designs for a universal archive called the ‘Mundeneum’ developed by
Paul Otlet in the first half of the twentieth century. The Mundeneum bridged a

33
Introduction

material architectonic world and a new understanding of non-materially bound

information. The ability to transcend the limitations of topology found its ultimate
expression through the development of the computer and hypermedia, which
cause a separation of information from its material means. However, Chapter 2
will show that the digital revolution in information has seen a proliferation of
information spaces. Furthermore I will suggest that these information spaces
have been articulated by metaphors of the built environment and constrained to
geometrically limited topological structures.
Chapter 3 attempts to make sense of the multiple ‘information
spaces’ provided by digital information systems and to consider what may be
architectural about them. To this end, the chapter will develop definitions of
semantic space, screen space, and interaction space before developing a defini-
tion of ‘architectonic space’ in computer-based information, which ties in with the
definition articulated in Chapter 1. Chapter 3 will show that our attitude to and
definitions of information space depend on a conceptual embodied relationship
to the information as we interact with it.
Chapter 4 reflects on the consequences for the architectonic system
in an era of ubiquitous computing. It will be shown that, where metaphors of
architecture were used in the discussion of information space architects have
now started to use the metaphor of computing to describe the built environ-
ment. This chapter will examine the relationship between objects organized in
space where these objects are smart and their position in the world does not
necessarily coincide with their conceptual position in an undeclared virtual space.

The beginning
This book will set the scene for a battle between Babel and the Theatre of
Memory; between the will to communicate everything and the fear of not being
able to find or understand anything; between those who take comfort in discur-
sive formations of material organization and those who detest their limitations. At
the centre of it all we have the individual, navigating his or her way through the
no-man’s land of an ever growing territory of information. This is not a theoretical
battle, it is happening on computer screens every day. The Theatre of Memory
and the Library of Babel exist in contemporary society. Every time we access
information we enact, in some way, the story of the librarian and the scholar. We
enact these stories in a Theatre of Babel and I hope to show that our conceptual
world is, at least in part, constructed by a space that is physical, conceptual and
ultimately architectural, even when it is not material.

34
Chapter 1

The architectonic
system
Architecture organizes space surrounding man. It organizes this space as
a whole and with respect to man in his entirety, that is with respect to all
the physical or psychic actions of which man is capable …
(Mukařovský 1977: 240)

Part 1: Architecture and meaning

When we observe architecture in the twenty-first century, it is obvious to suggest
that buildings provide more than basic shelter. We know that the built environ-
ment of any modern human habitation is a cacophony of spatial compositions,
ornaments, styles and typologies. We know that the rich complexity encoded
in the built environment provides ample material for architectural critics, histo-
rians, theorists, anthropologists, sociologists and countless others. We know
that buildings are among the most sophisticated objects that humans create.
We know that buildings can express our political, moral and social ideals, but do
we understand how? Is architecture a medium for information? Are buildings a
language of communication?
Taken in their entirety, architectural ideas constitute an unimaginably
large corpus of material, with varying degrees of relevance to this enquiry. A
fully comprehensive approach would certainly overburden this relatively short
book, but selectivity is challenging as, to borrow a term from the Russian literary
theorist Jan Mukařovský, architecture exists on a number of ‘functional horizons’
(Mukařovský 1977: 241). It is not possible to look at a building as a functional object
related directly to its current use without cross-referencing with the assumptions
derived from historical typologies and norms, the social context in which design
decisions were made and the designer’s own creative process (Mukařovský
1977: 241–2). If we attempt to unpick any one of these functional horizons by, for
example, isolating a building’s practical function, then our understanding of the
building may unravel. Such a holistic view of architecture is, however, problematic
in relation to this enquiry, which simultaneously seeks a broader understanding
of architecture beyond the built environment and a more focused understanding
of space as a particular means of structuring communication.
This chapter will address this problem of architectural complexity by
seeking a fundamental understanding of how the human mind shapes space
and, conversely how the human mind is shaped by space at primitive, cognitive
The architectonic system

and conceptual levels. A range of sources will be used around common themes
of pre-historic architecture, architectural languages, cognitive architecture and
architectonic systems.

Pre-historic architecture
The forms and structures we now conceive as being ‘architectural’ have evolved
over time. That is to say that, using a crude but illustrative analogy with biological
processes, more complex architectural organisms have emerged from earlier
and simpler forms. A book on modern human anatomy could easily stretch to
thousands of pages, but, trace the human ancestral line back far enough, and we
arrive at the single-celled organism that represents our oldest relative and which
may be described in equivalent detail on one page. Recorded architectural history
is relatively recent and historical narratives tend to be limited to formal ‘civilized’
architectures which occur some distance up the architectural evolutionary tree.1
Archaeologists, however, are often interested in the earliest forms of settlement
and anthropologists have, in parallel, researched and written accounts of ‘primi-
tive’ settlements and communities that have been isolated from the progression
of formal civilizations. These accounts help build a picture of pre-historic archi-
tecture and the most basic motivations which drive us to articulate our spatial
environment. Through these studies, a lens is created through which we can
view the fundamental human need to pattern our environments.

Architectural languages
From these foundational accounts of the built environment, the notion of a ‘lan-
guage’ of architecture will emerge. To some extent, ‘language’ in this context is
simply another analogy. Like biological organisms, languages change over time,
becoming more complex and developing new ‘families’ derived from common
ancestors. Languages also encode a sort of DNA, allowing commonly inherited
features such as sounds, grammars and syntaxes to be traced from their govern-
ing structures. Applying the language analogy to architecture further implies that
information content can be written in stone and read in the form of architectural
space. This way of talking about architecture emerges from a structuralist tradi-
tion of philosophical thought and, while there is a tradition relating to various
historical ‘languages of architecture’ through description of stylistic convention,
the structuralist view of architecture attempts to uncover space as a communica-
tive medium by understanding, not how particular forms of architecture relate to
particular meanings, but rather, following the linguistic teaching of Ferdinand de
Saussure (1857–1913),2 how a language or meanings is enabled by a particular
structural system. What I hope to show is that architecture and language have
developed from the same origin and remain intertwined but essentially distinct.
Following the structuralist theme, I will draw on sources which read social
structure as it is written in the form and configuration of the built environment.

Cognitive architecture
As well as addressing the emergence of forms which humans have developed
to shape and pattern their environment, this chapter will also develop an account
of a cognitive architecture by looking at theories on how the brain structures

36
 The architectonic system

knowledge by defining patterns of ideas, both consciously and subconsciously.

These descriptions of knowledge representation, which are borrowed from cog-
nitive psychology, reveal the bind between the cognitive propensity to spatialize
knowledge and the imprints of that knowledge as articulated in the spatial world.

Architectonic systems
Building on these first three themes, I then wish to develop a fourth, which
isolates a particular type of ‘architectural’ thought and moves the discussion of
architecture beyond the concrete material of the built environment and toward
something much broader. This ‘something’ I will describe as ‘architectonics’,
which exists because of what Kojin Karatani describes as the ‘will to architecture’
(Karatani 1995: 5). This ‘will to architecture’ defines a particular view of, not only
how and why we articulate physical space through buildings, but also how we
impose structure on our mental classifications of the world through recourse
to architectural metaphor. By understanding the origin and affect of the ‘will to
architecture’, the domain of architectonics is revealed as something which simul-
taneously shapes and is shaped by the world as it exists and as it is perceived.
Following this introduction, this chapter is split into four parts. Parts 2
and 3 tackle the concept of the architectonic system from two different angles.
Part 2 starts with a notion of spatial organization shaping the human mind and
discusses the relationship between early forms of architecture and so-called
‘primitive’ cultures and settlements and deals with the emergence of early
types of classification. It will be argued in this section that the human propen-
sity to structure our world into abstract classification systems is a result of the
organization of society into distinct groups of individuals with specific roles and
responsibilities. Such social groupings are then made manifest in the organization
of social space; through the configuration of rooms in a dwelling or the organiza-
tion of buildings in a whole village. Such patterns have been read extensively by
structural anthropologists following the work on the evolution of ‘primitive clas-
sification’ by Claude Levi-Strauss (1963; 1995) and Emile Durkheim and Marcel
Mauss (1963). By uncovering the deep connection between social space and
the emergence of classification, this section will seek to move the discussion of
architecture and meaning beyond the language of form and towards a separate
concept of space and its organization. In turn, the built form as an articulation of
space will be described as a means of structuring our mental as well as physical
worlds.
Whereas Part 2 starts with physical space and its role in shaping
mental spaces, Part 3 approaches architectonic systems from the opposite
point of view. This section will investigate how knowledge is represented in
the mind. The domain of knowledge representation has a long history and it is
studied in philosophy and, more recently, in cognitive psychology, neuroscience
and computer science. The study of knowledge representation has the aim of
both helping our understanding of the human mind and enabling systems to be
built that are capable of simulating aspects of the human mind through ‘artificial
intelligence’. This section will build on earlier studies of knowledge representa-
tion by distinguishing a model based on the three levels of cognitive structure
as they are currently understood: symbolic, conceptual and subconceptual

37
The architectonic system

representations. Focusing on ‘conceptual’ representations, it will be shown

that, with reference to the work of Peter Gardenfors (2000), types of knowledge
representations called ‘conceptual spaces’ exist and provide a bridge between
the way our bodily senses receive information about the world and the pattern-
ing of knowledge in the mind. By elucidating the theory of conceptual spaces, it
will be shown that the mind cannot be considered something which is entirely
abstracted from our physical interactions with the world. Furthermore, drawing
on the work of George Lakoff and Mark Johnson (Lakoff 1990, Johnson 1990,
Lakoff and Johnson 1980, 1999) on the problem of knowledge representations,
I will show how human understanding of abstract concepts is often constrained
by the representations we use to describe them.
Part 4 will develop an historical example of the unification of spatial
organization and knowledge representation by focusing on ‘the art of memory’,
as described by Francis Yates (2001), and the ‘method of loci’, which involves
the use of physical environments to help in the memorizing of abstract facts and
concepts. The method of loci leads to the development of new types of art and
architecture based on the human propensity to recollect journeys through physi-
cal spaces (both real and imagined). More than the creation of new art forms, the
art of memory left us with a legacy of philosophical thought based on a relation-
ship between topos (place) and topic (subject). This legacy can be traced through
the development of the manifestation of conceptual spaces in typologies such as
museums and libraries, which are designed to act as explicit knowledge repre-
sentations. These building typologies not only act as storehouses for objects but
organize those objects and our interaction with them. Museums and libraries are
the most obvious types of ‘information architecture’ and provide sophisticated
examples of conceptual spaces made real.
Part 5 will conclude by drawing together the threads of the discussion
on architectonic systems and will provide the historical and philosophical basis
for the enquiry in Chapter 2 into the evolution of digital technologies and the idea
of separating information from its material means.

Part 2: Architectonic systems and the emergence of

categorization
Beyond languages of space
One problem with any discourse on architectural space and meaning is that archi-
tectural space is such a complex phenomena. Furthermore, the term ‘space’ is
found in such a wide range of contexts that discussions of spatial meaning can
lead to wildly divergent conclusions depending on what sort of space is being
described. Indeed, wading through the literature on philosophies of space it is
often unclear which thinkers base their observations on real physical spaces and
which thinkers understand space simply as a useful metaphor to describe social
processes or more abstract ideas.3
In the introduction, I hinted at the possibility of a semantic and a
grammar of architecture. The notion of architectural language has been preva-
lent in many texts on architectural theory and history4 and many relate directly
to structuralist theories of language and linguistics.5 Nevertheless, the language
analogy is a problematic one, particularly when an attempt is made to imbue the

38
 The architectonic system

built environment with a level of articulateness akin to oral language. Searches

for a ‘language of architecture’, therefore, tend to be confined to analyses of
specific styles of architectural form where there is a formal and systematic rela-
tionship between architectural formalisms and intended meanings. The use of
the language analogy in architectural analysis also has the potential to become
anachronistic. The problem is that most ‘meanings’ in art, architecture or indeed
written and oral languages are culturally specific, so that the language must be
known before meaning can be inferred (whether spatial, visual or symbolic). In
other words, meanings can only be read effectively in a specific time and place
by someone initiated into the language. Linguistic meanings tend to change or
get ‘lost in translation’ and, as a result, linking a specific language model with
architecture becomes difficult and is likely to become irrelevant as cultural inter-
pretations change over time.
By isolating the idea of Space Syntax in the introduction, I have
deliberately attempted to define architecture as a structural phenomena, and
extending this notion I will now attempt to provide a counterpart to the language
analogy by separating two distinct systems, which I will describe as the linguistic
system and the architectonic system. To do this I will start with a discussion of
the earliest types of architecture and of how the patterning of the environment
may have lead to the first patterns of categorization and social organization.

The emergence of the architectonic system

What was the first architectural gesture? It seems tempting to identify the origin
of architecture with a need for human shelter and to understand early buildings
as temporary enclosures, one evolutionary step away from cave dwelling. We
have little or no archaeological evidence for the very first dwellings and it is likely
that those where remains exist are comparatively advanced examples. It is pos-
sible that, well before the development of the relatively complex technologies
necessary to build architectural structures capable of providing shelter, human
beings began to pattern their environment by creating structures and imprints
that were capable of defining spaces.6 It is also true that humankind has always
existed in the cradle of a natural architecture, created by the structural forms
of the landscape (Kostof 1995: 21). The earliest practice of architecture may,
therefore, have been to invest natural structures found in an environment with
symbolic meaning associated with ritual or simply to delineate territory.7
With the development of the technologies necessary to construct
structures which were recognisable as buildings, more complex organizations
become possible and archaeological excavations have tended to reveal, not
simple or isolated dwellings, but collections of buildings and configurations of
internal spaces indicating divisions of use and structures of community.8 These
early buildings not only provide sheltered space but articulate it in order to convey
information about the pattern of relationships within a settlement. Seen in the
context of cultures whose archaeological remains include highly ritualized struc-
tures such as stone circles and basic temple buildings, it is clear that even early
building types are not simply functional shelters. While the meanings behind
many pre-historic buildings and environmental structures may have been lost,
that architecture was used to articulate meaning is not in doubt, in the same way

39
The architectonic system

that it is possible to hypothesize that pre-historic languages existed without us

having to find material manifestations of them.

The architectonic system as distinct from language

The process of associating meaning with the articulation of the environment
through buildings or other environmental structures is described by the art histo-
rian and critical theorist, Donald Preziosi, in Architecture, Language and Meaning
(1979), as the ‘architectonic system’. Preziosi rejects discourses on ‘languages’
of architecture that relate particular forms, scales, limited typologies or orna-
ments to what he considers to be arbitrary categories and styles of the built
environment.9 Rather, Preziosi searches for the foundations of an architectonic
system distinct from the linguistic system which he suggests

incorporates the entire set of place-making orderings whereby indi-

viduals construct and communicate a conceptual world through the
use of palpable distinctions in formation … [which are] addressed
to the visual channel, to be decoded spatio-kinetically over
time.
(Preziosi 1979: 4)

This architectonic system is independent of the material or technology of con-

struction and goes beyond the notion of buildings to include ‘all manners of space
and place-making activities’ (ibid.). If Preziosi’s architectonic system is viewed in
parallel with the linguistic system, then the definition above would correspond to
a description of oral language as a sequenced string of audible sounds, articulated
by changes in pitch and volume made by one person and experienced over time
through the ears of another. In other words, the atomic units of the linguistic
system are sounds and the units of the architectonic system are spaces.
Forms of communication do not exist in isolation from one another.
For example, written and oral languages tend, at least in many Western lan-
guages, to have a direct equivalent in written representations, binding oral
gestures with visual symbols. Before the evolution of oral language, however,
Preziosi offers an account of human linguistic communication that stems from
early types of material manipulation and, specifically, the origins of tools.
Early tools, for example flint knives or scrapers, are more than func-
tional devices in that they relate an abstract form to a functional activity (e.g.
cutting through meat or vegetation). Recognition of a form, therefore, relates
the tool to its use and distinguishes it from other tools or objects that are not
tools. When human beings first picked up a physical object and shaped it to
perform a particular function, they encoded a meaning within the object in addi-
tion to its material properties. Simply showing the tool would, in itself, be akin
to a verbal instruction conveying the intention of completing a particular physical
task. As tools became complex and part of ‘tool assemblages’ (Preziosi 1979:
20), early humans become articulate, not only able to shape the physical world
around them but also to build up descriptions of complex activities through the
organization of those tools and the actions they symbolized. As new tools used
to manufacture other tools evolved, the nature of the relationship between object

40
 The architectonic system

and action became increasingly abstract, and symbolic correspondences began

to form, which had only arbitrary relationships to the actions they represented.
These symbolic assemblages include oral languages, where the articulation of
sound becomes a tool in itself. Each stage in the process of language develop-
ment involves a degree of separation from the physical reality of the object or
action the language represents. However, the formation of a communication
system cannot be seen independently from the material technologies that
shape it.10
Buildings need to be seen separately from tools, although the same
notion of material articulation and its relationship to abstraction can be observed.
Where a hand tool is directly related to an action or set of actions, i.e. a verb, a
building can also be seen as a noun. A building can both denote and can classify
an activity. For example, it can turn a place for eating and sleeping into a ‘home’
but also, by defining territories, it can classify the contents of a place, whether
it be people or objects. While Preziosi focuses on the verb of architecture, what
makes the architectonic system distinct from the linguistic system is its articula-
tion of nouns.

The architectonic system and classification

Tools and buildings constitute the foundations for two intertwined but distinct
forms of communication, the linguistic and architectonic systems. Where the
linguistic system is well understood (or at least well researched), the ‘architec-
tonic system’ is much less understood and, indeed, architecture as a form of
communication in its own right is contested by a number of theorists, not least
because the architectonic system has remained in many accounts inexorably
linked to language.11
The challenge for linguistic or semiotic theories of architecture is
to analyze architectural form in the same way as one would a verbal language.
However, since the architectonic system does not have the same origin as the
linguistic system should it be treated differently? And, if so, how?
One rich vein of research on architectonic systems emerges from
structural anthropology and the writings of Marcel Mauss, Emile Durkheim and
Claude Levi-Strauss. While these authors do not use the term architectonic
system explicitly, their understanding of the built environment as an expression
of meaning is clearly distinct and relevant. In developing structural anthropology,
Levi-Strauss attempted to understand commonality across apparently diverse
cultures by cross-referencing evidence of social structure, ritual art practice,
oral testimony and architecture. Material manifestations, in his view, became
‘projections’ or ‘reflections’ of mental processes (Hillier 1996: 241). Less well
referenced, in architectural discourses, is the proceeding work of Durkheim and
Mauss, which Levi-Strauss has cited as a key influence (Levi-Strauss 1963: 5).
While they never directly address the notion of an architectonic system, their
paper Primitive Classification (Durkheim and Mauss 1963)12 does give some
clues as to how the architectonic system may have evolved as a form of com-
munication, distinct from oral language.
For Durkheim and Mauss, the absence of written accounts of certain
‘primitive’ cultures means that we must ‘read’ other material manifestations to

41
The architectonic system

understand them, notably the built environment. Even in what might be consid-
ered ‘primitive’ settlements, there is often a complex relationship between basic
functions (providing shelter and material security) and the use of a spatial order to
define social space. Simply put, the built form has a role in articulating territory,
in defining who can be where by separating, at a macro scale, the territories of
whole tribes and, at a micro scale, the hierarchy of seniority of social groups and
divisions between family members based on age, sex, etc. While the role of
space and territoriality in defining a social system is not, in itself, a dramatically
original insight, Durkheim and Mauss set out with the ambition of accounting
for the emergence of the human capability and propensity to classify the world.
While philosophers such as Aristotle and Plato are considered to be the fathers
of our understanding of scientific classification by grouping concepts or objects
based on shared properties, Durkhiem and Mauss suggest that the concept of
classification, itself, as it relates to a broad human propensity to subdivide the
world, is neither an invention of Aristotle nor a natural/pre-programmed state
of mind:

[H]umanity in the beginning lacks the most indispensable conditions

for the classificatory function. Further, it is enough to examine the
very idea of classification to understand that man could not have
found essential elements in himself. A class is a group of things; and
things do not present themselves to observation grouped in such
a way. We may well perceive, more or less vaguely, their resem-
blances. But the simple fact of these resemblances is not enough to
explain how we are led to group things which resemble each other,
to bring them together in a sort of ideal sphere, enclosed by definite
limits, which we call a class, a species, etc.
(Durkheim and Mauss 1963: 7–8)

Durkheim and Mauss make their case with reference to studies of three con-
temporary ‘primitive’ cultures, from North America, Australia and China. While
the tribes they described have evolved without contact with one another, their
systems of classification are remarkably similar and originate from a separation of
their territory into regions of space governed by polar orientation. Classification,
Durkheim and Mauss suggest, emerged because, as social groups grew and
extended, they were seen as distinct both in terms of their social identity and
their spatial location (Figure 1.1):

It was because men were grouped, and thought of themselves in the

form of groups, that in their ideas they grouped other things, and in
the beginning the two modes of grouping were merged to the point
of being indistinct. Moieties were the first genera; clans, the first
species. Things were thought to be integral parts of society, and it
was their place in society which determined their place in nature.
(Durkheim and Mauss 1963: 82)

42
 The architectonic system

‘Society first’ approach to spatial meaning

Levi-Strauss also observed structural similarities between apparently diverse
social groups though his pioneering work on structural anthropology. While a
plethora of different cultural practices could exist between separate societies,
Levi-Strauss suggested that there is a recurrence of certain structural themes
that do not emerge through inheritance (i.e. they are not passed through cultural
transmission) but arise because of a limited number of ‘institutional possibilities’
(Levi-Strauss 1963: 133).
In one notable example, Levi-Strauss provided evidence for what
he called ‘dual organization’, with reference to native cultures in Indonesia and
America. Dual organizations refer to social groups who naturally tend to divide
themselves into two sub-groups so that, for example, the American Winnebago
tribe divided their society into wangeregi ‘those who are above’ and manegi
‘those who are below’ (Levi-Strauss 1963: 133). Citing the work of Radin (1923),
Levis-Strauss isolates two recurring village structures that articulate dual organi-
zations, which he describes as:

1 the concentric structure, consisting of inner and outer circles (Figure 1.2a)
(Levi-Strauss 1963: 135)
2 the diametric structure, consisting of a circle bisected in the middle separat-
ing the two dual functions (Figure 1.2b).
These patterns do not necessarily exist in isolation, indeed in the
case of the Winnebago tribe both seem to exist simultaneously; however, duali-
ties form the starting point for classification by first separating one group from
another and then making that pattern visible through the distribution of buildings
within a village. In the first instance, Levi-Strauss suggests, dual organization is
reflected in patterns of social life around certain basic practices, for example the
storage of cooked or raw food,13 the division of society into men and women
or adults and children (Levi-Strauss 1963: 137). Such practical organizations are
imbued with symbolic significance and become part of a structural belief system
separating sacred and profane, married and celibate, etc. These dualisms appear
to emerge from ‘aspects of the physical world or between moral or metaphysical
attributes’ and through apparent thematic oppositions (Levi-Strauss 1963: 141).
Dual organizations are, therefore, not only manifest in relationships in society
but are expressed through a particular way of articulating our material world,
notably through the appearance of variations on diametric and concentric village
patterns. Such structural patterns both describe and make possible daily ritual
and the organization of conceptual and physical space.
The importance of Levi-Strauss’s structural methods cannot be over-
estimated and they have had a profound impact on fields outside anthropology.
Perhaps because architecture is the most observable patterning of social struc-
tures into material form, Levi-Strauss’s work has found favour with architectural
theorists attempting to move beyond art historical analysis of buildings toward
something more akin to a science of architectural form. However, the translation
of Levi-Strauss’s work into the field of architecture is not without its problems,
notably the fact that the idealized forms described by Levi-Strauss are far from
universal.

43
The architectonic system

1.1
The first types of category
formed in the human
mind as a result of the
division of social groups,
their distribution in the
environment and their
association with a society of
objects and concepts

‘Space first’ approach to social meaning

Where Levi-Strauss’s work may be described as society first, reading spatial
configuration as a by-product of social structure, architectural approaches to
reading space tend to be space first – seeing space as an active agent in forming
social structure. Post-Levi-Strauss, a group of architectural theories has emerged
which tend to de-emphasize the geometry of the environment in terms of its
shape and orientation and to focus on its topology in terms of spaces and their
connection to one another.
I introduced topological representations of space in the introduction
through Space Syntax. A clear starting point for Space Syntax is the idea that
space can be analyzed in order to provide an understanding of social structures
and, to this end, Levi-Strauss is credited as the originator of the methods which
allow one to consider such structures systematically (Hillier 1996: 89–90).
However, a recognized problem, isolated by Levi-Strauss’s discourse on spatial
organization, is that, despite his proposal that there are specific identifiable mod-
els of, for example, village organization, there are many examples that contradict
these idealized representations:

Seen from a spatial point of view, societies vary, it seems not only in
the type of physical configuration, but also in the degree to which the
ordering of space appears as a conspicuous dimension of culture …

44
 The architectonic system

1.2
Diagrams to show two
common types of dual
organization in village
structure

Some societies appear to invest much more in the physical patterning

of space than others, while others have clear global, even geometric
forms; and some societies built a good deal of social significance
into spatial form by, for example, linking particular clans to particular
locations, while others have recognisable spatial forms, but lack any
obvious investment of social significance.
(Hillier and Hanson 1984: 5)

The Space Syntax approach to this problem, as elucidated in The Social Logic
of Space (Hillier and Hanson 1984), is not to seek top-down readable geometric
patterns on to which, for example, rituals and societal structures are literally
mapped on to (e.g. the disposition of dwellings in a village). Instead, Hillier and his
colleagues at University College London aggregate space using statistical models
to analyze spatial relationships, which are mapped as connected networks. Seen
in this way, architectural space is an emergent phenomena where individual
spaces are agents within a system of connected spatial sequences derived from
an individual’s need for visual and physical access to different parts of their spatial
world. In other words, these representations illustrate (in part) the effect that the
fixed elements of architectural form have on the dynamic patterns of spatial life.

The architectonic system

What constitutes an architectonic system? Based on the discourse described
above, it is possible to draw some tentative conclusions before tackling the
problem from a different angle in Part 3.
First, in understanding architecture and meaning, a clear distinction
must be drawn between oral language as the articulation of sound, written
language as the articulation of symbols, and architecture as the articulation of
space. It has been proposed here, using Preziosi’s discourse on architectural
meaning, that the linguistic system and the architectonic system have funda-
mentally different origins, where language is an extension of the ‘verbs’ of tool
use and the architectonic system is fundamentally associated with the ‘nouns’
of categorization. This does not discount the possibility of analysing architectural
meaning in terms of language, as both communication systems are entwined so
that, for example, buildings carry symbolic meanings, which allow them to be the
subject of semiotic analysis. Particular styles of architecture are also sufficiently

45
The architectonic system

consistent in their form to allow regular grammars to be uncovered. However,

the architectonic system, as described here, needs to be treated independently
from the notion of language.
Second, the concept of classification, which is a keystone of our
intellectual representations of the world, is not a natural product of the brain
but an affect of social humans organizing their world into discreet contained
spaces, separating different objects (initially people) based on their physical or
metaphysical characteristics. The process of the evolution of classification starts
with the division of social groups who are distributed territorially throughout a
community’s environment. Eventually this tribal classification becomes the basis
for an abstract organization of objects as well as people.
Third, following on from Durkhiem and Mauss (1963), material and
mental processes cannot be subdivided. The structures that govern thought are
inseparable from those that shape our physical environment. This theme will be
developed further in this chapter in relation to cognitive psychology and studies
of knowledge representation. However, what becomes clear through the diverse
structuralist literature on architecture and meaning, from Levi-Strauss through to
Hillier (1996) is that our mental processes and physical environments are closely
aligned and, whereas it might not be possible to simply read an environment to
understand the social structure of a given society, deep meaning can be uncov-
ered by understanding the relationship between the articulation of spaces and
the social processes enacted within them.
Finally, studies of topological architectural theory reveal a representa-
tion of space that discounts symbolic meanings encoded in geometric form and
describes space as topological, i.e. connected through networks of spatial cells.
Topological descriptions of space attempt to represent spatial configuration, as
it is perceived and built by an individual on the ground, as part of a larger system
rather than as a god’s eye view or a top-down plan. As will be shown, topological
descriptions of space seem to match with certain models which describe how
the brain organizes space and indeed, how the brain is capable of representing
knowledge.
Having defined the architectonic system as it relates to the built envi-
ronment, I will turn to the brain, starting this time with how the brain represents
and infers knowledge.

Part 3: Conceptual spaces: categories in the mind

The idea of an architectonic system is compelling and, coupling this with
Durkheim’s and Mauss’s (1963) discussion of the origins of classification, a
clearer distinction between the tools of language and the tools of space begins
to emerge. The concept of architectonic systems laid out in Part 2 of this chap-
ter, however, is far from proven. To add detail and weight to this proposition I
will look more at the nature of classification from a different angle. Rather than
starting with the concept of space or social structure, this section will start with
the mind or, more specifically, the embodied mind (see Figure 1.3). Mindful that
the reader may not be fully literate in cognitive psychology, this part begins with
a brief introduction to knowledge representation as it is currently studied.

46
 The architectonic system

Beyond Cartesian dualism

In the seventeenth century, Rene Descartes proposed a model of human thought
in which the mind and body were separate entities so that the corporeal world
could be treated independently from the mental processes of higher logic and
reasoning (see for example Descartes 1968).14 This philosophy forms part of what
we now call Cartesian dualism. However, a large and growing number of philoso-
phers and cognitive psychologists have challenged this view and the mind–body
problem is constantly being revisited, particularly in the post-digital era where
information and its representation and meaning become important in the design
of computational systems. The possibility of ‘artificial intelligence’, coupled with
new techniques for understanding how the human brain operates, notably various
forms of brain scanning, have lead to an interest in how the brain represents and
processes knowledge about the world. The key problems tackled by research into
knowledge representation within cognitive psychology are threefold.

1 How is knowledge about the world encoded and captured in memory?

2 How is it possible to infer knowledge about the world through a process
involving the recall of memory?
3 How can computational systems be built which simulate the way the mind
works with the aim of creating artificial intelligence (AI)?

Inference about the world is primarily encoded in the form of categories and
these categories occur at different levels. For example, in order to recognize an
object such as a chair, we must be able to match the properties of the object
we are observing with a mental category so that we recognize the object and
its function as distinct from other objects we may encounter. Furthermore, the
category for chair is, itself, a member of a parent category (furniture) and con-
tains within it many subcategories (dining room chair, couch, baby seat, etc.)
(Goldstein 2005: 269) (Figure 1.4). To recognize an object such as a chair, we
need to define the characteristics of that object so that we can infer to which
category it belongs and thus knowledge is encoded by binding together groups
of similar objects based on their shared properties. The study of knowledge
representation is the search for mental structures of categorization which allow
an agent (be it human or computational) to accumulate knowledge through a
process of inference based on the properties of entities within the world. We
not only make use of categories for simple objects, such as chairs, but our lives
are also filled with many abstract categories we use to frame our thoughts, only
a few of which we are ever explicitly aware of and this makes understanding
knowledge representation a particularly complex problem.

Symbolic, conceptual and sub-conceptual knowledge

representations
Within the domain of cognitive psychology there are three competing, although
not mutually exclusive, paradigms of knowledge representation summed up as
symbolic, subconceptual and conceptual representations (Gardenfors 2000: 1).

47
The architectonic system

1.3
In addition to the notion that
social groups distributed in
space articulate categories,
there is also evidence to
suggest that the mind
shapes abstract concepts
through spatial reasoning
through what are known as
conceptual spaces

Symbolic representations

Models that propose symbolic representation have evolved from classical sys-
tems of categorization, from the philosophies of Aristotle onwards, and have
developed through modern computational systems, which encode and logically
parse information in, primarily, hierarchical structures of inference.15 Symbolic
representations are an ‘implicit methodology’ and form the basis of much AI
research (Gardenfors 2000: 35).
In symbolic representations, structures of inference can be repre-
sented either through linear notation, in other words written language, or through
graphical notations involving visualizations of semantic networks, often charac-
terized by graph representations consisting of nodes and links. While examples
realized in computation didn’t emerge until the 1960s, examples of semantic
networks were developed as early as the thirteenth century.16 Using such graphi-
cal representations it is possible to infer the category of a particular object from
its properties by tracing down more detailed information from one category to
another in a process similar to the game of ‘Twenty Questions’, where the player
must discover the mystery object within 20 questions by asking a series of true/
false or either/or questions, usually starting with ‘Animal, vegetable or mineral?’.
These links are called predicates and encode properties that organize subjects
into different groups.
Semantic networks have distinct limitations, however. For example,
how is a coherent set of properties inferred for the category of chair? We might
suggest that the category ‘chair’ is defined by a shared set of properties or
predicates ‘a piece of furniture consisting of a seat, legs, back and often arms’.
While this definition of the chair category may suffice for many chairs, it won’t
work for all. Objects such as baby seats and bean bags may also be considered

48
 The architectonic system

1.4
Diagram to show the process
of inference based on the
categories for different types
of chair

to be chairs but do not fit all the criteria of the chair category (Goldstein 2005:
269). The category ‘chair’ may therefore be fuzzy. Where hierarchy becomes
conceptually fuzzy, it is not possible to identify a chair through simple inference
based on a set of universal properties of chair. Other models for symbolic repre-
sentation that extend the notion of predicates and semantic linking, for example
Collins and Quillian (1969), allow for the use of predicate exceptions, preserving
hierarchies while allowing for deviations in their structures. For example, in the
chair category, a bean bag may be considered a chair without legs. In this case
the chair category would still contain the property with legs but the bean bag
category would contain a property that overrides this aspect of the parent chair
category.17
Symbolic representations all share common features in that all are
composed of ‘symbol manipulation according to explicit rules’ (Gardenfors
2000: 35) and are limited in a number of ways.

49
The architectonic system

1 The Frame Problem, which refers to the combinatorial explosion in terms

of the number of logical inferences that need to be programmed in order
to build even a modest knowledge system and the inflexibility of a seman-
tic network to evolve to suit the limited information needed in particular
domains (Gardenfors 2000: 37–8).
2 While symbolic representations can account for the evolution of cat-
egories, they do not account for the evolution of the predicates that
bind categories together, posing the question ‘where do predicates
come from?’ (Gardenfors 2000: 38). In other words, who interprets the
labelled links that constitute connections between concepts (Goldblum
2001: 38)?
3 Semantic networks lack flexibility. Once a concept has been learned by
these systems, it is very difficult to change their structure and this does
not account for the known human ability to adapt to new concepts as they
are learned and evolve over time.

The essential argument against symbolic representation is that, while its meth-
ods explain possible patterns of connection based on predicate relationships,
they don’t explain where meaning comes from.

Sub-conceptual representations
Symbolic representations are considered to occur on the conscious level of
thought. We are consciously aware of the hierarchies present in the way we
categorize the world. However, a greater understanding of the structure of
the brain and its neural networks has lead to the emergence of sub-conceptual
representations through the notion of connectionism. Connectionism is defined
by Dellarosa:

Events that co-occur in space or time become connected in the mind.

Events that share meaning or physical similarity become associated in
the mind. Activation of one unit activates others to which it is linked
and the degree of activation depending on the strength of associa-
tion. This approach held great intuitive appeal for investigators of the
mind because it seemed to capture the flavour of cognitive behav-
iours: When thinking, reasoning, or musing, one thought reminds
us of others.
(Dellarosa 1988: 29)

While a full account of connectionism is well beyond the scope of this book,18
there are some characteristics of connectionist networks that are worth noting.
Concepts are not stored in individual nodes (e.g. brain cells) in a system, but
rather in patterns of activity. This means that there are no discreet mappings
of single concepts onto defined neurons. Instead, a concept is composed of
a pattern of active neurons working together. For example, the category birds
may light up a cluster of neurons and similarly, but independently, the category
of feathers will also cause a cluster of neurons to fire. It is likely that both con-
cepts make use of and share a similar region of neurons. The two categories are,

50
 The architectonic system

therefore, related. By thinking of a feather, the concept of bird is also brought

into our conceptual view (Goldblum 2001: 44) (Figure 1.5).
Connectionism is not without its problems. In implementations of
connectionist systems, often through computer programmes called neural net-
works, it takes a massive number of training examples before a network picks up
useful knowledge of even a limited domain and there is an unavoidable challenge
in the complexity of turning on and off huge numbers of neurons in a complex
multidimensional system (Gardenfors 2000: 42–3).
Connectionist and symbolic representations are not mutually exclu-
sive. Connectionist models are often used to construct symbolic representations
(Goldblum 2001: 35). The two differ, however, in that connectionist repre-
sentations are sub-conceptual; i.e. we don’t comprehend the structure of a
connectionist system directly, in contrast to symbolic representations that we can
describe easily through external representations (notably language and diagrams).
Gardenfors (2000) describes these two systems in the context of
three levels of representation in which symbolic representations are at the top
and connectionist models are at the bottom. While the connectionist model
relates to the basic biological processes that constitute the building blocks of
thought, symbolic models exist as a product of a conscious interpretation of the
affect of the underlying structure. However, between the two, there needs to
be a bridge.

Conceptual representation
Between sub-conceptual and symbolic knowledge representations, suggests
Gardenfors, is an additional layer of representation, which he calls the ‘concep-
tual layer’ and which he describes by the use of an analogy, imagining that the
brain and its neural paths of reasoning are like a society of people living in a
jungle and following paths of inference to navigate through the jungle’s complex
undergrowth (Gardenfors 2000: 33–4). The question of understanding the differ-
ent levels of representation for those journeys, using this analogy, becomes a
question of how we represent the jungle dweller’s journeys.
A ‘myopic’ view would look at individuals and their interactions with
the jungle in terms of their immediate and local activity. Such descriptions would
include an understanding of the limitations of the environment and the person’s
ability to move through it; i.e. the individual cannot climb up steep cliffs, pass
through solid objects, etc. These material constraints define the class of pos-
sibilities for the individual. However, such localized descriptions of the jungle
are very limited in what they tell us about the environment or the individual’s
journey through it. This is the sub-conceptual level, analogous to the descriptions
of individual or small groups of neurons and their relationships as defined by the
physical reality of their connection to one another.
Over time, people’s movements will leave certain patterns in their
environment. Paths that are easily traversed become established and it becomes
possible to find directions that allow for a journey to be planned and for others
to follow the same paths. This tendency to move in certain patterns, to follow
the routes that others have followed, mean that it is possible to describe the
jungle, not holistically with relation to every topographic feature, but with relation

51
The architectonic system

1.5
Diagram to show the firing
of similar groups of neurons
when a human being thinks
about the categories of bird
and feather

to paths and the structures they create. Established paths make it possible for
individuals to make maps of their environment since, in the established jungle,
there are enough defined paths to make it possible to create descriptions that
are schematic; e.g. turn left at the tree stump, carry on north until the bush with
the blue flowers, etc. These sorts of descriptions rely on the existence of known
spatial dimensions that orientate the description in relation to polar coordinates
and distances, etc. This Gardenfors (2000) calls the conceptual level.
Finally, as different paths cross and become discreet locations in the
system, naming conventions can be implemented to efficiently describe the

52
 The architectonic system

most commonly used paths with instructions such as ‘start at A’, ‘go to B’, ‘turn
left at C’. This Gardenfors describes as analogous to symbolic representation,
where the most common and obvious pathways are revealed through efficient
representations.19
Conceptual representations bridge the gap between symbolic and
sub-conceptual representations by defining the underlying governing structures
of representation as a product of the structure of sense experience. Each domain
of classification, Gardenfors suggests, has a particular conceptual space attached
to its representation. So, for example, colour categories exist on the symbolic
level as named entities red, blue, green, etc., and, on the sub-conceptual level, as
the result of direct sensory input of different wave lengths of light. However, on
the conceptual level, colour categories are represented by a continuous spectrum
of colour variations in a geometrically defined mental space. This means that to
describe, for example, red as being close to pink we are describing its similarity
as a function of the proximal relationship of the two colours in a conceptual space
consisting of a colour wheel (Figure 1.6).20
The conceptual level of knowledge representation is comparatively
less understood than either the symbolic or sub-conceptual levels and Gardenfors,
while providing a spectrum of examples of different conceptual spaces, acknowl-
edges that he falls short of providing a definitive theory (Gardenfors 2000: 259).
However, the theory of conceptual space and the recognition of an intermediary
level between symbolic and sub-conceptual representations is an important one
for this study.
The theory of conceptual space is a useful introduction to embodied
theories of cognition in that it proposes a model for knowledge representation
that is not structured by a computationally driven and disembodied mind, but
is governed by our embodied relationship to the world and the structuring of
information as it is received through the senses. The discussion of ‘conceptual
spaces’ also hints at an underlying spatial cognitive function for categorisation;
i.e. the possibility that the parts of the mind that govern spatial cognition may
also act as a frame for organizing other concepts.

Evidence for conceptual spaces and schematic representations

in language
Returning to our ‘jungle’ of knowledge representation, if we want to understand
more about conceptual spaces then we might find clues to their structure in the
symbolic level of representation i.e. through the language we use to describe
abstract concepts. Cognitive approaches to knowledge representation frame
this discussion of architectonic systems and give access to an interesting vein
of literature that looks at how conceptual representations are externalized or
made visible through symbolic representations. In particular, a focus on lan-
guage has led to the development of theories of cognition based on the study of
metaphors.
While I have spent much of this chapter steering the argument away
from the linguistic system and toward an alternative ‘architectonic’ system,
language does, perhaps, offer the best insight available into the structure of our
conceptual world. In parallel to Gardenfors’s discussion of conceptual spaces, an

53
The architectonic system

analysis of category systems through language has revealed a conceptual basis,

or more accurately a number of conceptual bases, for knowledge representation
at the conceptual level. Notably, George Lakoff and Mark Johnson have detailed,
in a number of influential publications,21 the relationship between categories in
thought and cognitive structures as revealed through the use of fundamental
metaphors in language.
Metaphors, Johnson suggests, make use of patterns from our ‘physi-
cal experience to organize our more abstract understanding’ (Johnson 1990: xv).
In particular, Johnson isolates our physical interactions as a frame for our ability
to understand and communicate abstract concepts in the following way:

1.6
Diagram to show an example
of a sub-conceptual,
conceptual and symbolic
representations for colour
categories red and yellow

54
 The architectonic system

Understanding via metaphorical projection from abstract to concrete

makes use of physical experience in two ways. First our bodily move-
ments and interactions in various physical domains of experience are
structured … and that structure can be projected by metaphor onto
abstract domains. Second, metaphorical understanding is not merely
a matter of arbitrary fanciful projection from anything to anything
with no constraints. Concrete bodily experience not only constrains
the input to the metaphorical projection but also the nature of the
projections themselves, that is, the kinds of mappings that can occur
across domains.
(ibid.)

From the beginning, we experience constant physical containment in

our surroundings (those things that envelope us). We move in and out
of rooms, clothes, vehicles and numerous kinds of bounded spaces.
We also manipulate objects, placing them in containers (cups, boxes,
cans, bags, etc.) In each of these cases there are repeatable spatial
and temporal organizations. In other words, there are typical sche-
mata for physical containment.
(p. 21)

The physical relationships identified in metaphor reveal what Johnson terms

‘image schemata’, which are conceptual structures that organize patterns of
mental representation. Mental schemata allow us to operate in the world by
applying the conceptual structure of one domain to another. The earliest image
schemata are likely to have been those that govern basic motor functions. For
example, once an object (e.g. a stone) has been grasped by the individual’s hand,
the physical process of grasping does not need to be relearned for each new
object encountered. A conceptual model will exist for the properties of stone
and the actions required to pick up a different stone effectively. Image schemata
allow us to predict what is likely to happen given a new situation based on past
experiences.
Of particular interest here are a group of schemata isolated by Lakoff
and Johnson under the general heading of ‘Spatial-Relation Concepts’ (Lakoff
and Johnson 1999: 30). These emerge from our evolved ability to make sense
of space. Spatial-relation concepts are built around our ability to perceive spatial
relations in terms of objects which are in, on or across from other objects (1999:
631) and a common example occurs in the following statements:

Tell me the story again but leave out the minor details (the story event
becomes a container).
I give up, I’m getting out of the race (race event as container).
Whenever I’m in trouble, she always bails me out (state as
container).
(Johnson 1990: 34)

55
The architectonic system

The container metaphor maps to a single schema. In each statement the person
considers an event or state to be a container.
Johnson’s work is, along with Gardenfors’s, another articulation
of the conceptual level of knowledge representation but externalized through
(symbolic) linguistic representations. Like conceptual spaces, image schemas
are based on an embodied perception of the world, not only in terms of the
way in which the senses receive information about the environment, but also in
relation to how physical interactions with our environment are used as a frame
for abstract reasoning so that ideas are ‘objectified’; i.e. given physical qualities
and constraints and then manipulated. As is the case of intrinsic representations,
we would expect image schemas to place constraints on the domain being
described.

Toward a cognitive theory of architectonic systems

By starting to define the concept of the architectonic system from the per-
spective of the mind rather than from that of external representations, I have
attempted to articulate the other side of a relationship between structures of the
environment and structures of mental categories. While Part 2 of this chapter
seeded the idea that the organization of our physical world may be the basis
for the organization of categories in the mind, Part 3 has proposed a model of
categorization in the mind where categories belong, at the conceptual level, to
defined conceptual spaces. The conceptual level of mind is neither a product
of the structure of the brain (which is better explained through connectionist
theories at the sub-conceptual level) or of external representations alone (i.e.
semantic networks or logical inferences through language). The conceptual
level of knowledge representations is shaped and constrained by limited dimen-
sions; it is structurally restrained but is also fluid in terms of the way in which
content can be mapped on to different conceptual spaces through the use of
image schemas. I have also introduced the notion that the conceptual level of
representation emerges through structures of symbolic representations such
as language. Along with the idea that there are intrinsic types of representa-
tions which are confined by their source domain, I have used theories based
on the study of metaphor to contrast with mental schemas which are a way
in which representations can constrain the nature of the object we are trying
to represent.
These sources form part of a cognitive theory of architectonic sys-
tems but this work must be mapped on to the observations made by the research
discussed in Part 2. We cannot claim to have a reliable model for how the mind
works. However, taken together, these approaches suggest a picture of a mind
that is shaped by its environment and constrained by deep structural features
and as the result of which, is effective at making sense of space and structuring
abstract thinking.

Part 4: Topic and topos

An understanding of the cognition of classification, coupled with the appearance
of image schemas through metaphor, implies a dialectic relationship between the
mind and its tendency to spatialize concepts (through conceptual spaces) and the

56
 The architectonic system

organization of the physical world as a means of communicating. The dialectical

relationship between the embodied mind and our environment as a representa-
tion system reveals itself not only in patterns we impose on our environment, but
in the way we talk and the diagrams we create. However in this part I will return
to architecture as the focus of my attention revealing a relationship between a
particular architectural tradition and a philosophical tradition, which binds mental
space to physical space in an explicit manner (see Figure 1.7). I will develop a
narrative around a particular fragment of architectural history as it relates to the
manifestation of a particular, and mostly forgotten, architectural form: mnemonic
architecture; i.e. buildings designed to support the recollection of knowledge,
where a building’s typology emerged as a function of the cognitive process of
remembering. This formalization of architectonic systems begins with ‘the art of
memory’. Through this art, a way of considering the relationship between build-
ings and the mind developed which acts as a potent illustration of the cognitive
function of architectural space.
The history of architecture is invariably linked with epistemology, with
each historical epoch generating new architectural forms and the development of
specific building typologies. These include the museum, archive, library and labo-
ratory, each of which supports research processes and the storage of knowledge
artefacts. Understanding the built form as a way of organizing spatial experience
and articulating meaning has not only been the subject of the study of primitive
classification systems, but also modern building typologies and their evolution,
including the study of Renaissance memory palaces (Yates 2001) and the notion
of narrative architecture which is present in modern museums (Psarra 2009).
In addition to the study of physical architectures, the study of archi-
tectonic systems has lead to a corollary interest in the structure of philosophical
thought in relation to formal structural systems, which often make use of meta-
phors of the built environment.22
Part 4 will examine the legacy of mnemonic architectures in the con-
text of a world shaped by the cognitive necessity of ordered categorization. In
so doing the term architectonics will be extended to include not only its use in
discussions of the built environment but also its more common use to describe
the structures inherent in many philosophical systems.

The Art of Memory

In her seminal book The Art of Memory, Frances Yates (2001) describes how the
necessity for the efficient recall of facts and ideas that dominated pre-printing-
press cultures, led to the development of techniques to support memory. The ‘art
of memory’ thus emerges as a collection of material artefacts that act as memory
aids and that are based on mnemonic systems, which have been developed to
aid the structuring of knowledge and its systematic recall. All these systems
involve the structuring of memory into connected places, and can be roughly
divided into two sets, those which use language through rhymes or vivid episodic
stories23 and those involving the method of loci, which uses images placed within
built environments. The legend behind the ‘method of loci’ is introduced by Yates
in The Art of Memory with reference to the Roman philosopher Cicero’s account
of the poet Simonedes:

57
The architectonic system

At a banquet given by a nobleman of Thessaly named Scopas, the

poet Simonides of Ceos chanted a lyric poem in honour of his host
but including a passage of praise of Castor and Pollux. Scopas meanly
told the poet that he would only pay him half the sum agreed upon for
the panegyric and that he must obtain the balance from the twin gods
to whom he had devoted half the poem. A little later, a message was
brought in to Simonides that two young men were waiting outside
who wished to see him. He rose from the banquet and went out but
could find no one. During his absence, the roof of the banqueting hall
fell in, crushing Scopas and all the guests to death beneath the ruins;
the corpses were so mangled that the relatives who came to take

1.7
Architecture is a primary
means of communicating
abstract categories through
spatial organization

58
 The architectonic system

them away for burial were unable to identify them. But Simonides
remembering the places at which they had been sitting at the table
was therefore able to indicate to the relatives which were their dead.
The invisible callers, Castor and Pollux, had handsomely paid for their
share in the panegyric by drawing Simonides away from the banquet
just before the crash. And this experience suggested to the poet the
principles of the art of memory of which he is said to have been inven-
tor. Noting that it was through his memory of the places at which
the guests had been sitting that he had been able to identify the
bodies, he realized that an orderly arrangement is essential for good
memory.
(Yates 2001: 17)

While the development of the method of loci in ancient Greece tended to empha-
size the use of imagined fictional spaces, its evolution in ancient Rome was
heavily ‘grounded’ in physical spaces and places, through a tradition summed
up by Cicero who stated that ‘an object cannot be understood without a place’
(Cicero quoted in Small 1997: 97). Small argues that the difference between the
recorded Greek versions of the Method of Loci and the known Roman writings
on the subject is that the Romans offer a corporeal vision where real physical
spaces are necessary for remembering, in contrast with the Greeks’ emphasis
on a natural structuring of knowledge, which was not necessarily supported by
physical places. Under the Roman mnemotechnic system, the physical environ-
ment becomes a method of structuring information, and even whole cities can
become the basis for memory:

The word Mnemotechnics hardly conveys what the artificial memory

of Cicero may have been like, as he moved among buildings of
ancient Rome, seeing places, seeing the images stored on these
places, with a piercing inner vision which immediately brought to his
lips the thoughts and words of his speech.
(Yates 2001: 20)

The tradition of finding settings for the structuring of memory reaches its logical
conclusion during the Renaissance when the mnemotechnic tradition becomes
the genesis for a new art concerned with creating specific memory devices, most
notable of which was the audacious proposal of Guillio Camillo to build the vast
Theatre of Memory (Figure 1.8).
At the time the Memory Theatre was conceived, the intentions of
these new building types was no longer simply to create artefacts to support
memory. Instead, the relationship between the architectonics of philosophy and
what appear to be ‘natural’ rules for the organization of memory come together
in a new form of representation system, an ars combinatoria.
In its purest intention, the ars combinatoria is an attempt to find an
architectonic language distinct from text which, through the spatial configura-
tion of objects, would find the perfect correspondence ‘between words and
things, between logic and ontology’ (Rossi 2000: 61). Reading through the many

59
The architectonic system

post-Yates discourses on the art of memory as it developed into the Medieval

and Renaissance periods, it is striking that, in an attempt to discover a universal
art, notions of architecture and literacy were mixed together so that it became
possible to think of a building as written and read and at the same time consider
a sentence to be built and navigated.24

Constraint in logic, space and ontology

The method of loci works because of the human capacity to remember space to
an intricate level of detail and the capacity to use one mental structure (schema)
to support another. Yates describes this mental structure, with reference to
an early text on rhetoric, as an ‘inner language’ where the places themselves
become a medium:

For the places are very much like wax tablets or papyrus, the images
like the letters, the arrangement and disposition of the images like
the script, and the delivery is like the reading.
(Yates 2001 quoting from A Herennium: 22)

However, to describe the spaces used as the equivalent of papyrus scrolls or

wax tablets is not to recognize the impact that space itself has as a structuring
phenomena, creating unique constraints as a schematic representation.
The idea of the method of loci constraining representation appears
in a particular tradition of mnemotechnics described by Aristotle. In his text on
memory, De Memoria, Aristotle discusses the use of images as a way of imprint-
ing memories but does not describe the method of loci. Rather, he stresses the
importance of a natural structural organization of images into logical sequences
which can be traversed through a mental journey (Sorabji 2004: 31). In other
words, the objects of memory or fragments of information are composed in
a purely mental space, no setting is required other than the natural ordering
of ideas into coherent and interrelated patterns of meaning, which can be re-
collected. This sort of mental architecture, while having no apparent equivalent
in physical space, is structured by limited dimensions, where topos (Greek for
place) and topic (subject) become linked:

Aristotle’s application of the word ‘topos’ to general patterns of argu-

ment is the source of the name of his treatise, ‘The Topics’. And this
use of the word, along with the related use in rhetoric, is the source
of the English expression ‘topic’ and ‘commonplace’. If the above
suggestions are correct, these words will have come via Aristotle
ultimately from the system of place memory.
(Sorabji 2004: 32)

The term architectonic is much more likely to be found in relation to philosophical

concepts than to architectural space. Architectonic denotes the idea that, while
the subjects of philosophical discourse are important, it is also vital to understand
how they are constructed and to understand the underlying patterns of related
ideas which form a philosophical system.

60
 The architectonic system

1.8
Imaginative reconstruction
of the Memory Theatre by
Ruth Dickie

Architectonics and the patterns of ideas

The concept of architectonics in philosophy is profoundly complex. Philosophers

have for some time tried, both implicitly and explicitly, to break away from the
confines of philosophy’s architectonic systems, and discourses in post-structural
and postmodern philosophy in particular have sought relief from imposed
systems of an architectonic order. For example, in Lines of Thought (Lacour
1996) and Architecture as Metaphor (Karatani 1995), both authors attempt to
highlight the folly of the architectural metaphor as a way of ‘grounding and
stabilizing … otherwise unstable philosophical systems’ referencing the work
of philosophers as diverse as Plato, Descartes, Hagel and Kierkergaard (Karatani
1995: 5). However, ‘The Will to Architecture’ (Karatani 1995: 5), through the drive
to structure knowledge in relation to an allusion to material constructions and to
associate topos with topic remains pervasive and powerful. In simple terms, the
creation of architectonic structures is derived from a desire to create discreet
containers for specific concepts so that their nature can be inferred from their
position within an overall system.
The relationship between topos and topic creates representational
constraints. These constraints are, furthermore, expressions of the schema that
frames them, in that, in each case, concepts are treated as objects with defined

61
The architectonic system

spatial relationships to one another. In mnemonic architectures these relation-

ships are explicit because ideas are literally mapped on to images that reside
within physical spaces. However, in the philosophical system of Aristotle, they
are implicit through the semantic networks created by hierarchical categories.

Constraint in architectonic systems

The spatial relationship schema supposes that the represented domain consists
of metaphorical objects and containers with relationships to one another. These
spatial relationships are topological in that they refer to the relative positions of
the objects in space; i.e. of one object being next to or contained by another.
Semantic associations are then mapped on to these spatial relationships. Thus,
closeness represents similarity, sharing a container denotes a sharing of proper-
ties to create a category and an object seen as being above another may denote
a difference in value or status. These metaphorical projections are constrained by
the realities of 3D space; i.e. a single object can only exist in one location at a time
and two objects cannot reside in the same place simultaneously. Some spatial
relationships are also constrained by 2D space, for example when we consider
the structures created by the method of loci where images are distributed in an
environment and recalled by a process of navigation.
Navigation tends to occur on 2D planes and most buildings have
plans consistent with the rules of 2D topology, where only certain combinatorial
configurations are possible. This restriction on 2D space is often illustrated with
relation to puzzles such as the three utilities puzzle. The puzzle involves three
houses and three services (gas, water and electricity), each represented as a
node. The challenge is to connect all the houses to all the services without cross-
ing any of the pipes. In other words, the ground is considered to be a 2D plane
so that service pipes cannot be placed over or under one another. We could
equally modify the puzzle to describe the movement of three people. Imagine
three poles placed in a field and three people, who start from designated points,
create paths by walking to each of the poles. It is inevitable that one of the
walkers will be forced to cross the path of another. These puzzles both describe
non-planarity and can be represented in graph form by what is referred to as a
K3,3 graph, where a group of three nodes are perfectly connected to three other
nodes. The other non-planar graph involves five nodes that are perfectly linked
together, known as a K5 graph (Figure 1.9).
Planarity has significant implications for architectural space because
of our tendency to plan on 2D planes. If we consider each node to be a room and
links between nodes to be thresholds between rooms (i.e. doorways), we create
a ‘dual graph’ which could represent a floor plan (Figure 1.10). It would then be
impossible to have a configuration of rooms in which a group of three rooms was
adjacent to a group of three other rooms. Similarly, five rooms cannot be adjacent
to one another. This means that certain functional relationships between spaces
are impossible and the study of planarity in graphs is highlighted by Steadman
as critical to understanding the disposition of functions within a building and the
constraints placed on what can be next to what (Steadman 1976: 101).
If, as envisaged in mnemonic architectures, ideas have places and
their associations are articulated as spatial relationships, which must be navigated

62
 The architectonic system

to be recalled, then topological restrictions which apply to 2D topology might also

apply to the construction of mental categories.

Part 5: Conclusion
Having rapidly traversed human history and the domains of architecture, linguis-
tics, anthropology and cognitive psychology, it is necessary to conclude with a
brief summary of the argument so far.
In Part 2 I focused on two texts. First, following Preziosi, I distin-
guished between two roots of early communication, the linguistic system and
the architectonic system. The linguistic system originated from the early use
of tools where inanimate objects acquired meaning through the association
between their form and their function. These tools thus became the first lin-
guistic gestures.
Early buildings and settlements can also be interpreted as functional
tools, however, their functions were not directly related to specific actions but
rather to the designation of territory and social space. The primary function of
early building types was, therefore, to subdivide space and thus categorize
activities and people. There is a clear distinction between the role of space as a
functional tool and the symbolic relationship of material form to function, and the
architectonic system is therefore separate from the linguistic system.
In parallel, structuralist approaches to the analysis of ‘primitive clas-
sification’ attempt to map patterns of settlements against known classification
systems and Durkheim and Mauss’s analysis of primitive classification reveals
a narrative which suggests that the conceptual human propensity to classify
emerges from 1) the organization of social groups; 2) their distribution in the
physical landscape; and 3) the association of other objects (animate and inani-
mate) with those social groups. While the actual classifications vary significantly
across different cultures, the method of their formation does not, with com-
mon practices being identifiable for organizing objects and social groups into
orientations of space (up, down, north, south, east, etc.). Long after the actual
distributions have ceased to become illustrative, the systems remain, capturing
a conceptual space rather than a physical one.
In Part 3, I approached the concept of classification from another
direction, as an emergent mental phenomena, through a discussion of knowl-
edge representation. I highlighted three levels of knowledge representation:
symbolic, conceptual and sub-conceptual representation. It was noted that
symbolic representations tended to result in externalizations of knowledge
through, for example, language and diagrams, that sub-conceptual representa-
tions describe mental processes at the level of neurons in the mind and that
conceptual representations shape our knowledge of the world through what
Gardenfors described as conceptual spaces. I suggested that conceptual spaces,
as a structuring phenomena, are revealed in language through the use of certain
metaphors and, following Lakoff and Johnson, isolated spatial relation metaphors
as pervasive in conceptual thought. I proposed that the architectonic system and
the notion of categorization may be constrained by schematic properties. That is
to say that limitations in space are imposed on representations which use space
as a structuring phenomena.

63
The architectonic system

1.9
Diagrams to show (from
left to right) a K3,3 and a
K5 graph

1.10
Diagram to show a graph of
six nodes, its dual graph and
a plan with interconnected
rooms generated from the
dual graph

It has been noted that the relationship between the physical world
and mental classifications can be seen in the origins of philosophy. The term
architectonic turns up in relation to the work of Aristotle and discussions of the
relationship between philosophical discourse and techniques of remembering,
in particular the method of loci, in which facts and ideas are remembered with
relation to a logical pattern of spatial relationships. We have seen, in Part 4,
how the relationship between topic and topos was developed by the Romans
and was revealed through the work of Cicero, who proposed a method of loci
based upon physical locations and paved the way for ‘the art of memory’, in
which physical architectures were designed as specific ‘memory palaces’.
Much more than mnemotechnic devices, however, we have seen how these
new constructions were descriptions of the world through the arrangement of
physical objects and their mapping of subject domains through a relationship
between topos and topic.
The influence of the architectonic system has continued both in
philosophy and in architecture. In philosophy, the ‘will to architecture’ has been
characterized by the recurrence of architectural metaphors in many western
philosophical texts from Aristotle to Descartes and, in architecture, it is revealed
through the emergence of building typologies such as libraries and museums,

64
 The architectonic system

which mediate the relationship between people and objects through the articula-
tion of space.
This selective historical narrative is useful but far from complete.
From the seed of primitive buildings to the tree of modern architecture I have
chosen only a few branches through which to analyze the relationship between
the built environment and meaning in an attempt to refine the idea of the archi-
tectonic system as distinct from the linguistic system. While I want to maintain
the distinction between architectonic systems and linguistic systems, it would
be naive to see them as completely separate. As has been shown, language can
give us an insight into the nature of architectonic systems because, in describing
space, language reveals and informs our perception of our physical world and,
through metaphor, shapes our spatial and cognitive interactions. This is particu-
larly true where the architectonic systems are not explicit.
The architectonic system is based on the idea of articulating space
and thus the objects and people within space. It therefore frames the concept of
classification, which has developed through the emergence of patterns of social
space and a human will to impose order on the world. Furthermore, while some
instances of the architectonic system remain in the mind, others are made visible
and real in the form of buildings and assemblages of objects. I have proposed
here that the seed of the architectonic system was planted with the first build-
ings, in other words the first physical interventions in the landscape, and has
been continued through a plethora of building typologies.
In the following chapter I will consider what happens when the rela-
tionship between information and its material manifestation is broken by digital
technology. In other words, what happens to the architectonic system in the
post-digital world?

65
Chapter 2

Between city lights

receding and the
non-space of the mind
Think of a Digital Artefact, shaped by software operations, made up of
data assemblies. Although lacking in physical substance, it is a thing
with an appearance, spatiality, structure, workable properties and a
history.
(McCullough 1998: 155)

Part 1: Loosing the ballast of materiality

There are, suggests Michael Benedikt citing Karl Popper, three worlds. World
One is the physical world, characterized by ‘natural things and their physical
properties’. World Two consists of the thoughts and imagination of our mental
domains and World Three consists of

the world of objective, real and public structures which are the, not-
necessarily-intentional, products of the minds of living creatures,
interacting with each other and with the Natural World 1.
(Benedikt 1991: 3–4)

Until the latter half of the twentieth century we had lived with the reality that to
articulate World Two, we needed the materials provided by World One. This has
tended to mean that our representations of World Two have been constrained
by the physical properties that define our material existence. Digital technology
has, however, changed this status quo by allowing, for the first time, information
to be separated from its material means. As Benedikt put it, we are ‘loosing the
ballast of materiality’ (Benedikt 1991: 4).
In Chapter 1, I showed how, through representation, our material
articulation of the world also shapes our cognition of abstract concepts; how
conceptual objects are constrained by spatial metaphors; how categorization may
have emerged from the organization and distribution of social groups and how
knowledge representation is shaped by a conceptual space of representation.
In contrast to Benedikt’s interpretation of Popper’s theory, however, Chapter 1
did not articulate two or three worlds but systems of communication character-
ized as the linguistic and architectonic in which the mental and the physical are
Between city lights receding and the non-space of the mind

in constant dialogue. Digital technologies question this dialogue by proposing

a material-less world which is free from physical constraint. This new world,
described generally in the 1990s as ‘cyberspace’, is made visible through the
fibre optic cables of telecommunication, is made manifest on the screen of
personal computers and a wide range of mobile devices and is sensed through
the electronic climate of invisible but pervasive wireless networks. But, while
digital technologies rely no less on physical materials than any other technology
of communication in terms of their implementation, they propose a model of
non-material engagement. Cyberspace, suggests Margret Wertheim, is seen
by some as an attempt to manifest ‘an electronic space of mind’ (Wertheim
1999: 39):

[C]yberspace itself is not located within the physicalist world picture.

You cannot pinpoint it on any cosmological map. You cannot deter-
mine its coordinates in Euclidian or relativistic space. As complexity
theorists would say, it is an emergent phenomena whose properties
transcend the sum of its parts.
(Wertheim 1999: 38–9)

This transcendental ‘cyberspace’ offers only half the story. There is inevitably
a place in which cyberspace itself must become manifest and, even if it can’t
be mapped, cyberspace continues to be conceptualized as ‘something’ existing
‘somewhere’ even if it is, in the most fantastical science fiction imaginings, ‘the
nonspace of the mind’ (Gibson 1995: 67). Have we really been able to break free
from the ballast of materiality?
This chapter will describe some key paradigms in the pre-history and
early history of computational technology through the lens of the mind/mate-
rial dichotomy, which is inherent if we adopt a cyber-spatial view of the world
when considering human interaction with information. To this end I will develop
a narrative that shows how, despite the technical possibilities of dematerialized
information, the realization of computation through various interfaces has sought
to use metaphors of space, or even to simulate aspects of the physical organi-
zation of material objects through virtual spaces and visualizations, which often
make use of architectural metaphors.
Part 2 will trace a descendent of the ‘art of memory’ (which was
introduced in Chapter 1) through the development of logic and the search for a
universal language. While in architecture, the building typologies of the museum
and library continue a tradition of mapping classification systems on to physical
spaces, a branch of ars cobinatoria (the art of combination) turned its attention,
through the Renaissance period, to overcoming the challenges posed by topo-
logical restriction. In this context, Part 2 will focus on Ramon Lull’s attempts
to overcome these restrictions with the creation of mechanisms for generat-
ing combinations and permutations. These machines were early forms of the
computer and show that, even in a pre-digital era, the limitations of physically
bound classification systems were understood and that the evolution of logic
was an attempt to capture information in dynamic mechanical processes rather
than in static spatial structures. This part traces the conflict between notions of

68
 Between city lights receding and the non-space of the mind

structured and unstructured information and their relationship to architecture,

through the work of Paul Otlet and his designs for the ‘Mundeneum’, a universal
archive that contains, simultaneously, an echo of the Theatre of Memory and a
progressive understanding of information as separate from its material manifes-
tation. This part concludes with a discussion of the last of the unrealized grand
plans for the organization of knowledge, the Memex, a machine proposed by
Vannavar Bush. While the Memex was never realized, the philosophy behind it
would have a substantial influence on the early pioneers of digital information
systems.
Underlying this evolution of the human interaction with information
is a change in the way that the human mind is understood, from the topic/topos
structures of memory palaces, which proposed a mind where memory was
given fixed representations in a cognitive map, to the notion of the brain as an
association machine where knowledge is enacted through mechanical processes
(see Figure 2.1).
This shift in understanding of information is partially responsible for
modern computing. However, the implementation of computer systems in two

2.1
A revision in our
understanding of the human
brain and the enacting
of information through
mechanical processes gave
rise to computing as an
alternative way of organizing
and processing information

69
Between city lights receding and the non-space of the mind

of the key paradigms for human computer interaction, the GUI and hypertext,
appear to contradict the apparently limitless technological possibilities provided
by dematerialized information.
Part 3 will trace the emergence of GUIs as a way of interacting
with computational information. GUIs are widely held to be a vital computing
innovation and they made computing accessible for non-specialized users and
ultimately led to the revolution of personal computing. The visual metaphors of
the desktop of the PC GUI, with the ‘direct manipulation’ of physical objects in
constrained spaces however, not only made computers easier to use (compared
to their command line based ancestors) but have also introduced constraints,
which have nothing to do with the underlying digital system itself. The GUI
has sought, not necessarily to augment man’s intellect through computation,
but rather to simulate aspects of the real world in order to make computers
accessible.
Whereas the GUI’s use of visual metaphors overtly constrained the
presentational space of digital information, the paradigm of hypertext (extend-
ing Bush’s notion of the Memex) offered an apparently different vision of a
networked topology of information through the WWW. Part 4, however, will
examine the emergence of the first hypertext systems through the lens of a
material/virtual dichotomy that has, I will suggest, restricted hypertext topolo-
gies and the way in which information on the web is authored, by introducing a
conceptual model for hypertext exploration based on navigation.
Part 5 will conclude with a discussion of the underpinning theoretical
notion of ‘remediation’, examining the dominant role of a spatial understanding
of digital media in discussions of the remediation of text. This discussion seems
to close a circle, rebinding the relationship between topos and topic, apparently
reinstating the conceptual basis of place based representations of information
in an age of synthetic memory.
Through these instances of the GUI and hypertext, this chapter aims
to draw out a fundamental and repeated problem relating to the way that digital
information has been conceived and the way it is presented. Although taking a
broad brush approach to a number of different concepts and technical paradigms,
I want to set the stage for a revised understanding of ‘information space’, which
will be developed in Chapter 3.
This chapter offers a uniquely architectural view of the discourse
and I run the compounded dangers of oversimplifying the philosophical basis of
logic and combinatorial mathematics and of choosing, seemingly idiosyncratic
examples from the broad history of computing. I will plead guilty to both these
things but suggest that this chapter should not be read as a comprehensive
history of computing or an in-depth enquiry into logic in the context of know-
ledge representation, but rather it should be seen as the beginnings of a search
for how an architectural understanding of space has been and can be used to
understand the wider issues of digital information and its representation. At its
most fundamental level, this relationship is about an evolving information space
unable to escape from its embodiments.

70
 Between city lights receding and the non-space of the mind

Part 2: Spaces between logic and ontology

In Chapter 1, I introduced the ‘art of memory’ in the context of ars combinatoria, a
method which combines the art of placement with a notion of logical inference so
that conceptual objects can be constructed to form logical statements, thus bind-
ing physical placement with semantic relationships. The challenge, as defined in
Chapter 1, is that in physical space there are a limited number of such relation-
ships that can occur. This problem would challenge the philosopher Leibniz in
the seventeenth century. Employed as a librarian by John Fredrick of Hanover,
Leibniz sought not only to understand knowledge as communicated through
language but also to understand how the categorization of objects (specifically
books) articulated a philosophical discourse. The problem for Leibniz, who was
also a pioneer of logic, is that the fixed organization imposed on classified objects
was ‘contrary to logic’ (Markus 1993: 174) and that, in his own words, ‘It is usu-
ally found that one and the same truth may be put in different places according
to the terms it contains’ (Leibniz quoted in Markus 1993: 174).
This challenge has now been answered by digital technologies but
the push to extend our representations of information beyond the materially
bound world is not unique to the pioneers of digital technology and, in this
part, I give an overview of the development of computation as it emerges from
the place-based method of loci and is ultimately bound up (as will be shown in
Parts 3 and 4) in a complex relationship with material representations; a shift from
knowledge as static organizations of objects to dynamic systems of logic and a
changing notion of how humans interact with information and its representation.

Combinations and permutations

There are two ways of overcoming the problem of the geometric restriction
inherent in physical organizations of objects and we can illustrate these methods
in relation to books in a library.
The first method would be to realize different instantiations of the
same group of objects, each implementing a different category structure, e.g.
having multiple libraries. There are, indeed, multiple libraries carrying similar con-
tent throughout the world but no librarian would consider suggesting that each
one should be organized differently or that a single library was extended to carry
multiple copies of itself to express the many possible permutations of books.1
The second option would be to mechanically intervene by, for exam-
ple, providing mechanized shelves in which the books could be arranged and
rearranged at will. One can only imagine the machine that would be necessary
to perform this operation for an entire library, but mechanical intervention in
certain circumstances is not a ludicrous notion if the information objects can be
reduced to a manageable scale and the appropriate combinatorial patterns can
be mapped on to a mechanical device.
A property of combinatoriality is that even a modest set of objects can
be combined in a huge number of different permutations. In fact, as the number
of objects grows, the number of possible permutations grows exponentially. A
useful way to understand the scale of combinatorial explosion is introduced by
Umberto Eco in his text on universal languages. He compares the number of
permutations in a three-letter anagram. There are six possible permutations of

71
Between city lights receding and the non-space of the mind

the letters ABC, which consist of: A B C, A C B, B A C, B C A, C A B, C B A. Now

imagine an anagram consisting of all 26 letters of the alphabet. The result would
be reminiscent of the Library of Babel and the world would not be large enough
to contain all the possible permutations, which number:

371,993,326,789,901,217,467,999,448,150,835,200,000,000.
(Eco 1995: 54)2

In language, however, the number of meaningful combinations is constrained.

We can limit the number of permutations by devising rules of expression. For
example, Umberto Eco uses the case of four soldiers A, B, C and D and the
challenge of finding how many different patrol groups are possible where a
patrol group consists of two soldiers. In this case, many of the permutations are
identical (group AB is exactly the same as group BA) since the order in which
they are expressed contains no meaning in this example. Rather than a problem
of permutation, the example of the soldier patrol is a problem of combination
(Eco 1995: 55). Ars combinatoria is the art of uncovering combinatorial rules and
restrictions that govern the placement of information objects and realizing these
patterns of relationships through systems in which these combinations can be
expressed and in which all allowable permutations are meaningful. In fact, all
information systems can be reduced to combinatorial patterns in this way. It is
restrictions in the combinatorial possibilities of letters from which we can derive
meaningful words and it is the combinatorial rules inherent in the grammar and
syntax of language, which allow us to construct meaningful sentences. There are
vastly more possible permutations of letters and words than there are meaningful
combinations of them.
Pioneers of ars combinatoria, such as the eccentric thirteenth-century
Majorcan philosopher Ramon Lull, sought to understand the power of com-
binations and permutations to express knowledge and to develop universal
languages. For Lull, combinatorial systems offered a way of presenting the
fundamental moral code embedded in the Bible but, for this study, the content
of Lull’s system is less interesting than the mechanism of its representation. In
developing suitable representations to support the Method of Loci, the challenge
had been to find an ideal organization of information through a fixed spatial organi-
zation of objects (as epitomized by the Theatre of Memory). Lull also envisaged
a place-based system of information representation but he developed it beyond
a fixed organization of places toward ‘chambers’, which could be reconfigured.
This approach reached its zenith in the Lullian Circle, described by Eco as ‘a
mechanism formed by three concentric circles of decreasing size, inserted into
each other, and held together usually by a knotted cord’ (Eco 1995: 59). Each
circle contained nine letters relating to statements held in an associated index.
By turning the wheels independently, Lull could build statements consisting of
84 possible combinations.3
A simplified version of the Lullian Circle can be illustrated (Figure 2.2)
using the example of the three-letter anagram. If we are interested in the per-
mutations of letter combinations A, B and C, then we could create three wheels
each with A, B and C written along their circumference. We could then express

72
 Between city lights receding and the non-space of the mind

2.2
A simplified version of the
Lullian Circle containing
three concentric rings and
three letters which can be
recombined by turning the
wheels independently from
one another

any permutation we wished by turning the wheels on an axis at their centre and
lining up the different letters.
We can also create a similar mechanism to express the combinations
of the two groups ABC and DEF (Figure 2.3), where combinatorial constraint
means that each of the letters A, B or C must be paired with one of the group
D, E and F. This combinatorial constraint is represented in the machine by fixing
the ABC group to a single circle and DEF to another. The combination A B for
example, is impossible since the letter A can only become associated with a
letter from group D E or F.
For Lull, such combinations allowed him to create machines for the
production of logical statements capable of producing tens, or even hundreds,
of possible combinations based on the creation of sets or categories of state-
ment, each of which occupy different levels of the wheel. Lull had escaped from
the combinatorial limitations of a fixed 2D topological space. Insomuch as Lull’s
approach to ars combinatoria belongs to the tradition of artificial memory, Lull is
thought to have considered the units of combination as representing places, as

73
Between city lights receding and the non-space of the mind

evidenced in the use of the term ‘chambers’ to describe them (Eco 1995: 59) and, 2.3
A variation on the Lullian
to this end Lull’s machines are designed to mechanically manipulate adjacencies Circle in which two separate
between places.4 Lull therefore extends the notion of a language of fixed places groups of letters can be
and spatial relationships through a process of mechanical change and through recombined

the creation of complex systems, which can be interrogated to answer questions

that were not predetermined. In other words, Lull has created an early type of
computer, where meaning is encoded through the potential of the system rather
than persistently expressed as a static spatial representation.

The last universal architecture

Lullism represents an evolution of the representation of information yet is still
based in the tradition of the method of loci through an, apparently, place-based
understanding of the representation of structured information. The technological
possibility of ‘escaping place’ would not appear until the late twentieth century
yet, before then, our understanding of information and its relationship to physical
objects and place continued to evolve and architectural space was increasingly
contested as both a restriction on information and a metaphor for the way it
must be organized.
One particularly illustrative example of the contradiction raised in
the context of architecture as metaphor for systems that are fundamentally
non-architectural, comes through the work, at the turn of the century, of the
great Belgian intellectual and entrepreneur Paul Otlet. Otlet’s work was only

74
 Between city lights receding and the non-space of the mind

2.4
Imaginative reconstruction of discovered in the late 1960s by a graduate student who stumbled upon an
the Mundeneum based on Le archive of his work in Brussels (Wright 2007: 184). What he found has made
Corbusier’s original designs
Otlet a cult figure in the history of the information society. Otlet envisaged two
significant building projects. One was called the ‘Mundeneum’ (Figure 2.4) and
was a proposal for a building to provide access to all the world’s knowledge. The
other proposal was an even more utopian vision of a World City which would be a
‘colossal Book, whose buildings and their arrangements – not just their contents
– will be read, in the same way as the stones of cathedrals “were read” by the
people of the Middle Ages’ (Otlet 1934 quoted in Chabard 2008: 108).
Like so many ideas for ‘universal archives’, the Mundeneum was
never built in its full incarnation and the World City was not built at all, but some
of Otlet’s ideas have survived and are remarkable because of their anticipa-
tion of today’s technological reality. Otlet’s work on documentation and its
organization and retrieval can be seen from two, arguably contrasting, points
of view.
On the one hand, Otlet was an ‘information architect’ (perhaps one
of the first), and through his plans for the Mundaneum, he saw buildings as
representation systems that could break, what he believed to be, the scourge
of the book and the limitations of written language. Otlet saw, argues Chabard
(2008: 111), architectures as determining the structural relationships of their
contents through topoi (the structured collection of places). Architecture was
both a metaphor for Otlet and a physical reality, and his relationship with the

75
Between city lights receding and the non-space of the mind

great modernist architect Le Corbusier and the plans for the Mundeneum link
him with the modernist tradition in architecture which considered the world
through plan, analysis, classification, abstraction, standardization and synthesis
(Heuvel 2008: 129).
On the other hand, the modernist era of which Otlet (ostensibly
through his relationship with the architect Le Corbusier) was a part, also con-
vinced him to be more progressive in his visions for an information society and,
prefiguring the WWW, he describes knowledge as an assemblage in ‘open and
in continual mutation’ (Heuvel 2008: 128).
Otlet battled with these apparently contradictory positions. He sought
with Le Corbusier to find the ideal form for the expression of the Mundeneum
yet he would seek to fill it with devices, not unlike computers, where knowledge
was no longer organized but performed through mechanical devices connected
together in a ‘universal network’ (Heuvel 2008: 140) of telephone lines and
presented graphically on screens (Heuvel 2008: 145). Otlet also imagined and
implemented one of the first faceted classification schemes which allowed him
to separate the object of the book from the valuable information it contained
(Wright 2007: 187).
The Mundeneum is one of the last grand proposals for a universal
archive and exists at a critical transition point in the history of emerging informa-
tion architecture. It represents the last stand of the universal and the structured
architectonic system. It also represents an emergent networked sensibility, a
new understanding of the role of architecture from being bibliographic to sceno-
graphic, i.e. from a structuring medium to a stage set; and from the institutional
to the personal. This transition would be realized more fully as a concept by
Vannevar Bush and would be made reality by Tim Berners-Lee.

Extending memory
In 1945, in the wake of the Second World War, the scientist and engineer
Vannevar Bush, writing an article entitled ‘As We May Think’ in The Atlantic (later
published in 2003 in The New Media Reader among other places) pondered,
in a similar vein to Liebeniz, the explosion of knowledge and the challenge of
organizing it. He worried that great scientific insights may be prevented because
relevant research was buried underneath the mass of information that had grown
since the beginning of the Second World War.5 Bush realized that the old catego-
rization systems would no longer suffice, both in terms of the divisions between
disciplines and the way knowledge was stored in its physical form (Bush’s focus
was on scientific papers). Bush had emerged after the war having overseen mas-
sive engineering projects, notably the construction of the first atomic bomb, and
had witnessed firsthand the advances that could be made with concerted effort
and the collaboration of specialists across fields. He worried that, without the
impetus of war, the technological strides which had been achieved during the
Second World War would not continue to be made as scientists and engineers
returned to their respective disciplinary ghettos and to a world in which knowl-
edge between fields and individuals became incommensurable.
Bush’s unique position as the US Government’s Director of Scientific
Research and Development allowed him an overview of the technological state

76
 Between city lights receding and the non-space of the mind

of the art in terms of new computational machines, optical systems and displays,
and he envisaged the creation of a machine called the Memex, which could sup-
port the production of knowledge through a system of semantic relationships
which did not rely on spatially coherent places:

When data of any sort are placed in storage, they are filed alphabeti-
cally or numerically and information is found (when it is) by tracing it
down from subclass to subclass, it can only appear in one place ….
The human mind does not work in that way. It operates by associa-
tion. With one item in its grasp, it snaps instantly to the next that is
suggested by the association of thoughts, in accordance with some
intricate web of trails carried by the cells of the brain.
(Bush 2003: 44)

Like the pioneers of the method of loci, Bush was interested in the representa-
tion of knowledge in human memory, and the word Memex is a portmanteau of
‘memory extender’. Rather than creating an external aid to cognition, however,
Bush felt that information systems needed to reflect what he believed to be the
structures of thought, separating topic from topos and creating, in the Lullian
tradition, a synthetic memory. Like Lull, Bush proposed that the bridge between
the mind and the physical storage or material information objects could be
achieved through mechanical means. Bush, however, did not propose a univer-
sal machine for permutation, but imagined a flexible system for the creation of
much less structured semantic relationships. Bush’s proposal took the form of
a desk that stored microfilms, cameras, readers and electromagnetic controls
combined to become, using my own crude analogy, a cross between a juke box
and a microfilm reader:

Consider a future device for individual use, which is a sort of mecha-

nized private file and library … It consists of a desk, and while it can
presumably be operated from a distance, it is primarily the piece
of furniture at which he works. On the top are slanting translucent
screens, on which material can be projected for convenient reading.
There is a keyboard, and sets of buttons and levers. Otherwise it
looks like an ordinary desk.
(Bush 2003: 45)

The two screens would be able to present two documents simultaneously with
the user being able to assign a code to one document which provided a link to
another.

Thereafter, at any time, when one of these items is in view, the

other can be instantly recalled merely by tapping a button below the
corresponding code space … any item can be joined into numerous
trails.
(Bush 2003: 45)

77
Between city lights receding and the non-space of the mind

The physical arrangements of document objects stored within the Memex were
separate from their semantic association with one another and, critically, this
semantic arrangement, freed from physical constraint, could become multidi-
mensional, allowing for a single document object to occur in multiple contexts
or, in Bush’s terms, recombined into new ‘trails’.
The Memex machine was never realized but it offered a radical
understanding of the relationship between human internal and external informa-
tion representations. Through Memex, it would no longer be possible to think
of information as residing only in materially bound external representations
or only as mental constructions. The Memex introduced a third layer, realized
through a mechanical code, unbound by material constraints yet outside the
mind, visible only at the point of interaction with a mechanical system. It was,
importantly, proposed as a mechanical way of simulating an aspect of the
function of the mind and this mind was not fixed to places and topoi but was
associational.
Bush’s article, ‘As We May Think’, can be read online now. Googling
its title and reading it through the window of my computer’s web browser is a
spine-tingling experience. On the one hand, the proposal outlined in the article
is, with hindsight, remarkably modest in terms of its technical implementation.
On the other hand, it seems to exemplify a blueprint for a new understanding
of information. ‘As We May Think’ is an acknowledged forerunner of the digital
revolution and is made all the more remarkable by the fact that Bush did not have
access to the technologies that we now take for granted and that now shape our
understanding of information, as evidenced by the very technologies that can
now be used to read his article.6

Extending man’s intellect

The underlying thinking that was driving Bush would continue throughout the
latter half of the twentieth century, influencing a new generation of computer
scientists who were increasingly aware that the emerging digital technologies
were capable of radically realizing a world of information that was separated from
its material means. The history of the computer and of digital media has been
told countless times and many of the pioneers of computing, such as Alan Turing
(an early pioneer of computing), Doug Engelbart (the inventor of the mouse) and
Tim Berners Lee (the inventor of the WWW) are, if not household names, then
certainly part of a popular consciousness.7 These pioneers envisaged systems in
which computation could provide new types of information representation and
new ways of interacting with them.

Liberated information?
The transition of the representation of information from object to process, as
I have shown here, has been a gradual one that started as an offshoot of the
‘art of memory’; through ars combinatoria; through to the evolution of logical
systems of inference and eventually to digital systems where permutation and
combination are no longer restricted by spatial locations. This should be seen as
a departure from the architectonic system as it was proposed in Chapter 1. After
all, a disconnection between information and its material means necessitates a

78
 Between city lights receding and the non-space of the mind

disconnection with architecture as a form of material articulation for systems of

categorization. Furthermore, the realization of a mechanical mind negates the
mind material dialogue proposed by the evolution of the architectonic system.
However, there is a dichotomy at the heart of this digital revolution which occurs
at the point of human interaction with digital systems.
You may have accessed this text online through the browser on your
computer and a digital copy might have been uploaded and sits on the file sys-
tem of your hard drive and what you will become aware of is that, at the point
of interaction, the information object presented to you, whether it be online or
stored on their computer’s ‘desktop’, is not connected in a multidimensional
hyperspace but, rather, is restricted to either specific screen positions within
a hierarchy of web pages or files and folders. Digital computational informa-
tion, therefore, has not fully escaped from its material world but, as will be
shown, resides in a virtual world that appears to mimic many of the material
world’s properties. The propensity for mimesis occurs across a range of digital
artefacts and here we will focus on two, GUIs and hypertexts as realized on
the WWW.

Part 3: ‘Display becomes reality’

Twenty years after the birth of the WWW, we are barely grappling with the
profound implications that digital media is having on our lives. The internet is
the single most complex system that humans have created and its centrality to
the distribution of information means that its effects have been felt in almost
all aspects of society within the developed world. The era of the networked
society has begun. But what of the architectonic system, the history of which
begins with the first architectural gestures which bound classification to spatial
organization?
The opportunity to think about information as a ‘something’, inde-
pendent of us and of the material world, comes about largely because of
computers. Suggesting, as I have in the previous chapter, that the way we rep-
resent knowledge in the mind is structured in part by our environment and by our
comprehension of the spatial world, only becomes an insight once we consider
the alternative possibility that a ‘space’ of information exists independently from
its expression and material representation. In considering this alternative, how-
ever, we note that interfaces to computational information systems often appear
to adhere to spatial conventions. This adherence to the conventions of spatial
organization can be seen most clearly in the emergence of the paradigm of GUIs.

Bridging the gap between bodies and information

Doug Engelbart (1962) wrote a seminal report entitled ‘Augmenting Man’s
Intellect: A Conceptual Framework’, in which he proposed a new discipline
concerned with augmenting the human capability to handle information through
computer technology. In his report, Engelbart discusses the importance of pro-
cesses and objects8 that support human cognition and recognized the potential of
the computer as a radically new tool. Englebart realized, however, that computers
as they were then conceived used representation systems, which, at the point
of their human interfaces, were not easily compatible with human cognition and

79
Between city lights receding and the non-space of the mind

natural forms of interaction. To work more effectively, computers must support a

fluid human–machine dialogue and thus become more useful tools to ‘augment
man’s intellect’.
In order to overcome what they considered to be a physical and
temporal separation of human and computer, at a time when the dominant
interaction paradigm was through programming languages, Englebart and his
research team at Stanford University needed to radically rethink the means of
both computer output and input. In 1968, Englebart gave a demonstration of his
oN-Line System (or NLS), which represented a breakthrough in human computer
interaction, exhibiting many of the principles and technologies that are taken for
granted in modern (2011) computing, including the, now familiar, configuration
of a ‘mouse’, which translated hand movements into the visible movement of a
cursor, with a visual display unit, which allowed the user to select representations
of objects or functions directly rather than through programming using text-based
interfaces or punch cards.
With the mouse, Engelbart had created a physical bridge between
the computer user’s body movements and computer input and had opened the
door to new, more engaging, interaction methods.
Seeing the commercial potential of Engelbart’s invention, Xerox were
the first to develop it into a personal computer system using a GUI based on what
we now know as the WIMP (Windows Icon Pointer Menu) system, integrating
not only Englebart’s mouse and keyboard interaction, but calling on a host of
other newly invented graphical symbols and methods, notably so-called interface
‘widgets’ such as ‘icons’, used to represent files and applications as graphical
objects. The convergence of these ideas and systems resulted in Xerox’s devel-
opment of the Star office computer operating system (Bewley et al. 1983) and
the modern PC was born in 1982 with the Xerox Star 8010 Document Processor.

Virtual objects and spaces

While the invention of the mouse by the Stamford team opened the door for
Xerox, the first experiments with the GUI, the desktop interface of Star, didn’t
necessarily reflect Engelbart’s vision. Within their proposal for a GUI, Xerox con-
ceptualized the virtual information processes within the computer as objects,
albeit virtual ones, to be displayed graphically and to be ‘directly manipulated’.9
It may, in part, relate to Xerox’s history as a photocopying machine company
that they chose to build their GUI around a virtual office, or what would become
known as a desktop. However, the office metaphor, which was considerably
diluted in later versions of the GUI, notably from Microsoft and Apple, was not
Xerox’s key contribution. By taking a user-led approach to the design of their
computer interface, personal computing was tied to the logic of a simulated
material world. The contradiction of this endeavour appeared to frustrate the very
flexibility that digital systems create, as illustrated by a colourful tirade given by
the inventor of the term hypertext:

Paper. We’ve got to have media which are better than paper … there
are so many millions of graphical user interfaces possible and yet we
are stuck with one in which we have a single fixed little area called

80
 Between city lights receding and the non-space of the mind

the desktop – I don’t know why, I’ve never seen a vertical desktop –
where the icons are a fixed size and then they open to flat windows
which don’t have any perspective and don’t connect.
(Nelson 1998: 4)

For Nelson, the GUI of the modern personal computer was a sell-out; a product
of Xerox’s aim to simply simulate paper, disabling the user with the same restric-
tions that are placed on the physical organization of objects. However, although
not a commercial success, Xerox Star was to have significant influence on a new
generation of personal computers. Through these developments, abstract digital
information found a representation and a space of interaction that were united
through the GUI, a paradigm that has changed little since.

The re-emergence of the architecture

Its commercial success has made the GUI of the modern personal computer,
in all its flavours, ubiquitous and it is difficult to imagine another paradigm for
visualizing and organizing PC-bound information, but Nelson’s tirade reminds us
that the GUI is, in many ways, a compromise that has constrained our vision of
what computer-based information is and how we interact with it. However, much
of the research that followed Xerox, in the new domain of HCI, experimented,
not with alternative paradigms of human computer information representation,
but with extensions of the existing visual metaphors of the GUI. In an effort to
make computing accessible to a wider public, the search for intuitive computer
interfaces sought ever more elaborate interaction metaphors. Rather than a
desktop as the space of personal computing, why not, for example, have a whole
room or set of rooms (e.g. Henderson and Card 1987) or allow tasks to be spread
over multiple galleries (Robertson et al. 2000). Proposals also emerged for user
interfaces based on virtual offices, museums and even whole cities10 and some
even began to reflect, again, on the method of loci as a design paradigm.11 This
movement was based on a principle summed up by Kuhn and Blumenthal in
their 1996 tutorial called ‘Spatialization: Spatial Metaphors for User Interfaces’:

Space as we experience it daily, from our desktops through the

rooms and buildings we live in, to the cities and landscapes of our
environment, has essential properties required from source domains
of general-purpose interface metaphors.
(Kuhn and Blumenthal 1996: 346)

Based on the notion that space offers an intuitive interface for the representation
of information, Kuhn and others focused on architectural space as a metaphor
through which visual screen space can be articulated based on what they con-
sidered to be the naturally communicative properties of space.
Kuhn and Blumenthal’s work seems to represent the logical des-
tination for efforts to bridge the gap between computers and human beings.
However, in bridging this gap, there is also a contradiction. While information
is no longer restricted by its material manifestation, the paradigm of the GUI
places spatial restrictions back on information objects through its interaction

81
Between city lights receding and the non-space of the mind

and presentational rules. Information objects are conceptualized as existing in

one place and with spatially coherent combinations in the form of files contained
within folders, etc.

Part 4: The web is not a tree

If the GUI has been constrained by its visual space and has been, in the opinion
of some, stifled in its evolution, then the story of hypertext should surely offer
a different vision of a new type of information space. Hypertexts were invented
with the explicit intention of supporting multidimensional relationships between
information objects; in other words to be beyond combinatorial constraints.
However, even in the context of hypertext, the desire to create user-centred
systems has led to an implementation constrained by a requirement to find pat-
terns in their topological structure.
The term hypertext was coined by Theodor Nelson12 and the con-
cept was conceived as the basis for an open publishing network which would
not only act as a repository of documents but would support the processes of:
‘reconfiguration, comparison, and interconnection’ and the addition of ‘complex
version management and powerful user interface conventions’ (Wardrip-Fruin
and Montfort 2003: 301).
Although hypertext is a term commonly used to relate to the ‘chunk-
based structure’ of the WWW, Nelson envisaged hypertext as something more
sophisticated:

The web’s type of ‘chunk-style’ hypertext – static links that allow the
user to jump from page to page has been around for decades and has
been criticized for just as long. For Nelson, chunk-style hypertext is
just one subtype of hypertext, a term he introduced to mean ‘a body
of written or pictorial material interconnected in such a complex way
that it could not conveniently be presented or represented on paper’.
The ‘hyper’ in Nelson’s neologism does not mean ‘link’ but rather
connotes extension and generality: cf. ‘hyperspace’.
(Wardrip-Fruin and Montfort 2003: 301)

Nelson envisaged a system that was not only free of place but also able to
update in real time and be rapidly reconfigured and constantly updated. Nelson’s
hypertext project was never fulfilled but the term ‘hypertext’ stuck. It was
Berners-Lee’s linking of the physical structure of the ‘internet’, with his http
(hypertext transfer protocol), which popularized and defined what is now known
as the WWW. The conjunction of the internet and http lead to the emergence of
chunk-based hypertexts defined as:

[Information organized] as a network in which nodes are text chunks

(e.g. lists of items, paragraphs, pages) and links are relationships
between nodes (e.g. semantic associations, expansions, definitions,
examples …)
(Rouet 1996: 3)

82
 Between city lights receding and the non-space of the mind

The convergence of the internet and the hypertext of the WWW allowed infor-
mation to become de-spatialized in two ways. First physically, in terms of the
network infrastructure where information is stored and accessed through remote
servers, and where physical location and distance became largely irrelevant to
the speed and quality of access; and second, in terms of the representation of
information to the user. There is no need for hypertexts to be topologically limited
and through the technology of the WWW, a new type of ‘information space’
was born, a material-less ‘space’, shaped not by physical location but by the
relationship between the internet’s structure and the virtual hypertext network
of the WWW. This web of information is potentially capable of almost infinite
permutations and combinations.

The WWW’s pattern language

It is against this aspirational context for the development of hypertext that Tim
Berners-Lee claimed, in the late nineties, that the WWW was a ‘web not a tree’,
in reference to the conception of hypertext as ‘a non-structured set of connec-
tions, links that are not limited by a pre-determined pattern’ (Durand and Kahn
1998: 1). This statement is an echo of Christopher Alexander who, 30 years
earlier, had expressed a similar sentiment when discussing the designs of new
towns in his paper A City is Not a Tree (Alexander 1965).
What Alexander had described was an approach to town planning
that restrained the design of new urban areas to street patterns and zoned blocks
of cores and peripheries of branching, tree like structures. Alexander criticized
this prevalent approach to city design, comparing the top-down design of new
city developments to those built environments that had evolved over time,
embedding networks of social relationships into their fabric. Whether Berners-
Lee was aware of Alexander or not, the similarity between both their statements
is not coincidental. Both men were discussing the same type of structure and
the same conceptual problem. Alexander’s concern was that hierarchical urban
planning divided the environment into spatial units that were limiting and did not
reflect the social relationships that formed part of a city’s system and which are
manifest in its pathways, allowing different parts of the city to communicate with
one another. In Berners-Lee’s discussion of the WWW, he is describing informa-
tion objects, in this case hypertext documents, and their semantic relationships
to one another; the structure of associations that define those relationships
and the problems inherent in limiting them to fixed predetermined patterns. In
the case of the WWW, these are hyperlinks. The context of Berners Lee’s and
Alexander’s work is different but the two discussions can be united by separat-
ing, as Alexander did, fixed spatial relationships from the more abstract dynamic
patterns that govern the way in which these spaces are experienced.

The geometry of topological semantic space

The comparison between Alexander’s discussion of the city and Berners-Lee’s
discussion of the web is more than a useful analogy. The method of splitting
architectural elements into atomic components that express a reduced rep-
resentation of the environment’s geometry has been mirrored by those who
interpreted topological systems, such as hypertexts, in relation to apparent

83
Between city lights receding and the non-space of the mind

spatial geometries, despite the fact that there is no technical reason to do so.
Bernstein points out, in his history of hypertext patterns, that once hypertext sys-
tems began to be used outside technology development laboratories, the early
hope for systems that would allow for the free association of documents and
other information objects gave way to a realization that structural geometric pat-
terns were necessary if such systems were to be used effectively. The challenge
of hypertext interaction has been commonly described as ‘the lost in hyperspace
problem’ (Edwards and Hardman 1989). This problem gave rise to speculation
on how structures of hypertext could be constrained into coherent patterns:

Before 1987, hypertext writing tools were laboratory curiosities; after

1987, systems like Guide, HyperCard, and Storyspace became readily
available and were widely employed. Early systems often reflected the
prevalent concern with navigation. For example, Peter Brown, Guide’s
developer, argued that emphasizing a hierarchical structural backbone
would render hypertexts more comprehensible to users (Brown 1989).
To the widespread emphasis on tree-structured hypertexts was added
speculation on the utility of hypercubes, toruses, and lattices (Parunak
1989) and Petri nets (Stotts and Furuta 1989), while Polle Zellweger
argued in an immensely influential paper that guided tours along
clearly-marked paths help keep readers oriented (Zellweger 1989).
(Bernstien 1999)

These patterns of hypertext result from the notion that authorial intent can be
expressed through the pattern of relationships as well as the content of hyper-
text documents themselves. In addition, it is also possible to discover recurring
patterns in the way in which hypertext topologies are perceived by computer
users from the analysis of user behaviour in interacting with websites (see for
example Parunak 1989). From such studies, common patterns can be uncovered
and examples include linear (path), hierarchy and hypercube and, by far the most
common navigation strategy, the ‘hub and spoke’ navigation (Figure 2.5), where
a user starts from a main page and navigates deeper into the site before click-
ing on the back button repeatedly to return to the start point before embarking
on another path (Pirolli et al. 1996). These patterns are all characterized by their
adherence to topologies supported by the rules of 2D topological geometry. In
other words, such patterns represent coherent spatial structures where each
node could constitute a place within a planar topological geometry.
This tendency to seek patterns within hypertext structures has
undoubtedly effected the design of websites. Visit any major website now and
it is likely that it will have been created by a team of ‘information architects’
and you will be presented with a similar template of tabulated menus offering
hierarchical ‘navigation links’. This presentation may or may not be indicative of
the actual structure of the website. A study, undertaken by the author in 2005
graphically illustrates the gap between actual hypertext structures and their pres-
entation through individual web pages. The study uses a web tracker to retrieve
link information for specified websites; the data obtained is then visualized using
a force-directed graph that presents the pages as nodes and the links between

84
 Between city lights receding and the non-space of the mind

2.5
Diagrams to show common
hypertext patterns. From top
to bottom/left to right: linear,
hierarchy, hypercube and hub
and spoke

them as lines. The force direction algorithm forces the nodes apart, uncovering
hierarchies between pages and attempting to simplify the readability of the
graph. In the cases shown in Figure 2.6, a conventional hierarchical layout for
the pages for both sites disguises complex networks of links, because the links
highlighted for navigation, through tabulated menus, are separate from other
semantic links, which are embedded in the site’s text.13
Critical to our understanding of these websites’ presentational
and primary topology is an understanding of the term ‘navigation’. Navigation
may simply be a metaphor for the process of link traversal, but the idea that
hypertexts are navigable would also account for the topological restrictions
reconnecting topos and topic in ways that have little to do with the technical

85
2.6
Diagrams to show the
results of two web-mapping
experiments conducted
on websites at Cambridge
University. Both snapshots
were taken in 2004 and
the resulting graphs are
drawn using a force-directed
graphing algorithm
 Between city lights receding and the non-space of the mind

potential of the hypertext system itself. Inevitably, these topological restrictions,

along with the notion of navigation, give way to conceptions of hypertexts as
being akin to architectural space, with hypertexts considered as adjacencies in
an information system.

Part 5: Remediating space

While the emergent spatiality of digital information systems has not been criti-
cally analyzed in architecture, it has been explored in relation to a more general
understanding of media developed for the analysis of digital text. Through a
concept described as remediation, Bolter and Gromala have sought to under-
stand the underlying nature of digital media in relation to historical media types,
stating that ‘Like the printed book, film, and television before it, the computer
is not a neutral space for conveying information’ (Bolter and Gromala 2000: 77).
A key tenet of their approach is an understanding that a characteristic
of digital media is its potential for mimesis, in other words to copy aspects of
other media types. Digital technologies have not reinvented the media landscape
overnight, instead they simulate patterns found in other media forms. Bolter
(2001), Aarseth (1997) and Landow (1997) have focused on hypertext and the
remediation of textual media and the possibilities of multilinear reading and
writing. In particular, the discussion of literary hypertexts has necessitated the
development of conceptions of reading which lie outside literary study and New
Media theorists such as Bolter (2001) and Snyder (1997) have looked increasingly
toward spatial and visual metaphors to describe the unique characteristics of
textual media. In studying, for example, storytelling in literary hypertexts, narra-
tive structures are often connected with the process of journey, with hypertext
theorists such as Rossello quoting philosophers such as de Certeau, who states
that ‘Every story is a travel story – a spatial practice’ (de Certeau 1983: 115).
Through their wanderings in hypertext, Rossello argues, the reader is a flaneur,
able to ‘invent new paths toward old destinations’ (Rossello 1994: 134).

Writing space
Focusing on the work of Bolter and his unambiguously entitled Writing Space
(2001), it becomes clear that, although his work is ostensibly on the remediation
of text, this process is inexorably linked to the space of textual representation:

In about the 8th century BC, the Greeks began to refashion the space
of oral mythology and heroic legend into the more precise linear
space of the papyrus roll …. In late antiquity the shift from papyrus
roll to codex refashioned the space again, making more effective
use of the two-dimensional surface to deploy text. In Western
Europe the shift from handwritten codex to printed book was
another such refashioning, and the shift to electronic writing is yet
another.
(Bolter 2001: 23)

Just as ‘fashioned’ spaces define the papyrus role and the printed book,

87
Between city lights receding and the non-space of the mind

electronic writing carries its own spatiality, made clear in the language used to
describe the WWW:

We tend to conceive of hypertext spatially: the links constitute a path

through a virtual space and the reader becomes a visitor and traveller
in that space. We say that the reader or user ‘visits’ web pages in
California, Germany, or Japan, when in fact we could just as easily
say that the pages come to her.
(Bolter 2001: 29)

The notion of remediation causes Bolter to reflect on the emergence of space

as a means of organizing texts, which reflects the classical binding of topos and
topic:

A text as a network may have no univocal sense. It can remain a

multiplicity without the imposition of a principle of domination. In
place of hierarchy, we have a writing space that is not only topical;
we might call it ‘topographic.’… It is not the writing of a place, but
rather a writing with places as spatially realized topics … The reader
and writer can create and examine structures on the computer screen
that have no easy equivalent in speech.
(Bolter 2001: 36)

Here Bolter conflates two conceptions of space. Topography, as it is used in

contemporary language, refers to an ‘accurate and detailed delineation and
description of any locality’ (OED) and implicit in the definition of topographic
space are the properties of metric distance and absolute position, in other
words a Cartesian sense of space as it is measured, represented and, in this
case, mapped. This topographic space is in keeping with the notion of a screen
space where objects are given positions based on their display on the pixel grid
of the screen’s surface. In his book, Bolter’s examples are clearly topographi-
cal, particularly in his chapter ‘The Breakout of the Visual’ (Bolter 2001: 47–75),
where his concept of remediation focuses on the topography of the screen as
a visual space, the rules of which define its graphical presentation, and on the
graphical language that defines its communication. For Bolter, this visual space
is a natural territory through which he is able to articulate the visual language
of assemblages of text and image and the new role for the ‘symbolic image’
(particularly in graphical user interfaces). The result of this apparently singular
definition of the reading and writing space of hypertexts is that Bolter chooses
examples of literary hypertext where there is an explicit relationship between
visual images or diagrams and the text associated with them, or what he terms
‘image maps’.
For Bolter, space and language are inexorably connected. The sys-
tem of textual symbols is among many formal sign systems that rely on their
organization in a topographical space in order to communicate. Digital media has,
Bolter suggests, created a new, more flexible type of space through the visual
topography of the screen. The screen is, in turn, the material surface on which

88
 Between city lights receding and the non-space of the mind

textual representation is possible, in the same way that the material page is
the surface of textual communication for the book, which is articulated through
printed letters. However, Bolter’s linking of topos and topic, where the writing
of ‘places’ through ‘spatially realized topics’, is much closer to the conception of
a topological space encountered in the discussion of mnemotechnics, and the
method of loci, as was set out in Chapter 1.

Semantic space
Bolter is not the only theorist to have noticed the prevalence and importance
of space in new media. Kaplan and Moulthorp have attempted to rationalize
apparent conflicts in spatial definitions in hypermedia in a similar way to Bolter,
but using different terminology, by distinguishing ‘architectonic space’ from
‘semantic space’ where architectonic space is the

space of the screen … defined in terms of pixels and other coordinate

systems … a regular, mathematically precise space in which absolute
principles apply: objects are uniquely identified, have exclusive loca-
tions, and obey rules of geometry and perspective.
(Kaplan and Moulthorp 1994: 267)

This is as opposed to what they describe as semantic space, which is:

deeply connected to the production of meaning, interpretation and

other activities involving symbols.
(Kaplan and Moulthorp 1994: 267)

The effect of Kaplan’s and Moulthorp’s definitions is to separate the spatial

structure of information (semantic) from the physical or visual representation
(architectonic). The problem, Kaplan and Moulthorp argue, is one of dimensional-
ity since it is inconceivable that a potentially multidimensional information space
can be represented in the limited dimensions of onscreen representations. The
problem of realizing representations of semantic space through architectonic
space is usefully summed up by Dillon et al., who discuss semantic space as an
‘abstract psycholinguistic concept’:

We cannot navigate semantic space, at least not the way we navigate

physical environments, we can only navigate the physical instantia-
tions that we develop of the semantic space.
(Dillon et al. 1993: 187)

Although Kaplan and Moulthorp and Dillon et al. are correct to isolate the problem
of translation between architectonic space and semantic space, there is another
problem since if, as Dillon et al. suggest, semantic space cannot be navigated,
other than through ‘physical instantiations’, then how is it made manifest – is it
really a space at all?

89
Between city lights receding and the non-space of the mind

Part 6: Conclusion
In this chapter, I have traced the erosion of the place-based method of infor-
mation representation as an evolution rather than a revolution. Through the
combinatorial machines of Ramon Lull, described in Part 2, I have shown that,
while basing his ideas on a place-based understanding of information, Lull cre-
ated machines within which meaning can be encoded in the potential of a system
as opposed to a fixed spatial structure. The idea of information as enacted
through mechanical processes was further refined and developed in relation to
the evolution of logic and ultimately the invention of computation. I also showed
that, in attempting to escape the information object of ‘the book’ Otlet and Bush
were able to consider information as a ‘something’, independent from its material
manifestation and, although Otlet was interested in the a literal architecture of
information through his plans, with Le Corbusier, for the Mundeneum, his new
building typology shifted the role of architecture from that of organizing objects
towards one of setting the scene for information gathering and retrieval.
The digital revolution should have allowed for the computational
realization of Bush’s ‘Memex’, where information objects could be associated
into patterns that were not restricted by spatial coordinates or relative locations.
As I showed, however, two paradigms in the representation of information, GUIs
and hypertexts challenge the potential of the computer as a way of representing
information, through their user interface conventions.
In Part 3 I showed how GUIs have sought to control the space of
human interaction with the computer through ever more elaborate interface
conversions, from the desktop through to metaphorical rooms, buildings and
cities.
I also illustrated, in Part 4, how hypertexts, which represented the
most direct vision of Bush’s Memex system with the possibility of a networked
multidimensional topology, were from their earliest incarnations, limited by the
apparent necessity to pattern them. This patterning contradicts the notion of
hypertexts as vehicles for free associations but, rather, imagines hypertext docu-
ments as ‘places’ which are ‘navigated’.
Finally in Part 5, I reviewed the concept of ‘remediation’, where old
media forms are used as a way of understanding our interactions with new
media. Starting with Bolter’s remediation of text, I suggested that many of the
features of digital media and its representation of structured information are
spatial. This spatiality makes itself manifest through the appearance of architec-
tural metaphors in developing GUIs and the notion of patterning hypertexts and
constitutes, I suggested, a remediation of architectural space.
It is clear from this broad view of the dematerialisation of informa-
tion across the paradigm of the GUI and hypertexts that while digital technology
has had a profound effect on the organization and distribution of information, by
separating information from its material means, the interfaces through which
information is perceived and interacted with, present a new set of architectonic
systems that still seem to be constrained by recurring metaphors of space.
Indeed, in his discussion of the writing space of digital media, Bolter appears to
have brought the discussion full circle by proposing that a type of spatiality found
in the method of loci is the key to understanding digital media.

90
 Between city lights receding and the non-space of the mind

Far from freeing information from the ballast of materiality, attempts

to design interfaces for digital information seem to be searching for new spaces
of interaction, whether they be realized as icons distributed across the desktop
of a computer screen or through the navigational topology of the WWW. Far from
the non-space of the mind, computational architectonic systems have revealed
a proliferation of multiple spaces constrained by our conception of the physical
world. Even when machines allow us to lose the ballast of materiality, World
Three seems to retain the ghosts of World One.
The persistence of these ghosts has frustrated some, but the pro-
liferation of spaces for interacting with digital information, I will suggest, is not
accidental or simply the result of weak spatial and architectural metaphors.
In Chapter 3, I will attempt to provide an understanding of these spaces of
information by revisiting architectonic systems and searching for the space of
computational information.

91
Chapter 3

The spaces
of information
Space is not a setting (real or logical) in which things are arranged, but
the means whereby the position of things becomes possible. This means
that instead of imagining it as a sort of ether in which all things float, or
conceiving it abstractly as a characteristic that they have in common, we
must think of it as a universal power enabling them to be connected.
(Merleau-Ponty 2002: 234)

Part 1: Spaces of information

In Chapter 2 it was shown that, despite information and its representation hav-
ing lost the ‘the ballast of materiality’, whether through the GUI of the personal
computer or the conceptualization of the WWW and hypertext, the notion of
information as articulated in space, and through the organization of material
objects and patterned places, has evolved but has not been lost. In fact, the post-
digital era has seen a proliferation of ‘spaces’ of information and human computer
interaction, described variously as cyberspace, information space, semantic
space, topographical writing space, etc. Furthermore, the conception of digital
information as spatial has promoted the use of spatial, including architectural,
metaphors in attempting to articulate it. However, while spatial metaphors have
persisted, architectural metaphors have been limited in their success.
In this chapter, I will define ‘information space’ in relation to the
nature of a computer user’s interaction with the digital system. Following Borges’
story, ‘On Exactitude in Science’, which tells of the obsession of accuracy in
map-making, we might conclude, in relation to the cartography of real spaces,
that the map is not the territory. However, in considering information space, we
might have to conclude that the map is the only territory. Just as maps simplify
and filter space, information space is revealed only through alternative patterns,
made visible through the processes of analyzing and visualizing/mapping informa-
tion and through the interactive dialogue between human and computer, which
changes how the information space is viewed. I will address the key underlying
questions: How can information spaces be considered spatial? What kind of
spaces are they? I will analyze three spaces of information and suggest a fourth
(see Figure 3.1).
This chapter will begin by assessing the comparative failure of the
architectural metaphor as it is applied to the development of GUIs (both for the
The spaces of information

3.1
In an attempt to mediate
computer-based information
architectural metaphors have
often been used to give a
structure to an otherwise
structureless information
space

PC and as ways of visualizing the WWW). It is now acknowledged that attempts

to articulate digital information as architectural space were short-lived and super-
ficial, neither reflecting the complexity of the architectural spaces on which they
were based or the information spaces they represented. I will look specifically
at the notion of navigation and instances of the use of Kevin Lynch’s The Image
of the City (1960) as a way of understanding ‘navigation’ when interacting with
computer-based information. Through this analysis, I will examine the problem
of considering information spaces as homogenous phenomena.
In response, Parts 3, 4 and 5 will examine information space as a
heterogeneous phenomenon, using three separate spaces of information, which
I call semantic space, screen space and interaction space (see Figure 3.2):

94
 The spaces of information

3.2
In reality, information space
is mediated by multiple
types of space described in
the middle section of the
diagram, from left to right,
as semantic space, screen
(visual) space and interaction
space

1 Semantic space: This term is derived from the work of Kaplan and
Moulthorp (1994: 267) and defined as the structure of information held
within a computer. It is potentially multidimensional and, in practice, con-
sists of information objects and their relationships to one another described
through, for example, hypertext links or database fields. Semantic space
can be created by individuals, as was demonstrated in the description of
hypertext patterns in Chapter 2, but also through automated processes.
Semantic space can therefore emerge from computational systems used
to analyze information and find meaningful patterns.
2 Screen space: The concept of ‘screen space’ emerges as a result of sepa-
rating out the visual component of information space and defining the space
of the screen as separate from the space of interaction. Screen space is

95
The spaces of information

constrained by the number of available pixels and the visual language used
to display information. It can include web pages, a desktop GUI, a complex
graph-based visualization or a 3D world, and each type of representation is
constrained by metaphors and their visual rhetoric.
3 Interaction space: This describes the input actions of a user that change
the computer’s output. For example, clicking on a link causes a web page to
change its content; typing in a search query will cause a list of web pages
to appear in Google; pressing an arrow key may cause a virtual camera to
move through a 3D space; grabbing a document icon with the mouse curser
will cause it to move as if it is being directly manipulated. Different types
of interaction are mapped on to different types of screen space so that, at
best, the two reinforce one another.

Finally, in Part 6, I will seek a new understanding of architectonic space as a

fourth type of information space, derived from the three spaces of information
described above. I will focus on the idea of architectonic space as a result of the
relationship between the screen display of information and a specific method of
interaction, namely ‘navigation’. Part 6 proposes that ‘navigation’ is more than a
metaphor but is, rather, a mental schema.
In defining each of these spaces of information, I will show how
our interaction with each representation, whether it be semantic space, screen
space or interaction space, is governed by modes of address that relate each
one to our body. Indeed, I want to reveal how the fact that we use space at all
as a method for conceptualizing these different spaces is indicative of a need to
engage with digital information by using the same frameworks with which we
measure, analyze and perceive our physical world.

Technological context
In addition to introducing these spaces of information, this chapter will continue
the somewhat fragmented timeline begun in Chapters 1 and 2 by referencing
some examples that bring the story of the architecture of information more
up to date. To do this, I will introduce the disciplines of IR, InfoVis, interaction
design and HCI. This chapter cannot provide an exhaustive description of all
of these disciplines but, as in previous chapters, I will attempt to provide an
overview of illustrative projects and concepts that paint a broad picture of these
fields and their relationships to one another. It should also be noted that, in the
context of this chapter, I am deliberately limiting the study to screen-based
information, and interactions via mouse and keyboard, and latterly, touch screen
displays. In doing this, I do not wish to claim that screen-based interactions
are the only ways of interacting with digital information, far from it, but by
focusing on these types of interactions I am laying the foundation for a more
detailed consideration of ‘embodied computing’ freed from the visual screen in
Chapter 4.

Part 2: The failure of architectural metaphors

While the desktop of the PC’s GUI has evolved we don’t, for the most part,
interact with digital information through images of a virtual building or city. The

96
 The spaces of information

‘hysteria of total simulation’ as Johnson (1999: 73) put it, which gave rise to
many GUI proposals, has been greeted by some with incredulity and scepti-
cism. For example, in his review of 3D extensions to graphical user interfaces,
the influential HCI expert Ben Shneiderman cast doubt on GUIs which use 3D
spatial representations which simulate architectural spaces as well as other
material artefacts:

Users might find these interfaces enjoyable, recognizable and memo-

rable … but they can also distract and confuse because of increased
visual complexity … the compromises needed to produce 3D effects
undermine usability.
(Shneiderman 2003: 14)

It seems increasingly unlikely that the next generation of GUIs will be realized
through a metaphor of space in 3D. Although 3D GUIs would seem to be natural
extensions of the 2D office environments of common personal computer inter-
faces, there is a conflict between metaphors here. In reality, the GUI of a PC is
not limited to a 2D space, with objects arranged in physical ‘absolute’ dimen-
sions. Far from being absolute, defined by the edges of a screen or a spatial
territory (in 2D or 3D), the reality of the GUI’s representation of information space
is relative, driven not by locations and coordinates but by places and contexts.
The complexity of the real environments on which the metaphorical extensions
of the limited ‘desktop’ rely, has been lost in anaemic spatial representations that
extend 2D and 3D space, assuming that information objects must be represented
in a consistent Cartesian space.
The metaphorical spaces of the desktop GUI and its descendents are
not driven by scientific evidence on the effectiveness of metaphorical projec-
tions of real spaces when interacting with digital information (Dodge and Kitchin
2001b: 178) but, it seems, through cultural instinct. This instinct is driven through,
as Blackwell suggests, a presentation of the physical world in which, in the words
of the Xerox Star team, ‘display becomes reality’ (Blackwell 2006: 517).

Navigation in hypertext
At the core of digital media and its visual representation is the notion of naviga-
tion. Despite the fact that, for example, clicking on a hyperlink to a web page
has very little to do with the process of navigation (clicking on a link involves
sending a message to a remote server to send packets of information back), the
metaphor of navigation is pervasive and very rarely questioned. In fact, the notion
that the web offers a navigable infrastructure is at the heart of the way many
websites are designed. As was shown in Chapter 2, the ‘lost in hyperspace’
problem creates the necessity for limited topologies by associating the link with
an adjacency between two places. It seems counter to a definition of hypertext,
however, to consider a link to be the equivalent of an adjacency. Indeed, the
hyper component of hypertext should militate against this understanding, yet
the navigational metaphor persists.
The conception of hypertexts as a navigational medium results in
a trajectory which, in a similar way to extensions of the desktop metaphor to

97
The spaces of information

include more expansive spaces, leads towards a consideration of hypertexts as

equivalent in structure and form to urban space. The city might not be a tree and
the web might not be a tree, but could the web be a city?
Conceptions of the web as a city are illustrated most clearly by
Andreas Dieberger (1995, 1996, 1997), who sought to solve the lost in hyper-
space problem by referencing urban environments, drawing on the writings of
Kevin Lynch and basing a design proposal for hypertexts on textual virtual envi-
ronments (TVEs) based on multi-user domains/dungeons (MUDs).
Text-based MUDs are virtual representations of fantasy spaces where
the player is, usually, involved in a quest-like game where multiple players can be
involved at once. While modern MUDs tend to be played in 3D real-time virtual
environments (3DRTVEs), earlier (before sophisticated computer graphics) exam-
ples are text-based. A place within the game is not represented visually but rather
described textually, with exits from the place often indicated by the poles of a
compass. By playing the game, each location can usually be mapped into a coher-
ent spatial pattern,1 with the addition of occasional ‘magic features’, which allow
the player to teleport to remote points on the map without having to traverse the
intermediary spaces. These features are necessary in large MUDs, but are also
used sparingly in order not to destroy the continuity of spatial organization and
confuse the players, allowing them to navigate through the implied continuous
virtual space. A similar rule must be used in hypertexts, suggests Dieberger,
and by doing so navigational and teleporting links are separated. Extending this
link differentiation and by considering hypertext to be a navigational medium,
Dieberger is able to entertain a universal city metaphor for hypertext systems.
Dieberger bases his proposal on Kevin Lynch’s The Image of the City, and
uses the idea of ‘imagability’, which Lynch applied to urban environments, as
a way of articulating hypertext structures with an explicit aim of supporting
navigation.

The cognitive mapping of information visualization

Dieberger’s proposition was not unique and the logical trajectory of this research
leads towards a conception of hypertext as being visualized through spatial inter-
faces. The assumed necessity of navigational support for information space can
lead to more elaborate relationships being drawn between information space and
real spaces and The Image of the City is unique in offering a design template for
the creation of what Lynch describes as ‘imageable’ environments.
Central to Lynch’s theory is the fact that environments can be
designed in such a way as to be remembered vividly. This not only makes such
environments easier to navigate but also leads to a richer experience of inhabiting
them. In order to make his case, Lynch put forward a theory that is essentially
topological in that it involves articulating city elements as either nodes or the
edges which link them, as follows.

1 Paths are channels along which the observer customarily, occasionally or

potentially moves. They may be streets, walkways, transit lines, canals or
railroads …. People observe the city while moving through it, and along
these paths.

98
 The spaces of information

2 Edges are linear elements not used or considered as paths by the observer.
They are boundaries between two phases, linear breaks in continuity ….
They are lateral references rather than coordinated axes. Such edges may
be barriers, more or less penetrable ….
3 Districts are medium-to-large sections of the city, conceived of as having
two-dimensional extent, which the observer mentally enters ‘inside of’ and
which are recognized as having some common identifying character ….
4 Nodes are points, the strategic spots in a city into which the observer can
enter, and which are the intensive foci to and from which he is travelling.
They may be junctions, places of break in transportation, a crossing or
convergence of paths, moments of shift from one structure to another. Or
nodes may be simply concentrations which gain their importance from being
the condensation of some use or physical character ….
5 Landmarks are another type of point-reference but in this case the observer
does not enter within them, they are external. They are usually a rather
simply defined physical object: building, sign, store or mountain …. Their
use involves the singling out of one element from a host of possibilities.
(Lynch 1960: 47–8)

Lynch’s theories were aimed at an audience of designers of the built environ-

ment and yet his elements of the city have been appropriated well beyond
their architectural scope. Perhaps Lynch’s appropriation is less surprising in
the context, for example, of Dieberger’s pursuit of the city metaphor, but even
in systems that provide abstract or only loosely metaphorical visualizations of
information, Lynch’s elements still seem to provide some basis for design. For
example, the influential LEADS (Legibility for Abstract Data Spaces) (Figure 3.3)
project and VR-Vibe system, developed at the University of Nottingham to
visualize databases, are far from being metaphorical urban environments yet,
at their core, Lynch’s theories are still applied2 and, through such examples, the
association between urban environments and information environments can be
clearly seen.
The problem with using The Image of the City is that it does not pro-
vide a general framework for describing the cognitive map, which is a complex
and multifaceted phenomena and the book is very specific about the architectural
context of the theory. In fact it highlights, first, the distinctly individual nature of
the city image (Lynch 1960: 8–9) and, second, it describes memory and percep-
tions of the city as a collage of interrelated spatial objects that can be described
in terms of the five elements, but which only make sense when those elements
are combined together and experienced through personal perception and through
long-term learning and discovery (Lynch 1960: 91–117).
If Lynch does not provide a distinct design methodology, how is he
being used? The answer to this question differs in the two cases described
above.
For Dieberger, the urban nature of hypertext lies in its topology. The
metaphor of the web as a city comes about because hyperlinks are considered
as adjacent places, part of a planar network. It is not clear in Dieberger’s work
to what extent the necessity to limit topology is an emergent feature of the

99
The spaces of information

systems themselves or is a constraint that comes about because of the city 3.3
Screen shots from LEADS
metaphor. However, we can surmise, given the descriptions of the authoring of showing an untreated 3D
hypertext topologies described in Chapter 2, that articulating hypertext topology information space (left) and
as like a city probably reflects an already-present notion of topological patterning. the same space articulated
using Lynch’s elements of
Lynch’s work is useful because it provides, through the concept of imageability, environmental structure
a topological description of space. Landmarks and nodes can be inscribed as (right) and annotated to
show the articulation of
pages, with the ‘districts’ being web ‘sites’ which are connected by the ‘routes’
Lynch’s elements (bottom)
and edges of internet links.
In the case of Lynch’s application to the visualization of information,
the articulation of the screen is not topological but topographic. The environment
created by LEADS, for example, is no more like a city than any other diagram-
matic representation and here, Lynch’s elements are not used as support for
a cognitive map, but are translated directly from the diagrammatic notation in
Lynch to a diagrammatic notation on screen.
Landmarks, for example, become tall or visually distinctive objects,
rather like skyscrapers; edges become bounded regions separating districts
which, in the case of LEADS, are coloured spatial zones crossed by paths which
do not necessarily represent lines of travel, but which are independent from

100
 The spaces of information

edges. Finally, there are nodes that often represent the information objects
themselves as blobs or blocks scattered through a 3D environment (Figure 3.3).
Although the diagrammatic notations that Lynch uses are illustrative of a more
subtle set of elements within these information environments, the notations
themselves are being literally translated into objects on the computer’s screen.
Lynch is used as a graphic design handbook. The ‘elements’ no
longer function as Lynch suggests they do in real urban environments, but act
as notations, enriching the visualization system’s graphical presentation. The
cognitive map in information visualization is considered to be an add-on, simply
a way of articulating the topography of representation that makes the diagram
clearer. The role of spatial cognition in the representation of information, how-
ever, is clearly more complex.
Observing the adoption of Kevin Lynch’s theories in the context
of InfoVis and hypertext research reveals the now familiar recurring theme of
two different types of space: one is hidden in topology, the other revealed as
topography.
In order to understand the many and various spaces of digital informa-
tion, we need to have a better understanding of what they are and to rationalize
the many different definitions that exist. A starting point for this process is to
recognize that there is no universal ‘cyberspace’ and thus no single architecture
of information. Applying architectural theories such as Lynch’s imageability is
simply a way of expanding an already-present metaphor, in this case the meta-
phor of navigation, but without questioning the metaphor’s basis.
The separation of information from its material means has differenti-
ated between where and how information is stored and structured, and where
and how information is presented.

Part 3: Measuring – semantic space

Semantic spaces, as defined here, constitute structures of virtual information
held within computers. The notion of information held ‘within a computer’ is a
vague one. Information is generally stored on physical media within computers,
often on the hard drive, and, to the extent that these storage media represent
physical materials, computational information is stored as physical manifesta-
tions. As we have seen, however, the success of systems such as the WWW is
based on a conceptualization of linked information that does not depend on the
physical locations in which it is stored. The physical manifestation of information,
therefore, is distinct from the means by which digital information is retrieved,
processed and structured at the level at which the information is conceptualized
by the computer user. On the web, information structures are conceptualized in
terms of connectivity, where a hypertext page is defined by a network of rela-
tionships with other pages. This network topology can be contrasted with other
conceptualizations of information. For example, objects within a database may
be conceptualized in relation, not to their connectivity, but to their containment.
Containment still represents a topological relationship but, rather than an object
being linked to another, it is contained within the same set. A simple example of
where containment is made visible would be the database used to split books
into subject categories on a website such as Amazon. For example, a book such

101
The spaces of information

as Thomas Markus, Buildings and Power, is placed within Amazon’s database in

a hierarchy consisting of:

• Books > Art, Architecture & Photography > Architecture > Criticism &
Theory
• Books > Art, Architecture & Photography > Architecture > Planning
• Books > History > Academic History.

In other words, each field in the database consists of containers and sub-
containers and, in this case, Buildings and Power is located within three con-
tainers: ‘Criticism & Theory’, ‘Planning’ and ‘Academic History’. Buildings and
Power, therefore, could be said to exist in three locations in Amazon’s database.
By describing the Amazon database as a set of hierarchical containers, I am
describing it through a conceptualization. In the case of this database, I could
equally say that categories act as pointers to the book objects and that Buildings
and Power exists on three branches of a classificatory tree. The concept of digital
information tends to be structured by spatial schema which treats digital informa-
tion objects as if they were located in a topological space and the nature of this
conceptualization is governed by factors which include the following.

• The definition of the information object itself; i.e. how do we define the
minimal unit of information of a book? Do we stop at the book, a chapter, a
paragraph, a sentence or even an individual word?
• The nature of the organizing principle in terms of whether the information
objects are organized with relation to each other (for example a book might
be located as it relates to the semantic networks of books it cites and books
cited by it) or against a master categorization system.
• The process of structuring information in terms of whether the structure is
emergent; i.e. the result of computational analysis, or has been specifically
designed, as is the case of, for example, a library classification system.

Semantic space, Dillon et al. (1993) claimed, cannot be navigated directly but
is made visible through ‘alternative instantiations’. To this end, it might be sug-
gested that semantic space comprises the computer’s internal representation of
information and that this space has the quality of being topological, that is to say
it is composed of information objects and their associations to one another, inde-
pendent of metric distance and geometry. This potential for multidimensionality
does not, however, preclude the possibility that semantic space has a structure,
even where that structure is not directly interacted with by a human user. In fact,
meaning only emergences once information has been conceptualized as having
a spatial structure and where this structure constitutes a restricted topology. For
example, if we consider a linear string of words (where each word is considered to
be an independent information object) constituting a sentence, then the meaning
of the sentence would not be enhanced by the freeing up of topological restric-
tions. Imagine the same sentence but written in such a way that each word could
be read in conjunction with any other, in the form of a rhizome structure. Not only
would this structure be very difficult to read, but the words would appear as a

102
 The spaces of information

3.4 meaningless jumble. It is only when the topology of the word space is reduced,
Diagram to show the affect
so that they can be read as a linear sequence governed by the rules of syntax and
of topological incoherence
in an individual sentence grammar, that meaning is conveyed in a useful form (see Figure 3.4).
where each word constitutes In Chapter 2, I showed how authors of early hypertexts looked
information object
towards topological geometries as a way of giving structure to the reading and
‘navigation’ of their information spaces. In the twenty-first century, a new domain
for the structural understanding of information has arrived, where the emphasis
is on understanding emergent information systems through the study of their
content and the implicit links that information objects have to others as part of a
system. This development is perhaps best understood in relation to the change in
the way that the WWW is conceived and interacted with. Rather than navigated
via hypertext links, it is likely that many, if not most, interactions with the web
start with the input of a query into a search engine.

Semantic space as a tool for analysis

In his introduction to Visualizing Data (2008), Ben Fry identifies the stages in
extracting useful information from data as: acquire, parse, filter, mine, represent,
refine and interact. Each stage marks a programmatic point on a journey which
starts with raw data and ends in a readable display of information. Semantic
space, as an analytical method, exists through the process of parsing, filter-
ing and mining the information, where data is analyzed and given shape. Any
information retrieval system is effectively a series of translations from raw data
to ordered data, where some form of filtering has taken place to reveal salient
structures, sometimes with the view of visualizing the results in the form of a
diagram or to support the process of search and retrieval in complex information
space.3 Each translation involves a cost in terms of computing power and, by
association, time. Each translation also involves the act of leaving out information
and thus stripping back the system’s topological complexity by amalgamating or
ignoring some connections between information objects.
There are many ways of achieving this simplification, most of which
involve some form of topological manipulation based on finding spatial ways of
reducing the dimensions of relationships between information objects.4 Methods
of IR, for example, use notions of distance in discovering relationships between

103
3.5
Diagram to illustrate the
semantic distance model
applied to words in the
semantic network of a
thesaurus
 The spaces of information

text documents where distance indicates similarity between content as in, for
example, the Semantic Distance Model (SDM) (Brooks 1995), which analyzes
comparative meanings between words by using a thesaurus and finding groups
of words with a similar meaning. So, for example, a word such as ‘game’, may
be considered equivalent to ‘contest’ so, rather than creating two links between
documents a single link is created associating equivalent meanings rather than
just words. Using the thesaurus, meanings between words are measured by
their distance from one another in the thesaurus’s network of associations. For
example, taking a word such as ‘game’, its immediate equivalents are words
such as ‘contest’ and ‘sport’. If ‘contest’ is also referenced, we find that they are
associated with words like ‘match’. ‘Match’ is therefore considered to be two
places away from ‘game’, giving ‘match’ and ‘game’ a semantic distance of two.5
Whole documents can therefore be measured in terms of their semantic distance
from each other, depending upon the similarity of their content (Figure 3.5).

Embodying semantic space

In order to understand semantic spaces, we need to appreciate how they are
articulated, i.e. what are their intrinsic properties? What makes spatial location
meaningful? In the discussion of hypertexts and databases at the beginning of
this part, I isolated two conceptualizations of meaningful spatial structures as
they related to topological relationships in databases:

connectivity = meaningful relationships

containment = shared properties

To these, I wish to add a third type related to IR techniques which equate seman-
tic similarity with proximity:

closeness = similarity

While semantic spaces may seem far removed from physical spaces and objects
manifest in the physical world (because of the possibility for multidimensional-
ity), their conceptualization and, indeed, systems used to extract meaning from
them, are based around the methods we use to measure and map physical
space. Crucially, these methods conceptualize semantic space as consisting of
topologies of objects, containers and geometric relationships, where meaning is
articulated through degrees of connectivity, containment and distance.

Multidimensional semantic space

It is because of this spatial conception of semantic space that we can call an
information system multidimensional. Multidimensionality both articulates an
information system’s complexity and also frames it within an expectation of
spatial coherency. Multidimensionality implies that every information object
has a single instance and a discreet location that requires a space beyond 3D
to contain it. We can describe this space in geometric language following the
conceptions of semantic space as articulating meaning, through the three spatial
relationships described above.

105
The spaces of information

Connectivity
Multidimensionality, in terms of connectivity, is a product of the topological
restrictions that govern how places can be found to be adjacent to one another in
a 2D plane. The necessity for what Dieberger, for example, calls ‘magic features’
or ‘teleports’ in his discussion of hypertexts is due to the fact that navigation,
which tends to occur on 2D planes, restricts the topological relationships bet-
ween places to those not containing K3,3 or K5 topological geometry.

Containment
If an information object only exists as a single instance, it cannot sit in two
containers simultaneously, unless one container is held within another in a
hierarchical structure. Multidimensionality occurs when a single object sits in
two containers that are independent and must therefore exist in two different
dimensions.

Closeness
The notion of closeness implies a geometric limitation placed on objects that
have measurable distances between them. An object placed within a 2D or 3D
space eventually achieves geometric lock if its position is triangulated in relation
to other objects in the same system. This is illustrated in Figure 3.6, where a
number of objects are linked together by rigid connectors of a defined length in
a 2D space. The diagrams show the relationships defined as each new object
is introduced. Geometric lock is achieved when an object’s position has been
triangulated with relation to at least three other objects in the system. In the
illustration it is not possible to connect object (a) to object (d) because of the
existing relationships between (a), (b), (c) and (d). We can solve this problem
by projecting into 3D space, but a more complex set of link requirements soon
renders the geometry locked in 3D space as well. We must therefore seek the
addition of a fourth spatial dimension, and so on.

A definition of semantic space

Semantic space is more than information held ‘within a computer’. Through the
ways in which we conceptualize it, semantic space is considered as comprising
a set of information objects occupying discreet relative locations within a sys-
tem of interconnected topological relationships with other information objects.
Through their conceptualization, information objects can only reside in one loca-
tion; i.e. they are not considered to have separate instances of themselves and,
because of the need to assign them coherent spatial relationships, semantic
space is considered to be multidimensional. Semantic space isn’t available for us
to experience directly, but can be geometrically mapped and measured, where
meaning is articulated through the closeness, containment and connectivity of
information objects.

Part 4: Mapping – screen space

In describing semantic space it is difficult not to use graphical illustrations. In
discussing the spaces of information, it is virtually impossible not to attempt to
visualize these spaces as existing in terms that we understand and encounter

106
 The spaces of information

3.6 every day. No matter how complex and unimaginable semantic spaces are, they
Diagrams showing geometric
lock when four objects
only become useful once they are exposed though human-readable interfaces.
are arranged with fixed Semantic spaces are revealed, when they are, through the output of a computer
distances from one another and through changes in the computer’s output caused by the computer user’s
input. For example, the primary representation of a website is the visual display
of text on a computer screen where the user can change their view of the site
by clicking on a hyperlink and thus calling up a new page. While computers are
capable of receiving multiple inputs and presenting multiple outputs, our primary
means of interacting with them is through the mouse and keyboard and by see-
ing the results visually on screen. The type of interaction isn’t only defined by the
computer’s hardware input, but also by the relationship between the input and
the change in output and, to this end, I want to separate the visual rhetoric of the
screen’s display from the interaction with information, which I will return to in
the next part. This part will examine aspects of the visual display of information
through the concept of mapping, with a view to understanding how the visual
arrangement of information objects creates visual meaning.
I have shown, in the discussion of semantic space, that its concep-
tualization is bound by the three methods through which the distribution of
information objects is used to express meanings. When attempting to visualize
multidimensional semantic spaces, there is inevitably a process of dimension
reduction which must be undertaken. When using a web-search engine for
example, typing query terms into the search interface is essentially a process
of defining a limited set of coordinates in the search index and visualizing this
fragment of the index space as a list. The challenge of visualizing semantic

107
The spaces of information

spaces is to derive ways of capturing and presenting parts of it in two or

at best, three dimensional representations. Broadly this can be achieved in
two ways.

1 Through the development of systems which are able to traverse semantic

space so that the user does not have to. In practice this means creating
better IR algorithms that present ‘accurate’ results for our needs using a
process of search querying and retrieval. For example, the methodology
adopted by a search engine such as Google is to ensure that the most
relevant web pages relating to any given query are ranked highest on the
results page. Companies such as Google, who have become leaders in this
field, have done so because they are perceived to give unbiased, trustwor-
thy and, most of all ‘accurate’, results to given queries.
2 By aggregating semantic space i.e. to map it in such a way that, although
the represented dimensions might be limited, the essence of the underly-
ing information space is preserved. This method works on the basis that
humans are more effective at pattern recognition than computers, if infor-
mation can be shown in an appropriate visual format.

It is the second of these approaches that I will focus on here by examining the
process of mapping and diagrammatic representation. To achieve this, I will
divide the ways in which the visual space of the screen can articulate meaning
through spatial representations into two parts. I will show how visual space can
express meanings through the representation of relationships in either absolute
or relative space. By doing this, I want to show that the placement of objects
on the visual space of the screen leads to meaningful representations through a
computer user’s interpretation of diagrammatic relationships. As with semantic
space, I will show that computational screen visualizations of information express
meaning through a number of consistent geometric rules and thus our ability to
read meaningful relationships in screen based visualizations is derived from a
reduction in topological or topographical complexity.

The semantics of absolute space

Absolute spaces occur where the information objects have properties that can
be mapped onto either an existing space (i.e. a specific geographical location) or
quantitative dimensions, where coordinates assign values to positions in space.
Both types of space are sometimes called ‘principled spaces’ (Folz 1997: 1).
Principled spaces, both geographical and numerical, are the basis
for many of the key visualization systems that have become canons of InfoVis
research. For example, HomeFinder (Williamson and Shneiderman 1992), which
uses a map of Washington and the locations of homes for sale and Perspective
Wall (Robertson et al. 1993), where document objects (personal files) are placed
with relation to a timeline and file-type axes. In systems that use principled
spaces, there is a clear differentiation between objects and the spatial matrix in
which they are placed.

108
 The spaces of information

The semantics of relative space

As a contrast to absolute spaces, we can also isolate the semantic potential of
relative space. Relative spaces are defined by information objects themselves
and their relationships to one another, independent of the spaces in which they
are held. Relative spaces are a natural occurrence in our everyday experience.
For example, to say that the spade is in the bucket is to state the relative relation-
ship between the bucket and the spade. As has been shown, such embodied
relationships are fundamental to the way we interact in spatial domains, but the
vocabulary of such relationships extends beyond a binary ‘within or outside’
relationship. The knives are put in the drawer; the file is next to the telephone;
the cup is on the table; the cat is under the chair are all examples of relative
relationships. These relationships can be mapped accurately, if necessary, but
linguistic descriptions like ‘over’, ‘under’ and ‘next to’ are normally sufficient to
describe relative spatial relationships. It would be considered eccentric to say the
file is 2 cm to the northeast of the phone, but without this information it would
be impossible to provide an accurate absolute description of the position of the
file and the phone.
These relative relationships then acquire meaning, as was shown in
part one, by our conception of information objects in relationship to one another
through metaphor. Categories are containers, relationships are enclosures, simi-
larity is closeness, etc. However, realizing relative space visually is a non-trivial
task since there are no ways of naturally mapping them. Diagrammatically, the
more abstract the visualization, the more likely it is that we will read more infor-
mation into the diagram than is actually there. For example, Figure 3.6 ostensibly
shows the same visualization of six circles in two containers. Figure 3.6a shows
these objects as a bucket with marbles in the bottom. Figure 3.7b shows the
same visualization as a diagram where the container is represented as a circle.
In both cases, the round objects are clustered, yet in the bucket visualization
this clustering is more likely to be interpreted as random, corresponding to the
way in which the marbles may actually sit in the bottom of the bucket, whereas
in Figure 3.7b, the abstract diagrammatic nature of the clustering means that

3.7
Diagrams to show six objects
arranged in the bottom of a
bucket and in a diagram.

109
The spaces of information

we are tempted to interpret similarity as closeness and to imbue the clustering

with meaning.
To understand relative space, we need to understand that there is
no natural topographical mapping. As Wexelblat (1991: 260) points out, in rela-
tion to what he calls the quantum dimensions of space, spatial representations
that articulate relative relationships between objects are essentially embodied,
in that they relate to the world as it is experienced so that objects do not gain
meaning from their absolute locations but make themselves available for our
use by being brought together into groups; the knife is in the knife drawer, etc.
Of course, such objects exist within a topographical space, but their absolute
positions rarely have meaning; rather, in the knife drawer, it is their containment
and the proximity of other knives that articulates their place within a category of
cutting tools.
To say that a representation is essentially embodied, as in topologi-
cal representations of space, also carries with it an implication which I want to
develop in the next part. Specifically, where are we? Or, in other words, where
are our bodies in embodied representations? To understand this question it is
necessary to understand the space of interaction with a computer and also that
information space doesn’t become visible via a simple translation from semantic
space to a visual screen space but, rather, exists in the space between human
computer input and output, in the space, we might say, between clicks.

A definition of screen space

Screen space has been defined here in relation to semantic space and the pro-
cess of mapping common to many types of information visualization. In screen
space, information objects acquire meaning either through their position in an
absolute space, where spatial coordinates are ‘principled’ or through their relative
relationships with other objects as part of a system. Relative positions involve
understanding the meaning of an object’s position in relation to other objects; for
example, its proximity to other objects, its enclosure or its degree of connectivity.

Part 5: Exploring – interaction space

To understand the interaction between semantic space and the computer
user, we need to better understand the process of interaction and, importantly,
the user’s embodied relationship with information space. So far, in Part 3, I
have described the process of mapping semantic space through relationships
between objects in the visual language of diagrams, which express spaces as
either being absolute or relative.
What Part 3 has not taken into account is our embodied relation-
ship to the information being presented. Interacting with a personal computer
appears to be relatively straight forward in terms of our embodied relationship to
the artefact of the computer. We sit, immobile in front of the computer screen
where computer output is presented in a primarily visual display, pointed to with
a mouse cursor that mirrors hand movements on the mouse. But this is not the
whole story. In addition to the literal embodiment represented by our physical
interaction, there is also what might be described as our phenomenological
embodiment.

110
 The spaces of information

The process of mapping information to the visual display of the

screen implies a cognitive relationship between the computer user and the
information space, where the interactor separates him- or herself from the
information space through an intermediary device. The equivalent spatial condi-
tion can be found in the process of reading a map, or indeed any other visual
representation. The observer, when reading a map, is separated from the space
of presentation; they have an overview of a visual space and are not immersed
in it. In contrast, the space in which they are navigating allows them only a very
limited visual access. Things that are distant become indistinct and objects in
the space occlude a clear view. This physical space is cognitively pieced together
into a coherent whole through snapshots. In this part, I will propose that these
two spaces are also apparent in interacting with computers via screen-bound
PCs and argue that there is confusion between them in some discussions of
screen-based interaction with information.

Frames of spatial reference

Embodiment, as it relates to perceptions of space, is relevant to a problem which
has recurred in a number of different contexts in this book. On the one hand there
is a definition of space that suggests that space is absolute and realized through
measurable places and locations with accurately mappable coordinates within
the visual space of the screen. On the other hand, there is a definition of space
that is relative, consisting of places and associations in terms of a topological
space. The distinction between these two ways of describing space is present
throughout many philosophical enquiries into space and cognition. For example,
the philosopher of science, Henri Poincare, arguing against Descartes’s notion
of an objective mind, detached from the subjective body, by referring to the per-
ception of space and stating that ‘absolute space is a nonsense, and it is for us
to begin by referring space to a system of axes invariably bound to our body’.
Instead we should understand objects in space as position and state. Space is the
‘means that we represent to ourselves the movements which must take place
to reach an object’ (Poincare quoted in O'Keefe and Nadel 1978: 34).
O’Keefe and Nadel argue that perception is based on two compo-
nents, an egocentric spatial system which is both ‘represented in a relative
manner; that is, referenced to the organizm and built up through experience’
(O'Keefe and Nadel 1978: 60), and a prior unitary space which they call ‘local
space’ and which we might suggest is allocentric and ‘does not depend for its
existence on particular objects but which serves as a framework for relating
these objects to each other independent of the observer ….’(1978: 60).
The difference between these two proposed spaces of interaction
can be illustrated in relation to a question proposed by William James in 1907:

Some years ago, being with a camping party in the mountains, I

returned from a solitary ramble to find everyone engaged in a fero-
cious metaphysical dispute. The corpus of the dispute was a squirrel
– a live squirrel supposed to be clinging to one side of a tree-trunk;
while over against the tree’s opposite side a human being was imag-
ined to stand. This human witness tries to get sight of the squirrel by

111
The spaces of information

moving rapidly round the tree, but no matter how fast he goes, the
squirrel moves as fast in the opposite direction, and always keeps
the tree between himself and the man, so that never a glimpse of
him is caught. The resultant metaphysical problem now is this: Does
the man go round the squirrel or not? He goes round the tree, sure
enough, and the squirrel is on the tree; but does he go round the
squirrel?
(James quoted in Rohrer 2001: 50)

The answer to James’s question lies in the two alternative definitions of space
and the way it is perceived. If the space is considered to be perceived from
an allocentric frame of reference, in that it is independent of the bodies which
perceive it, then the man clearly does go round the squirrel:

If you mean passing from the north of him to the east, then to the
south, then to the west, and then to the north of him again, obvi-
ously the man does go round him, for he occupies these successive
positions.
(Rohrer 2001: 50)

However, if James is describing an egocentric version of the man’s spatial

perception, where his frame of reference is oriented towards the circling man’s
perception of his own position and orientation with relation to the squirrel:

if … you mean being first in front of him, then on the right of him,
then behind him, then on his left, and finally in front again, it is quite as
obvious that the man fails to go round him, for by the compensating
movements the squirrel makes, he keeps his belly turned towards
the man all the time, and his back turned away.
(Rohrer 2001: 50)

In the first explanation, a global perspective is being taken, in that both the man
and the squirrel are treated as objects with coordinates in a space with a global
orientation (in this case polar coordinates) (Figure 3.8a). This space can be judged
from a third point of view, independent of the man and the squirrel.
Conversely, the second explanation refers to an embodied concep-
tion of space. If the man only takes into account his position with relation to
the squirrel as a product of their orientation to one another, then their relative
positions are fixed since their orientation to one another remains the same
(Figure 3.8b). Regardless of whether the individual contemplating this problem
is the man, the squirrel or a third party, it is necessary to understand a position
with relation to the situation, to place oneself in relation to the bodies being dis-
cussed. An understanding of embodiment therefore requires a recognition that
a single situation can be viewed in different perspectives depending on one’s
orientation to what is being perceived.6

112
 The spaces of information

3.8
Diagrams to illustrate the
‘squirrel problem’ in which Egocentric space in interacting with information
the scene of a squirrel and Computer user interfaces tend to be conceived as mapping to an allocentric
man walking round a tree
can be either described frame of reference. The paradigm of the desktop, with the process of direct
in absolute terms (a) with manipulation, visualizes a space of objects over which the user has dominion
the man walking round the
through the grasping and moving of iconic representations of objects. Mapping
squirrel in terms of the polar
coordinates of the space information and the software used to visualize information, along with hardware
or relatively (b) with the and software that treat the computer screen as a surface, including a new gen-
man maintaining a relative
position to the squirrels body eration of graphical user interfaces that utilize touch on desktop, table-top and
at all times mobile devices, all present a space in which the user takes a god’s eye view of
an absolute space of interaction.

113
The spaces of information

As has been shown in the previous chapter, another dominant

schema for interaction is that of navigation, which is implicitly egocentric. The
most obvious examples often make recourse to the use of 3D user interfaces
such as those associated with simulation and video-game environments but I
also want to suggest that because the metaphor of navigation is so widespread it
exists in less obvious contexts through the paradigm of navigation in topological
systems, notably hyperlink traversal, as described at the end of Chapter 2. We
have already seen the emergence of limited topologies in such systems, where
their mode of interaction is described as a process of navigation, and the reason
for this becomes clearer in relation to a specific image schema.

Schematizing space
In Chapter 1, I introduced image schema as they assist in our intellectual
understanding of the world by binding, through metaphor, abstract concepts
to embodied actions. Fundamental image schemas are those that relate spatial
relationship to abstract objects and we have encountered them repeatedly in
this chapter: categories are containers, relationships are enclosures, etc. Image
schemas tend to be considered through metaphor in language, but they can also
be discovered in semantic and visual spaces, whether through the ‘similarity is
closeness’ metaphor employed in some IR techniques or the visual diagrammatic
mappings that articulate categorical relationships. To understand image schemas
in relation to interactive space, however, we must look at image schemas in a
different way.
The current list of image schemas can be divided into two groups by
separating, as Johnson does, schemas where bounded regions of space are act-
ing as containers for objects and where objects are placed ‘in containers (cups,
boxes, cans, bags, etc.)’, i.e. spaces that we conceive through allocentric frames
of reference and those spaces in which individuals themselves are contained
when ‘We move in and out of rooms, clothes, vehicles and numerous kinds
of bounded spaces’ (Johnson 1990: 21) that we conceive through egocentric
frames of reference.
The distinction in our perception of these two spatial situations has
been usefully observed by Tversky in her discussion of spatial representation in
language, where she separates the ‘space around the body’, which is ‘the space
within reach of eye or hand, a space we conceive of in three dimensions’ and
the ‘space of navigation’ which is ‘the space too large to be seen in a glance, the
space we mentally piece together from exploration, from maps, from descrip-
tions …’ (Tversky 2001: 12.2).
This fragmented perception of the space of navigation, because the
space is ‘too large to be seen in a glance’, is closely analogous to the situation
of interacting with topological spaces such as hypertexts. In fact, the notion of
navigation would be meaningless unless we were, to some extent, consider-
ing ourselves to be immersed within the spatial system created by hypertexts.
Meaning in such topological systems emerges, not through a visual association,
but by an act of translation, moving from one place to another. The fact that
the user has not physically moved doesn’t matter because, through the screen
transition, a change in viewpoint has occurred and this change in viewpoint is

114
 The spaces of information

associated with a change in location if a hypertext page is considered to be a

discreet place. The navigational schema, therefore, provides a means by which
a particular conception of space in the real world can act as a framework for an
understanding of topological spaces that appear to have little to do with the visual
world with which we are familiar.

A definition of interaction space

Interaction space, as it is defined here, refers to the space between mouse
clicks. Interaction space, in the case of hypertexts is revealed through the
process of navigation where a web page is conceptualized as a place and the
traversal of links is conceptualized as a change in view point.

Part 6: Navigating – architectonic space

In this chapter, I have separated the space of information (semantic space), the
space of visual presentation (screen space) and the space of interaction as it
relates to the relatively closed world of interaction with computational devices
(notably personal computers). In so doing, I have covered a large number of dif-
ferent types of presentation and modes of display and interaction.
The first striking characteristic of this discussion of the three spaces
of information is that a limited conception of space pervades each description.
Even in complex information systems, it seems that we need to put the human
body in the system and to ask where we are in relation to the information spaces
being presented. To conceive of semantic space, we use the same notions of
geometry used to measure our material world, we use metaphors of proximity,
connectivity and containment in the visual display of semantic spaces on the
screen and we attempt to articulate interaction with large information structures
as a process of navigation, akin to a real process of journeying.
By distinguishing between notions of absolute and relative space and
of allocentric and egocentric frames of reference, I have isolated the concept of
the navigational schema, which maps an embodied understanding of space to a
conception of topological systems such as hypertexts.
In the terms described above, architecture is concerned with an ego-
centric conception of space. Architecture differs from other design disciplines,
such as product design, in that buildings contain people and are experienced in
perceptual fragments that are built up to create a whole through the process of
physical movement. This contrasts with our use of objects such as household
appliances, where we adopt allocentric frames of reference where the space
of interaction can be viewed at once. We tend to consider personal computers,
in the most part, to be allocentric objects as we sit with an overview of the
screen unless we are literally immersed through devices such as head-mounted
displays. Yet, I have suggested here that some non-immersive computer inter-
actions are also egocentric and that computer users, at some level, ‘project’
themselves into the space of interaction.
I have shown how, in framing the process of navigation within hyper-
text documents, both the organization of personal information on the computer’s
desktop GUI and the visualization of information, have used the metaphor of
architecture and even adopted architectural concepts such as Lynch’s imagability.

115
The spaces of information

The metaphors themselves are much less interesting than their origin and the
implications of their use. The concept of navigation I suggest is more than a weak
metaphor and the navigational schema as described above does not necessarily
lead to the depiction of digital information as virtual buildings and cities. Rather,
the navigational schema helps us understand the constraints put on our digital
representations of information.
It takes a leap of faith to equate, for example, the process of click-
ing between hypertext links with the traversal of space and, consequently,
the process of link traversal seems far removed from navigating real spaces.
However, I will claim that the emergence of architectural and other spatial
metaphors to articulate information is not accidental and that the navigational
schema emerges because even the least visually spatial (if visually spatial is
considered to mean a 2D- or 3D-projected visual space) topological information
systems are perceived as a type of space. Furthermore, I will suggest that the
act of clicking on a link and changing a web page can be considered to be a tra-
versal of space because the nearest cognitive model that computer users have
at their disposal to understand an abstract computational process is that of their
topological model of spatial navigation. I will suggest that architectonic space,
once transposed from the material world, results from a relationship between
the computer’s interaction space, screen space and, crucially, the mental model
that interactors apply to understand the nature of their interactions. The basis
for this is to understand the interactor’s so-called cognitive map as a topological
phenomena.

Cognitive mapping
Cognitive mapping has scientifically grounded behavioural origins in the work of
Edward Tolman and his article entitled ‘Cognitive Maps in Rats and Men’ (1948).
In this article, Tolman published the results of a number of, now iconic, rats-in-
maze experiments in which the rats were given repeated exposure to a simple
maze, at the end of which was placed a food reward. Initially the rats used trial
and error, repeatedly taking wrong routes and backtracking until they eventually
reached the reward. It became clear, however, that after repeated exposure to
the maze, the rats began to learn the optimum route and would navigate without
error directly to the reward, no longer simply reacting to the stimuli of the maze,
but planning their route based on a prior knowledge of the spatial organization
of the maze. This process of environmental learning, which is prevalent in rats,
Tolman hypothesized, must also be present in human beings in a form of knowl-
edge that he defined as an ‘observer-independent representation of the external
world …’ (Tolman 1948: 429). In recognizing the existence of a cognitive map,
Tolman separated the egocentric model of spatial perception based upon direct
and synchronous perception from an allocentric memory of space independent
of direct synchronous perception and encoded abstractly in memory.
Cognitive mapping theory has grown, since its origins in the 1940s, to
become a core element of spatial cognition theories and it is now subject to its
own conferences and a number of highly influential collected works.7 In recent
years, cognitive mapping research has developed beyond the limits of cognitive
psychology to become a multidisciplinary study encompassing architecture,

116
 The spaces of information

urban planning, geography and neuroscience. Our ability to navigate success-

fully through environments, remembering routes and orienting ourselves in the
world, can be seen as one of our most sophisticated cognitive abilities and the
utilization of such abilities to new ends is a key focus of recent cognitive mapping
research.

Components of the cognitive map

The growing research into cognitive mapping, encouraged in part by Lynch, was
most evident in the edited book Image and Environment (Downs and Stea 1974).
This publication was cross-disciplinary and dealt with multiple topics, from map
interpretation to the quality of our mental environment, expanding upon, and in
some cases superseding, Lynch’s work and developing methodologies that are
now prevalent in the discipline of behavioural geography. By the 1990s, new
research efforts had begun to emerge, connected with fresh funding opportun-
ities and a renewed interest in multidisciplinary collaboration. Much of this later
work in cognitive-mapping research has set out to prove or disprove the existence
of an actual map-like representation in the brain. Where early researchers such
as Tolman, and later O'Keefe and Nadel, supported the notion of an independ-
ent, map-like representation of familiar environments, more recent research has
tended to use cognitive mapping as a metaphor for the structural processes that
govern navigational behaviour in space. Rather than being a cognitive map, mental
representations are divided into knowledge structures which are acquired, as an
environment is navigated and explored, through the following stages.

1 Landmark learning: This is initiated as an individual is introduced to an

environment and uses landmarks as a means of orientation with relation
to key features of the environment, such as distinctive land features or tall
buildings.
2 Route learning: This is acquired after more repeated exposure to an envi-
ronment and consists of knowledge of particular sequences of journeys
between certain departure and destination points, such as the route from
work to home. Often knowledge of urban environments, in particular, is
based upon this knowledge of key routes without detailed knowledge of
the area in between.
3 Configurational learning: This is acquired through long-term familiarity
with one’s environment and knowledge of the environment from external
sources such as maps or aerial photographs. Configurational knowledge
not only encodes landmarks and routes, but also integrates them into an
accurate description of the environment, allowing parts of the environment
which have not necessarily been explored before to be navigated with
confidence.

Knowledge of the environment is therefore built in stages, in some cases

starting with landmark knowledge (Siegel and White 1975), or environmental
memory is knitted together through the traversal of different routes. Other
related approaches include that of Cornell and Hay (1984), who develop the idea
of recognition of specific vistas or routes, suggesting the use of an episodic

117
The spaces of information

memory of sequences of events rather than a memory of overview, map-like

connections. The difference between the development of environmental knowl-
edge and the cognitive map is that, where the term ‘map’ implies a flat, single,
accurate image of an environment, analogous to a real map, environmental
memory is developmental rather than instantaneous, gained through the integra-
tion of multiple sources, both through direct experience of the environment and
through external representations (maps and diagrams for example). Memory of
environment is recalled, not as a topographically accurate whole, but as topologi-
cal knowledge representations akin, Kupier et al. (2003) suggest, to a skeleton of
relationships.

Topology and ‘ceptions‘ of the cognitive collage

Recent research into spatial cognition and navigation has brought into question
the existence of the cognitive map as a single and harmonious encoding of
environmental memory, and constructivist psychologists such as Barbara Tversky
have begun to develop cognitive mapping theory further. In particular, Tversky
has suggested an alternative to the cognitive map with what she refers to as a
‘cognitive collage’ (Tversky 2001).
In their development of the concept of the ‘cognitive collage’, con-
structivist psychologists have been attempting to develop holistic models of
cognitive systems, which take into account multi-source input and different types
of memory, and to develop an embodied account of the way in which environ-
ments are perceived and conceived. Humans are not simply passive receivers of
sensory input but are subject to conceptual schemas through which perception
is constructed. In Tversky’s paper to The Space Syntax Conference in 2001, ‘The
Structure of Mental Spaces’, she outlines a pluralist approach to the concept of
Cognitive mapping:

The mental representations that we form of space from these real

and imagined interactions differ from the external representations of
spaces of geometry or of physics or of maps … In human concep-
tions of space, the things in space are basic, and the qualitative spatial
relations among them form a scaffolding.
(Tversky 2001: 12.1)

In order, she argues, for us to operate effectively in space, we must consider

space as being, in part, perceived directly through embodied experience, and in
part conceived through a ‘scaffolding’ governed by prior knowledge, independ-
ent of direct experience. The terms ‘perception’ and ‘conception’ are conflated
by Tversy, following Talmy’s (1983) discussion of space and language, to the
single word ‘ception’:

To understand how ‘ception’ schematises space is to understand

that perception is not just bottom-up, determined by stimulus input
alone, but is in addition top-down, conditioned by what is already in
the mind.
(Tversky 2001: 158)

118
 The spaces of information

Ceptions are revealed to researchers of the cognitive map through consistent

errors in spatial memory. If memory of spatial locations was a simple matter
of direct perception, storage and recall, then we would expect errors to be
minimal and for them to be different for different people. However, where there
are consistent errors across different individuals, this may indicate a common
encoding of memory, based either on a shared perceptual or a shared conceptual
error (Tversky and Lee 1998). In particular, Tversky suggests that language is a
powerful system for encoding spatial memory. However, because language is
better able to articulate relative positions than specific global locations, memory
for one’s environment tends to encode relative topological properties of the
environment but not it’s global metric properties:

For example a term like ‘across’ can apply to a set of spatial con-
figurations that do not depend on exact metric properties such as
shape size and distance. Use of across depends on global proper-
ties and configuration of the thing doing the crossing and the thing
crossed.
(Tversky and Lee 1998: 158)

Through Talmy’s and Tversky’s concept of ception, topological representations of

space are an important dimension of spatial perception. `topological space in their
terms differs from topological space which consists of ‘size, length, distance,
angle or contour, as well as more substantive qualities like texture, materiality or
identity’ (Talmy 1983: 234). `rather topological space is described through proper-
ties such as ‘their type of structural conformation, degree of subdivision, number
of relevant dimensions, boundary conditions, and symmetry vs, distinguishability
of parts’ (Talmy 1983: 234).
The cognitive map, or in Tversky’s terms, the cognitive collage, is
encoded through a memory for topological relationships, which, though aided
by topographical representations such as maps, are composed of associations
between places as a network.

Topological cognition and the architectonic system

At the beginning of this chapter, I showed how, in the context of information
visualization and hypertext research, Kevin Lynch had been acquired to give cre-
dence to the use of architectural metaphors. Referencing Lynch in the context
of information visualization is, however, superficial. For Dieberger, for example,
the city metaphor is a way of taming semantic space. It does not matter whether
we call a hypertext page a room or a building, the result is to limit the topology
of semantic space and, therefore, to characterize a link between ‘places’ as an
adjacency in space. In the case of information visualizations, the use of Lynch’s
elements is a way of enriching the visual space of the screen.
A problem arises in the use of architectural metaphors because of a
failure to understand the core reason behind the metaphor of navigation. When
we discuss ‘information architecture’, it is unhelpful to take the notion too liter-
ally and to believe that a website, for example, is equivalent to a building or city.
However, I suggest that many systems for the representation are architectonic

119
The spaces of information

systems. These architectonic systems are not, and do not, need to be articulated
as buildings (real or metaphorical) yet exist because they are realized as a space
which is viewed as a sequence or ‘a set of place-making orderings whereby
individuals construct and communicate the conceptual world through the use
of palpable distinctions in formation … addressed to the visual channel to be
decoded spatio-kinetically over time’ (Preziosi 1979: 4). If a fundamental com-
ponent of the cognitive map is the topological representation of space, then in
finding systems that are abstractly topological it makes sense that we would
apply this component of our spatial understanding to interact with them.

Part 7: Conclusion
In seeking to answer the question ‘Where is the architecture in information
architecture?’, it is often difficult to move beyond the use of the analogy or meta-
phor of architectural space, whether through systems that present information
through ‘virtual reality’ constructs of buildings or cities, through the ubiquitous
desktop and office used for the arrangement of applications and file system of
a PC or through the patterning of hypertext. Although such systems present the
façade of architectural space, their presentations seem to have little to do with
architecture as it is experienced and considerable doubt has been placed on the
validity of such metaphors in the articulation of digital information.
Architecture, as traditionally practiced, is involved in the articulation of
space and it therefore follows that, in order to discover what information architec-
ture is, the nature of the space that is being articulated must first be understood.
We have seen that through the tradition of mnemonic architecture
there has been a long-standing relationship between architecture and the repre-
sentation of abstract meaning. In particular, mnemonic architectures, both real
and imagined, relate the organization of physical objects and spaces and the
way in which these spaces are experienced, to the communication of abstract
meanings and associations. This is a process that involves the uniting of topos
(place) to topic and relates the structure of ideas to the structure of space. This
tradition can be seen alongside, but distinct from, representations of knowledge
through textual, oral and purely visual means (e.g. diagrams and illustrations)
and this tradition has survived in modern storehouses of knowledge such as
museums and libraries.
Before the digital encoding of information, it was unnecessary to
discuss information space since information necessarily resided in representa-
tions that were grounded in, and presented through, material means. Since the
emergence of digital encoding, however, information has been separated from its
material means and thus the representation of information within the multidimen-
sional space of the computer’s memory and its visual and spatial representation
through the computer’s screen has divided the ‘space of information’. This divi-
sion can be characterized through different ways of interacting with information
systems where semantic space is measured and given geometric parameters.
The visual space of the screen maps semantic space and interaction space, which
is the ‘space between mouse clicks’, is understood through a conceptualization
about the dialogue between the user’s input and computer’s output.
In the context of representations that present information through 2D

120
 The spaces of information

or 3D environments, which are highly metaphorical, understanding the process of

shifting one’s viewpoint of the system as being akin to a process of navigation is
relatively straightforward since a direct analogy can be drawn between navigat-
ing in the real world and the simulation of navigation on the screen. However, in
terms of the representation of topological semantic spaces such as hypertexts,
the process of selecting a link and causing a screen transition is less obviously a
process of navigation. To overcome this, it is necessary to consider a topological
semantic space, such as a hypertext, as being represented cognitively to the
user through an embodied schema.
Embodied schemas are mental frameworks that allow individuals to
use their basic experience and cognition of the world as a means of conceptu-
alizing more abstract subjects. Such schemas are revealed through the use of
certain fundamental metaphors. In particular, there are a range of metaphors that
connect spatial experience to the formulation of abstract concepts (similarity is
closeness, etc). The use of navigational metaphors, it is suggested, is indicative
of such an embodied schema, where the user of the hypertext system conceptu-
alizes the process of clicking on links, with their own physical traversal of space.
However, as with all schemas, the navigational schema inevitably restricts the
subject it conceptualizes. This restriction can be demonstrated by associating
the abstract link with an adjacency in space and applying rules of topological
geometry that constrains the system’s link structure to 2D planar space.
The essence of the architectonic system, applied to digital informa-
tion systems, exists in this mode of interacting with a computer, where the
user associates the pattern of outputs generated by their input with a process
of travelling. This navigational schema is less about architecture as a metaphor
and more about the architectonic system as a way of considering spatial pattern-
making as a means of communication.

121
Chapter 4

Reality becomes
display
The universe is fundamentally composed of data, understood as
dedomena, patterns or fields of differences, instead of matter or energy,
with material objects as a complex secondary manifestation.
(Floridi 2010: 70)

Part 1: After information architecture

I began in Chapter 1 with a description of the ‘architectonic system’ as being akin
to, but distinct from, the linguistic system. The architectonic system developed
our capacity to categorize the world through the organization of social groups.
These social patterns became the basis for the organization of abstract concepts
and ideas. The idea of the architectonic system leads us to an understanding of
information as arising from combinatorial patterns, where physical constraints,
in terms of what can be next to what given a consistent 2D or 3D space, gener-
ate meaningful relationships between objects, concepts and categories. The
challenge posed by computational representations of information is that the
constraint on the placement of physical objects no longer applies to the organiza-
tion of virtual objects. Despite this, computational information has been framed
by spatial metaphors. I showed in Chapter 3 that there are, broadly, three types
of information space that we encounter when interacting with computational
information systems. I described these spaces as semantic space, visual space
and interaction space. I also developed a theory of the ‘navigational schema’
to provide a context in which, I proposed, the architectonic system proliferates
through the network topologies of systems such as the WWW. The tendency
to frame our interactions with digital information systems through the metaphor
of space has been characterized in previous chapters as a battle between the
potential of computational systems and the human requirements in interacting
with them. The earliest computational systems (in common with their descend-
ents) worked by hiding semantic content as a mechanical process rather than
as a static configuration of relationships. This meant that a mechanical system
needed to be activated before a meaningful relationship between information
objects could be read. The encoding of meaning through mechanical processes
results in the semantics of the system becoming hidden and only visible in the
process of interaction and the revealing of combinatorial patterns. Imagine,
for example, using the Lullian wheel, which was introduced in Chapter 2. The
Reality becomes display

machine, as I described it consisted of three concentric wheels with the letters A,

B and C inscribed along their rims. To visualize all possible relationships between
each of the letters would involve changing the state of the machine’s configura-
tion through many different turns. These relationships exist through the potential
of the system but not persistently. The system can know what the interactor
might not. This ‘knowing’ is not intelligence, however. There is no intention
behind a system’s knowing, but rather a necessity for the interactor to enter
into dialogue with the machine. This mechanical knowing might be described
as a shift in agency from users to the system with which they are interacting.
In this final chapter, I will look more closely at this idea of agency from
the starting point of the migration of computation away from the virtual world of
the screen and to systems which are distributed in real spatial contexts. Having
developed an argument in earlier chapters that focused on virtual information
objects, this chapter will discuss the rise of the computing paradigms described,
variously, as ubiquitous and pervasive computing.

Hypothesis revisited
In this final chapter I also wish to return to the hypothesis as stated in the
Introduction:

The spatial patterns that we make in our environment are a primary

means of human communication. We use space to organize people
and objects and those physical organizations are the basis of much
of our conceptual thought. Consequently, despite the potential cre-
ated by digital technology to transcend the material organization of
information, patterns that relate to the organization of architectural
space continue to have a role in the design of digital information
systems.

In Chapters 1, 2 and 3 I built an argument that our conception of space frames our
interaction with various different computational systems and, furthermore, that
a spatial conception of information necessarily constrains our representations
of it. In Chapter 4 I will invert this argument and I will suggest that computation
has the potential to constrain how we interact with space (Figure 4.1). To make
this argument, I am going to return to the origins of the architectonic system,
which I proposed in Chapter 1. My argument will be based upon a duality inherent
in the architectonic system, where space shapes categories and, at the same
time, categories shape space. In Chapter 1, I suggested that the emergence
of categories as a way of organizing abstract concepts started with the spatial
distribution of social groups and the association of those social groups with
more abstract identities. Our capacity to categorize objects, even those with no
physical manifestation, is thus shaped by mental schemas that literally objectify
information as discreet material objects organized into patterns. The architectonic
system emerged in parallel to the linguistic system so that, as tools became
associated with actions and ultimately linguistic gestures, places become asso-
ciated with categories. Architecture articulates these spatial differentiations but
is also, to some extent, differentiated by them. I suggested in the discussion

124
 Reality becomes display

4.1
While previous chapters
have discussed how
categories have been shaped
by our articulation of space,
this final chapter will look
at how computational
categories may, conversely,
shape our experience and
actions in space

of function, form and space in the Introduction, that a relatively neutral space
such as a sports hall can be transformed by changing its name (to a court room
for example). Architecture is made, therefore, through the combination of the
functions that the division of spaces articulates and the way in which different
spaces are named.
While this book has so far focused on the way in which architec-
ture articulates categories, in order to understand the new context in which
computation is embedded into the real places we need to look at how com-
putational categories shape space. Furthermore, we need to understand the
strength of connection between spaces and their functions, not at the abstract
conceptual level we have pursued elsewhere in this book, but in terms of the
place of our physical actions. To do this requires returning to the origins of the

125
Reality becomes display

architectonic system as a means of communication and digging deeper into its

complexity. So far in this book, notions of space and place have been treated at
an abstract schematic level. We have conceived space through its representa-
tions and I have emphasized types of representation that describe spaces as
topological networks of connected places. While I have argued that this way
of representing space relates to one way in which space is experienced and
remembered, it is only one part of a rich spatial cognitive collage. Such topo-
logical conceptions of space are useful in the context of information systems
such as hypertexts which are, by their nature, reductive compared to our real
experiences. When, however, computational representation systems become
part of our real spatial experience, we need to have a much greater understand-
ing of the rich nature of spatial experience and a more sophisticated definition
of place.

The evolution of the web

Before starting on a discussion of ubiquitous computing, it is worth tracing the
development of the web from the point at which it was described in Chapters 2
and 3, from an authored system of topological relationships, to one that is
emergent and, to some extent, automatically constructed. Understanding the
development of web technologies is an important step in understanding the
most recent paradigms in ubiquitous and pervasive computing because it gives
a background to the shift in emphasis from planned to emergent information
systems and from human agency to computational agency.
While the profession of information architecture remains, new web
technologies have altered the focus of the information architect’s work from one
of supporting the creation of appropriate categories through which to structure
a process of web ‘navigation’, to one of designing infrastructures in which new
types of information space can grow. These changes have been brought about,
broadly, by two developments in technology.
The first is the rise of ‘search’ as a paradigm for web-based interac-
tion, exemplified by the emergence of Google as the most important web portal
on the WWW. The obvious change brought about by this paradigm has been that
finding information no longer needs to simply be about finding the right place
to go to and then logically following a pathway of links to the page of interest
but rather, through the input of query terms, obtaining a ranked list of links that
will drop us directly into the right location. In addition the effect of interacting
with web-based information using search engines changes the link topology of
the web and, indeed, the reasons for having a link structure in the first place.
PageRank, the primary heuristic for delivering search results in a search engine
such as Google, sees a link, not as a means of navigation, but of validation. Web
pages that have a large number of inbound links (i.e. links from other web pages)
attached to them are considered to have a higher status, given that inbound links
represent an external validation of their relevance in context. Web developers,
mindful of this fact, will attempt, sometimes maliciously, to increase the number
of inbound links to a particular page, not as an aid to navigation but to encour-
age high ranking and, therefore, large amounts of traffic through to the page via
search. In addition, when inserting a new query into a search engine, the results

126
 Reality becomes display

of that search are cached in order that the search result can be quickly retrieved
if someone wants to perform the same search again. This caching essentially
creates new links with any new query connecting, through a search result page,
a series of pages which may never have been linked before.
The second evolution of the web has been loosely described as
Web 2.0 and refers to applications which simplify the publishing of user-generated
content. Web 2.0 applications simplify the creation and distribution of new web
content by hiding the complexities of both code and server in the authoring
process. Social networking applications such as Facebook and MySpace allow
for a personal identity to be created on the web. Sites such as YouTube and
Flickr allow for personal expressions of creativity or for sharing other people’s
content. Sites such as Wikipedia empower users to socially construct knowl-
edge. Much has been, and continues to be, written on the causes and effects
of Web 2.0,1 however, for the purpose of this book, it is useful simply to note
that the information spaces that appear from these new applications are emer-
gent and not designed. Flickr, for example, allows users to tag images with key
terms, which, in turn, create what are known as ‘folksonomies’. These key terms
can be used in the context of a search on the site or as a way of automatically
generating links between instances of similar content. The idea of a hypertext as
a page connected to other pages is transcended by these new types of linking
structure.

The challenge to information architecture

These new paradigms of web-based interaction undermine the notions of
information architecture. Both Web 2.0 and the rise of search are products of
semantic space. Both promote web behaviour that is associational rather than
structural and, therefore, emergent rather than planned, and Web 2.0 sites tend
to rely much less on navigation than on search as an interaction strategy. Joshua
Porter is a notable voice in a clamour for an alternative (to architecture) framing
of the process of constructing web-based information:

Whereas ‘architecture’ started off in the physical world, we now have

to imagine (after merely placing ‘information’ in front of it) what it
means in the conceptual world. The once solid word “architecture”
is now unclear.2

The obituaries for information architecture may be premature, but the fact that
former notable information architects such as Porter are questioning the term
‘architecture’, is a clear indication of a shift in the way web-based information
is perceived. Through search, it is possible to move away from navigation as
a mode of interaction. Instead, an agent takes the load of traversing semantic
space and returns a highly simplified cross section of it in the form of a list in
the visual space of the screen. An analogy would be to consider the difference
between using a library in which all the books are on open shelves and a library
in which the books are retrieved by a librarian who journeys into the hidden
library stacks. In the second case the work of finding the book is largely taken
by the librarian. However, this situation is more complex in using web search

127
Reality becomes display

engines because, depending on the nature of the query, we might not be look-
ing for a known piece of information (i.e. a particular book) but rather instructing
the search system in a way which would be the equivalent of instructing the
librarian to find us books on a general topic area without further specification.
The decision must then be taken by the agent on what is considered relevant
to that topic and we must trust the agent to get the answer right or, at least,
retrieve the most relevant answers. In either case, as users of the library system,
the space of the reading room3 may be of interest to us but the semantic space
of the book stacks is something we may never see. The librarian becomes an
agent of immense power and provides the only way of accessing an otherwise
hidden semantic space.

Two modes of interaction and the distributed internet

In the previous chapter, I made a broad distinction between two embodied
modes of interacting with computer-based information. These two interactional
paradigms were based on two embodied frames of reference.
The first conception, illustrated by the direct manipulation paradigm
of the personal computer’s graphical user interface, places the interactor outside
the system. This leads to conceptions of space in which the interactors imagine
the information objects they are manipulating to be in front of them (e.g. on a
desk) and for those objects to be grasped and manipulated as if they were close
at hand.
The second interaction paradigm is where the users project them-
selves into an environment that cannot be seen all at once from a single point of
view. The users conceive themselves to be within the system and, rather than
information objects to be manipulated, the information system is understood as
a network of places that can be navigated. I called this conception the naviga-
tional schema.
Despite the rise of Web 2.0 and search, modern technology develop-
ments have, in practice, made these two modes of interaction more prevalent
and vivid in an ever larger number of computer interfaces.
For example, the process of direct manipulation within a PC GUI is
actually performed through the intermediary devise of the mouse. Multi-touch
interfaces, common in devices such is the iPhone, iPad and Microsoft Surface,
essentially cut out the middle device altogether, allowing direct manipulation
to be literally embodied. Touch-based technologies have, in turn, led to a new
range of gestural manipulations in interacting with screen space. In addition,
as hardware is reduced in size, hand-held devices have become loaded with
context-aware components such as accelerometers, which are able to track
the pitch and roll of a device and turn this into a means of interacting with their
software interfaces.
The second mode of interaction, navigation, has also become more
explicit. Just as the notion of cyberspace was revealed to be a catch-all term
for a number of discreet paradigms in computing, the WWW, as it has been
understood as a unified environment accessed through the window of a single
software interface (the web browser), has also become more heterogeneous.
The internet, as a network of hardware nodes, is enhanced by an ‘internet

128
 Reality becomes display

of things’ in which any object can be computationally enabled with sensors

that monitor and report on their environmental conditions and are capable of
reacting both automatically and in response to other inputs. The days of the
deskbound personal computer have been, if not altogether superseded, then at
least subsumed within a whole range of computational devices with networked
capabilities. The PC increasingly looks like the Swiss army knife of computational
devices in the shadow of specialized objects and systems which make up the
world of ubiquitous computing and, in turn, make computation part of the mate-
rial reality of the physical world. In the context of computing, which is based in
the physical rather than virtual reality, we no longer need to talk of a navigational
schema since, by becoming embedded in the material world, computational
systems can be literally navigated. As we move through physical space, we may
also be moving through a computational interface.
The architectural-design research community, in addition to the digital
arts, has responded to the opportunities offered by new types of visual display
and the possibilities of networked infrastructure and embedded computation,
with a new ecosystem of architectures and architectural fragments that flash,
beep, fold, undulate, are capable of dynamic change and communicate in a
cacophony of electronic noise. Digital architects no longer promise cyberspace,
but rather buildings which are ‘intelligent’ or ‘smart’, which ‘understand the
user’s needs’ and which are reactive, responsive or reflexive.4 Through such
experiments, buildings themselves become computational devices, addressed,
in the same way that architecture is always addressed, through physical journeys
through space.

A space of flows
In the Introduction, I cited Christopher Alexander’s conception of the built envi-
ronment as structured and connected places. Through this understanding an
environment becomes atomic (Alexander and Poyner 1970: 308) and realized
through a design process that involves understanding the ‘large collection of
many small systems … [which] goes to make up a large and complex system ….’
(1965: 58).
Alexander divided these systems in terms of their propensity for
dynamic change. Traditionally, the built environment constitutes a fixed and
relatively unchanging framework for these dynamic systems, which include what
Hiller refers to as the ‘movement economy’ (Hillier 1996: 152) of people, goods,
traffic and information. Digital infrastructure, however, changes this relationship.
We are, suggests Castells, living in a space of flows ‘centred around information
technologies’ which have begun to ‘reshape, at accelerated pace, the material
basis of society’ (Castells 2000: 1).
The dynamic parts of the built environment have always been
anticipated in architectural design but are not explicitly designed. To design
the dynamic parts of an architectural system would be akin to behavioural
control and, instead, architects anticipate dynamic activity and, to an extent,
programme it by designating particular places with specific functions. The notion
of programming takes on a particular resonance in the space of flows, where
programmed software is as much a designed component as physical hardware.

129
Reality becomes display

Software-enabled architectures raise the prospect of a dynamic sort of infrastruc-

ture encoded by software, but over which the architect or urban designer must
have some design responsibility.

The future of the architectonic system

In the previous three chapters of this book, I have demonstrated how our per-
ception of digital information is framed by a conception of information objects
organized in spatially coherent places. While this story has largely been played
out on computer screens, the migration of bits to atoms seems, at first, to
validate concepts such as the navigational schema, which are now acted out in
real environments. The story is not so simple, however, and in this chapter I will
show that pervasive computing, in connection with an emergent semantic space,
which is constructed as much by computers as by people, offers the greatest
challenge to the architectonic system as it is now understood.
In this chapter, I will examine technologies that allow objects to have
identities outside their context of use or their physical location. I will argue that, in
the age of ubiquitous computing, the integration of semantic space into physical
architectonic space comes with the challenge of agency. Simply put, to filter and
simplify semantic space to make it manifest in real spaces involves information
being sorted by computational rather than human agents.
Whereas the previous chapters have been essentially historical, I
will focus, in Chapter 4, on current and future technologies. However, my aim
is not on futurism but on a coherent theoretical understanding of the human
relationship with certain types of technology in the context of current trends in
the evolution of computation.
Part 2 of this chapter will introduce the world of ubiquitous computing
in relation to Weiser’s prediction of ubiquitous computing described in his paper
‘The computer for the 21st century’ (1991). Ubiquitous computing gives rise to
the notion that any object can be augmented by computation and the idea that
a physical object can exist in both a physical and a virtual location through the
possibilities opened up by digital tagging. It will be shown that, whereas the
context of physical places and objects is discursive, when an object occupies
both a physical and a virtual location, its identity may be partly hidden and only
revealed through a computational filtering process.
In order to illustrate the problem of agency when interacting with
digital information, Part 3 will reflect on a debate between two computer sci-
entists at different ends of the human–computer interaction spectrum, on the
relative merits of direct manipulation versus software agents. I will show that, in
order to become useful in the messy, ‘in the wild’ contexts of the physical world,
computational information must be understood in the context of activities that
are encoded by a relationship between place and action.
Part 4 will look more closely at the issue of agency through the idea
of smart spaces, where the notion of programming occurs in the discussion of
both the ‘hardware’ and ‘software’ of architectural space. It will be shown that
the notion of programming, in software terms, constrains how we understand
space and place, using Paul Dourish’s distinction between tactical and strategic
uses of space and examples of ‘smart space’ installations.

130
 Reality becomes display

Part 5 will conclude with an account of how, ultimately, architectonic

space may be reshaped by computation that provides contexts through the
electronic coding of place. This section will close by citing Crang and Graham’s
(2007) description of ‘augmented space’ and ‘enacted space’ and the notion of
‘transducing space’ as a way of mapping the relationship between place and
agency.
This chapter can only be a preparatory sketch of a new world of
computation and its relationship with the architectonic system. It will not be an
exhaustive account but, rather, will conclude by attempting to bring full circle
the argument developed throughout this book, by tracing the potential impact
of computation on our experience of the built environment using the same
framework that I have used in tracing the impact of our perception of the built
environment on computational systems.

Part 2: Non-discursive formations

The danger of making long-term predictions about trends in technology is
illustrated by the myth of cyberspace and the fact that, in 2011, head-mounted
displays are not commonplace and the web is not experienced as a virtual reality
world. In this light, a set of predictions made by Mark Weiser and published in
1991 in his paper ‘The computer for the 21st century’, should perhaps be treated
with caution. However, Weiser’s claims are taken seriously, not only because, as
a researcher at the fabled Xerox Park research labs (which developed the first
commercially available graphical user interface), his ideas come with pedigree,
but also because 20 years of computational technology development have borne
out the accuracy of the trajectory of Weiser’s ideas.
Weiser’s claims are now synonymous with the computing domains
variously known as ubiquitous computing and pervasive computing and, particu-
larly in the popular press, are characterized by collective technologies such as
‘the internet of things’. Weiser’s proposal, articulated succinctly in the introduc-
tory line of his now legendary paper, was that ‘The most profound technologies
are those that disappear. They weave themselves into the fabric of everyday life
until they are indistinguishable from it’. Weiser described what he believes to be
a necessary shift away from the PC, with its multimedia one-stop shop of ‘laptop
machines, dynabooks and “knowledge navigators”’ (1991: 94) and, at the same
time, a shift away from the computer as the locus of our attention, relegating it to
the periphery of our lives. This technical achievement, he argued, would require
three conditions: ‘cheap, low-power computers that include equally convenient
displays, a network that ties them all together, and software systems implement-
ing ubiquitous applications’ (p. 100).
As I type this, with the background hum of my Hewlet Packard work-
station, which is placed solidly and immovably on my desk, it is worth recognizing
that the PC is alive and well in the twenty-first century. However, while my locus
of attention is still fixed to my screen, my periphery is also computationally ena-
bled through my iPhone (which has the potential to, and often does, compete
for my locus of attention).
The mobile phone represents a critical device in the erosion of the
PC’s supremacy and is slowly being joined by other recognizable computational

131
Reality becomes display

devices (among them netbooks and tablet computers), which attempt more
limited but specialized multimedia capabilities, and many embedded computa-
tion devices that are not recognizable as computers; for example, those found
in household appliances such as fridges, washing machines, etc. The prolifera-
tion of microprocessors thus leads to specialization, with focused information
appliances attempting to provide one or a few functions better, the most obvi-
ous examples of which are to be found in gaming and entertainment (Xbox,
Playstation, Wii). Weisner’s prediction may not yet have been fully met, but the
trends lead in the direction he foresaw.

Re-tethering the ballast of materiality

In Chapter 3, I elucidated the notion of spaces of information, proposing three
broad categories of semantic space, the space of interaction and screen space,
together with a renewed definition of the architectonic system, which is essen-
tially a mode of interacting with information space in which the interactor’s
body is metaphorically projected into information space through the navigational
schema. The migration of information space away from the desk or lap bound
personal computer has lead to significant changes in each of these information
spaces.
Physical spaces have their own opportunities and limitations and the
body is no longer rendered as a projection, but is sensorially in direct contact with
information space. There are pragmatic consequences of this. Screen space,
for example, may become smaller and more mobile (as in the case of mobile
phones), larger and less mobile/more situated (as in the case of large urban dis-
plays) or may be reduced in resolution and fidelity as their presentation becomes
limited by power requirements and specifics of use. This diversification of output
possibilities isn’t limited to screens. A renewed focus on the body as the locus of
interaction has lead to a generation of hardware devices which encode interac-
tion in multisensory ways – haptically, kinaesthetically, auditorily, etc.
Ephemeral digital information is essentially tethered to material
objects. Anytime, anywhere becomes here and now and the physical objects can
have an identity as part of a network. Patterns of information can exist virtually
but the key properties of objects within an ‘internet of things’ are their physical
contexts.

Cultural associations
To say that a physical object exists both as a material presence and as a system
of objects with semantic associations is not an innovation of digital technology.
Our perception of our world is not limited to its material properties but extends
to include rich cultural and personal meaning invested in material culture. But,
as I have shown, our interpretation of this culture is often communicated and
constrained by visible material formations.
The supremacy of positions, places and contexts in the communica-
tion of meaning is challenged by the process of computational tagging, where
an identification marker, for example a radio frequency identification tag (RFID),
can be attached to an object and, through the right computational device, hidden
information on the object can be unlocked.

132
 Reality becomes display

I observed that Vannevar Bush’s proposal (2003) for the Memex

machine was a mechanical realization of a tradition of academic referencing. It
could also be suggested that the process of identifying objects through digital
tags, of embedding them within computers and causing them to interact across
networks, is simply a mechanical realisation of an already rich semantic space
of cultural associations. As with Memex, however, the mechanical enacting of
semantic associations means that the patterns of information that relate material
artefacts together, can exist independently of an object’s material articulations.

The location of objects

In Chapter 3, I introduced the multidimensional nature of semantic space and
the problem of its visualization. We are not used to living in a multidimensional
world and our senses are tuned to understanding the world in three spatial and
one temporal dimension. The notion of a multidimensional semantic space,
however, is not unique to digital technology. As I pointed out at the beginning of
this chapter, it is well recognized that the discursive formations of, for example,
books classified according to their position on library shelves, are one amongst
many possible configurations, and such configurations may be realized in dif-
ferent patterns in libraries all over the world. A book, in use, actually occupies
multiple different locations. Imagine walking into a library, browsing the shelves
of the architecture section and narrowing one’s search to hone in on books on
Digital Architecture. You pick up The Architecture of Information (Dade-Robertson
2011) and at that moment the book appears in the context of the library’s clas-
sification with adjacencies to other books that are considered to belong to the
same subject. Now the context of use shifts as you sit down at a desk in the
library. Reading through the text you follow the references. You begin to collect
books which are mentioned in the bibliography, following the trails left by the
author’s research. Your desk now accumulates more books and a new discursive
formation develops as the books I have referenced are joined by books you have
discovered. These books will eventually become knitted into a new semantic
space created, perhaps, by the essay or book you are about to write. My book is
eventually joined by your book on the library shelf, shifting, ever so slightly, those
around it until eventually (perhaps optimistically) other books join ours, creating
a new classification that clusters into a new discipline. The book occupies these
many positions over time and in the context of its use, and this is not only true
of research objects such as books.
The simple act of picking up and using a mundane object such as a
pencil, for example, requires me to make, mostly subconscious, choices about
how I interpret that object as a functional device with affordances that allow me
to hold and use it in a specific manner, for example, to make marks on paper.
There are many other interpretations of the object. For example, an alternative
functional understanding of the pencil might be to pick my nose with it, or we
may take an interest in its production history in terms of the materials used in
its construction; i.e. the location where the trees were grown for its wood or
the type of graphite used in the lead. To place the pencil in each of these infor-
mational contexts requires knowledge of a broader set of semantic associations
and a conscious or subconscious decision to filter only those bits of information

133
Reality becomes display

required to act in different contexts of the object’s use. For the majority of the
time, this interpretation involves the limited understanding of the pencil as a pur-
poseful, functional device and nothing more. There are, in my life, no contexts in
which I need to identify the wood from which my pencil is made, in contrast to a
pencil historian or the curator of a pencil museum, for whom this information is
of primary importance when putting together a book or exhibition of the material
sources of pencil production. In other words, information is filtered in terms of
its context which, in turn, is given by the situation of its use and interpretation.
When an object also occupies a virtual semantic space, its ‘location’
becomes much less clear. Rather than searching on the library shelves for this
book, you are more likely to have searched the computerized library catalogue.
The book has multiple digital locations through the various dimensions of the
library’s online semantic space as well as a broader set of online spaces which
may include digital repositories and databases. While these resources are simply
indexes for the book object itself, they stay separate. But what about when the
material book itself becomes the locus for its digital representations? In other
words the book is tagged with a digital identity. Now the book has an identity that
is not necessarily discursive or revealed through its contexts of place and use.
There is a relationship between physical objects and semantic space of which
we can only be partially aware.

Part 3: Place–action
Computational semantic space and its relationship to physical spaces and objects
may be partially invisible, but must, at some point, be revealed through a process
of interaction. However, the process of interacting with semantic space through
real objects is significantly different from interacting with semantic space through
screen-based environments. This difference can be illustrated with relation to a
debate between two computer scientists at an ACM conference in 1997.

Direct Manipulation vs. Software Agents

The debate that took place between Patti Maes and Ben Shneiderman was
entitled ‘Direct Manipulation vs Software Agents’. Shneiderman, as a pioneer of
graphical user interfaces and the originator of the term ‘direct manipulation’ in
describing the process of direct user input used in most GUIs, put forward a case
for human agency as a means of interacting with information spaces through
the visual space of the screen. In terms we set out in Chapter 3, Shneiderman
was proposing the use of visual space as the primary way of interacting with
semantic space, utilizing visualization methods that would allow the complexity
of semantic space to be reduced and displayed graphically. Through the crea-
tion of these visual environments, it would therefore be possible for information
objects to be visualized in a coherent and recognizable 2D or 3D space and for
the user to have control over the system’s visual display.
Maes, on the other hand, presented ‘software agents’ as an alterna-
tive to information visualization. Software agents are systems capable of inferring
the informational needs of the computer user and filtering that information,
reducing the complexity of semantic space so that the right information can be
provided. We commonly associate software agents now with services such as

134
 Reality becomes display

recommender systems, which can build knowledge on an individual’s patterns

of interests, likes or dislikes, and turn these into recommendations. Examples
include the Genius system on iTunes, which can recommend music and playlists
based on an individual’s music collection and listening habits, or the ‘You might
also be interested in’ section of the Amazon retail site.
On his side of the argument, Shneiderman cited ‘the remarkable
human capabilities in the visual domain’ (Shneiderman and Maes 1997: 44),
rejecting the notion of machine intelligence but rather placing agency with the
computer user. This agency is enabled by good graphic design and interface
controls, which allow visual information spaces to be explored and changed
incrementally and in predictable ways. Shneiderman attacked what he perceived
to be the anthropomorphized notion of computational agents (Shneiderman and
Maes 1997: 56).
Maes, on the other hand, rejected Schneiderman’s caricature of
software agents, distinguishing the notions of a ‘software agent’ from ‘autono-
mous or intelligent agents’ and citing the limited domain in which information
visualization is applicable:

Twenty years ago, one typically had one user using one computer,
and everything in that computer, every file, every object, was in a
particular place because the user put it there …. It was completely
structured and well organized. Today, our computer environments
are completely different ….
(Shneiderman and Maes 1997: 50)

Fluid interactions
The debate between these two paradigms was, inevitably, not settled in the
context of the 1997 discussion, but it is worth noting the research trajectory
that each of the debaters took later in their careers. While Schneiderman’s group
maintained its success in the world of information visualization in traditional
screen-based PC contexts and in highly specialized visualization domains (par-
ticularly scientific visualization), Maes’s group at the Massachusetts Institute of
Technology (The Fluid Interfaces Group) has been characterized by significant
and high-profile implementations of ‘invisible computing’.
The most high-profile project undertaken by Maes’s group is known
as SixthSense (Mistry and Maes 2009) and its demo videos have gone viral on
video-sharing sites. SixthSense combines a camera and projector, which are
worn by the computer user. The projector is capable of projecting on to any sur-
face in front of the user and the camera (aided by markers worn on the figure of
the user) tracks the hand gestures of the user which, in turn, are used as a way
of controlling the projected images.
This combination of relatively simple input and output systems and
sophisticated software has made for compelling, almost magical, demonstra-
tions of direct manipulation as a physical reality, i.e. hand gestures are used to
directly interact with the computer without the need for cumbersome hardware
intermediaries (e.g. a mouse). At first sight, it seems as if Shneiderman has

135
4.2
Images of SixthSense
showing, from top to bottom,
the SixthSense hardware
device and applications
including: using gestures to
take photographs, projecting
a clock face on to one’s hand,
using fingers as a calculator,
projecting real-time
weather information on to
a newspaper and projecting
live flight information on to a
aeroplane boarding card
 Reality becomes display

won the debate and Maes and her group have followed him into the world
of direct manipulation. The SixthSense system makes use of a highly literal
understanding of direct manipulation through interaction techniques that allow
hand gestures to be used to literally grasp and move virtual objects projected in
space. However, in order to achieve this embodied and ‘fluid’ interaction with
a computer, the computer system rather than the interactor needs to filter the
projected information and infer the appropriate information to show in different
contexts. Simple examples from the demonstration video involve, for example,
holding up one’s hands with the index fingers and thumbs in a frame gesture to
instruct the camera to take a picture of the scene, or making a circular motion
with the user’s index figure on their wrist to instruct the computer to project
a clock face as a virtual watch. Objects also form part of the dialogue with this
system. In one example, a flight ticket becomes the surface on which information
about flight times and departures appears. For the interactor in the video, they
simply hold up the flight ticket and the relevant information is projected on to it.
For the computer system, it must go through a complex sequence of operations
involved in recognizing and tracking the ticket and inferring the relevant informa-
tion at that moment in time and in that place (Figure 4.2).
SixthSense demonstrates a principle that transcends the direct
manipulation vs intelligent agent debate. The project accepts that interacting
with computational information in the complex and dirty information-rich environ-
ments of the real world requires the agency of both humans and machines. For
the interaction to appear seamless and flowing, the information being projected
on to the world needs to be pertinent to its context.

Context meaning
Using the aeroplane ticket example in the SixthSense demo, we can consider
the implications of context-aware information being projected at particular times
and in particular contexts. The process of taking a flight, from picking up the aero-
plane ticket to collecting the boarding card and going through the security check
and boarding the aeroplane, is an example of a highly programmed sequence of
objects, places and events. The user picks up and looks at a ticket while travelling
in a taxi on his way to the airport. Updated information is then projected on to
the ticket about the aeroplane’s departure time. The aeroplane ticket is a means
by which the departure lounge of an airport, and ultimately the aeroplane, are
negotiated. Information pertaining to these place–actions is presented via the
ticket and, during the process of a journey from the ticket collection stands to the
aeroplane, different places are coupled to different information requirements. For
example, we need to know the flight number to get to the right gate, the boarding
time to alert us to the appropriate time to listen for the boarding announcement
and seat information when the aeroplane is being boarded so that we know
where to sit. Airports can be seen as amongst the most programmatic of building
types, where functions, places and actions are highly specialized.

Classification of action
However, if we broaden the discussion of context to include a wider range of
human activities, the idea of context becomes more elusive.

137
Reality becomes display

Place and action can rarely be tied inexorably together. We do pro-

gramme spaces, but the designation of an action to a place does not automatically
lead to an action taking place there. For example, we know that a kitchen is for
cooking, but it doesn’t preclude the possibility that the place might be used for
socializing, playing music or making love. The architectonic system comes about
because of a relationship between patterns and classification in the organization
of people and places; however, interrogated closely, the boundaries between
classification and action in the real world are fuzzy.

Part 4: Space agency

The notion of the architectonic system as a means for the classification of objects
and human action is an important one to understand when looking at ubiquitous
computing because of the nature of software agents in interacting in real-world
contexts. Representations of place are 2D or 3D but places themselves are
semantically multidimensional, when we consider place to be associated with
different conceptions and to be associated with diverse programmed actions.

Multidimensional place
We have seen through the concept of the architectonic system that ‘context’
has traditionally referred to the patterning and setting of knowledge in physical
and, predominantly, architectural places. Explicit spatial structures and implicit
metaphorical spaces (revealed through the use of spatial metaphors in language)
are part of the practice of everyday life yet, if they are joined by digital encodings
of information, which are essentially invisible, they can exist without forming part
of the material patterning of objects in the physical world.
In Chapter 2, I showed how the architectural metaphor had been used
to shape the otherwise formless information spaces of computer-based informa-
tion. Here, architecture is seen as something fixed and stable, pertaining to the
solidity of the built environment as a conveyor of meaning through physical organi-
zations in space. A strange parallel to this is that recent architectural discourse has
been infused with metaphors of computing. In the terms of Vicente Guallart, for
example, the house becomes a keyboard for interacting with the world; spaces
can be programmed and re-programmed; cities become populated by icons and
can be understood in terms of their resolutions and be designed as nodal networks
(Guallart 2009). Buildings also become visualizations of information and interac-
tional interfaces5 and we surf them as if we were surfing the web (Paulini and
Schnabel 2007). Ironically, the metaphor of computation thus becomes a way of
simplifying the otherwise complex experience of the built environment in the same
way as architecture as metaphor attempted to simplify computing. The built envi-
ronment is multidimensional, but those dimensions can be limited using methods
that represent it as a network, an interface or a form of information visualization.
Acknowledging that physical objects placed in real-world contexts
actually have multiple ‘locations’ with relation to their semantic associations,
allows the sorts of tagging and referencing systems created by digital technol-
ogy to become an invisible part of their identity. It is possible to think of physical
objects, whether pencils or books, as being tagged, not in any digital sense, but
rather as the product of knowledge held in the minds of the individuals who

138
 Reality becomes display

interact with them. The key difference, however, between the cultural position
of a material object and its digitally tagged counterpart, is that cultural knowledge
is played out as part of a social practice; for example, by using the pencil to draw,
by using the Roman pot to hold water or by placing either of them in a museum.
Context, in terms of a relationship between action and place, is the means by
which information on an object is revealed.
Digital tagging doesn’t require this relationship between action and
place. Information can exist ‘in the cloud’, without reference to the object’s con-
text of use or of place. However, at the point of interaction, the computer system
needs to understand the context of an action to work out what piece of informa-
tion is relevant. This process of contextual understanding is summed up in the
aeroplane ticket example, where information is projected onto the card depending
upon its context of use as part of the process of arriving at the airport and board-
ing the aeroplane. To infer the context of interaction and to filter information to
be projected back on to an environment requires an ‘intelligent’ software agent.

Smart space
The transference of agency from the practice of everyday life to computational
agents is best summed up by a paradigm that might be described as ‘smart
spaces’ and which include notions of ‘smart homes’, ‘smart offices’ and ‘smart
cities’. Smart spaces have come in for some, not unwarranted, criticism and are
contrasted to a design ideal that seeks to redress the notion of computer agency
and place humans at the centre of the computational design discourse.6 Despite
these criticisms, as I have shown in Part 3, smartness or intelligence must be a
characteristic of any computational system that is implemented in the complex
context of the physical world. This intelligence manifests itself as a filtering
mechanism and may be as simple as generating the appropriate distribution of
devices to support particular human activities or, as with the aeroplane ticket
scenario, providing appropriate information at the right time and place. However,
the challenge for such smart systems is that, in addition to being shaped by loca-
tion and context, the technologies themselves also shape the experience and
function of those contexts.
A useful illustration of the constraints of computing on our experi-
ence of space can be seen in projects which attempt to map context as a design
method for the development of distributed computing systems. One approach,
described by Lan (2008) as ‘situated life patterns’, involves using descriptions of
functional uses of space as a way of mapping places and activities. The results of
this method are described by Lan in the context of a smart office environment.
Lan uses a scenario description of an office worker’s daily habits and schedules,
a fragment of which I reproduce here:

Prof. Lan walks into his office. When he stands in front of his office,
he is identified and allowed to get into his office by the smart door at
the entrance. His status of ‘inside office’ triggers a spatial event to
display the daily schedule on the smart wall. He notices his students
will come to his office for a group meeting later ….
(Lan 2008: 169)

139
Reality becomes display

These descriptions allow Lan to distribute various computer screens throughout

the office, mapping actions to particular places and times and distributing relevant
information to ‘smart walls’, ‘smart doors’ and ‘smart tables’. Such descriptions
of place–actions are deterministic, turning physical space into an enhanced GUI
with locations connected to applications. In Lan’s description, the door becomes
an application for external communication of schedules, etc., the desk becomes
the application for collaborative work and meetings and the wall becomes the
application for presentations. An understanding of the physical space of the office
and the objects arranged in it is constrained by the distribution of computational
devices, which leave no room for the space to be appropriated in ways which
are not preconfigured. A highly programmatic and deterministic understanding
of space is tempting as a way of appropriately distributing digital information
in physical contexts. Indeed, the architectonic system as I have described it is
based on such programmatic understandings of space. However, space as it is
represented schematically is different from space as it is experienced practically.

Strategic and tactical interactions with space place

In questioning the idea of context in relation to distributed computing systems,
Harrison and Dourish, who are computer scientists with a wide gaze over disci-
plines such as anthropology and philosophy, cite the relationship between space
and agency through what they describe as strategic and tactical spatial practices
and through a distinction made between space and place. So far in this book, the
concept of ‘place’ has been described as it relates to mathematical descriptions
of topology. However, outside this mathematical abstraction, ‘place’ has a richer
meaning and is distinguishable from space where ‘space is the opportunity; place
is the (understood) reality’ (Harrison and Dourish 1996: 67).
Space structures ‘the material and geometric properties’ (Dourish
2006: 299) of interaction and, where there is a mapping of functions to spatial
locations, the space is described by Dourish as strategic (such spaces include
factories, airports, theme parks and smart homes). Place, in contrast, is a product
of tactics, not only of programmes. Dourish quotes de Certeau in the context of
tactics, where an urban space is appropriated through an act of walking:

The act of walking is to the urban system what the speech act is to
language or to the statement uttered. At the most elementary level,
it has a triple enunciative function: it is a process of appropriation
of the topographic system on the part of the pedestrian just as the
speaker appropriates and takes on language); it is a spatial acting-out
of the place (just as the speech act is an acoustic acting-out of the
language); and it implies relations amongst different positions.
(de Certeau quoted in Dourish 2006: 302)

The distinction between space and place and their strategic and tactical appro-
priation resonates with the idea of a spatial language in which space, as a
structuring phenomena, acts as a syntax where place constitutes an environ-
ment’s semantics. Extending de Certeau’s analogy of the act of walking as a
speech act, an environment can be imagined as having three voices. In highly

140
 Reality becomes display

programmed or ‘strategic’ spaces, the individual is silent while the ‘voice’ of

the architectural space has a firm and direct intonation that the user is com-
pelled to follow. In unprogrammed spaces, the user is free to roam and create
space by their actions through enunciations akin to improvised poetry. In this
situation, space is created through actions in an environment, which, itself, is
relatively silent. Computation has the potential to act as the third voice. In the
examples used so far, the tying of information to particular places reinforces
the programmed voice of the environment. It takes an act of dissent, or even
vandalism, to change this situation.7 Some spaces, however, are open to tactical
appropriation where the relationship between place and action can be flexible
and entirely personal.
All spaces sit somewhere on this spectrum; sometimes we simply
act out space as it is presented and sometimes we act creatively in space to
generate new opportunities and forms of expression.8 Computational agents are
better able to handle spaces that are highly programmed, where the space action
relationship can be clearly read and anticipated. In the case of smart architecture,
we need to understand which devices, sensors and output devices are needed
where. In the case of the interactive aeroplane ticket, for example, the computer
needs to know what the relationship is between the informational needs of
the user and their context. Programming (in terms of place and action), allows
context to become machine readable. The third voice of computation, therefore,
tends to reinforce the programmatic relationship between action and place. But,
what happens when the third voice sings a different song.

Part 5: Invisible architectures

From an architectural point of view, we tend to consider buildings as the mate-
rial articulation of programmes, but the relationship between place and action
can be implied by the creation of invisible contexts. The ability of computational
agents to interpret and thus reinforce the relationship between place and func-
tion must also be considered along with another possibility: that the third voice
of computation is capable of creating contexts in its own right. In other words, it
is able to programme space. I have shown that deterministic understandings of
space lead to the reinforcing of functions in the context of, for example ‘smart’
buildings, but as technologies of information filtering become more sophisticated
and invisible, this determinism may become more hidden and independent from
material forms.

Territorial fields
Dourish cites an example of the shaping of the world through spatial practices
that are independent of material forms, using the territorial ‘fields’ and ‘exclu-
sions’ written on to the landscape by aboriginal tribespeople through their
process of ritual journeying. These complex, but invisible, spatial phenomena are
overlaid on top of the starkly barren landscape of the Australian outback through
ritual acts rather than built artefacts (Dourish 2006: 302–3).
Digital technologies have the potential to encode ritual and reinforce
certain behaviours in ways that are not directly related to place or spaces as
they are currently conceived. In the example of the office environment used in

141
Reality becomes display

Part 4, there was a mapping of functions to space through the implementation

of ‘smart screens’ which were mutually reinforcing, but this does not have to be
the case. Imagine, for example, a recommender system, similar to the one used
to recommend books on Amazon, applied to the ritual of shopping on the high
street. Behaviours of consumption, which are shaped by the organization of the
shops and their contents, are supplemented by a mobile device which, having
spotted your shopping patterns, recommends other shops and goods for you to
buy. If you take the advice of the recommender system, your physical movement
through the shops in the high street becomes a product of the recommender
system and your experience of space is thus shaped by a computational agent
that is separate from the physical reality of the built environment in which you
are shopping.
The relationship between computational agents and space varies
depending on the nature of the system’s agency. Dourish suggests that ‘Empty
Space thickens when mixed with information, making space itself an interface,
and thus part of social space’ (2004: 380), but the degree of ‘thickening’ depends
on the relationship between information and our spatial practices. The relation-
ship between digital information and our behaviours in, and understanding of,
space have been classified by Crang and Graham into three types of mediated
spatial practice, which they term ‘augmenting space’, ‘enacting space’ and
‘transduction space’ (2007: 792–4).

Augmenting space
Augmenting space refers to the process of overlaying of our environment with
computational data. The technologies of augmentation are often associated with
data tagging through the use of markers, identification tags or location aware
hardware. A classic example of an ‘augmented reality’ (AR) installation, for
example, involves the use of gliffs (black and white printed markers), which can
be attached to objects and then tracked using a camera and software designed
to locate and calculate the orientation of the marker. Using such technologies,
an image of the scene through the camera’s eye is augmented with a 3D virtual
object, which is updated in real time so that movements of the gliff can be
tracked. The 3D virtual object then appears on screen (or sometimes through a
head-mounted display) as if it is part of the scene.
While classic AR is often related to the particular technology paradigm
described above, we can broaden this definition to any technology that reinforces
strategic space by mapping information to unique locations and known contexts
of place and action. Augmenting space refers to a direct and transparent one-to-
one mapping of place object and information

Enacting space
The process of enacting space builds on the technologies of augmenting space
but introduces an intermediary process of inference and filtering. A decision
is taken by a computer system as to which piece of information needs to be
presented in specific contexts. Agency shifts from the user’s activity in interact-
ing with an environment to the agency of the computer in occupying physical
objects and spaces. The material world thus has a dual presence through what

142
 Reality becomes display

Cuff describes as ‘cyburgs’, objects that exist simultaneously and are connected
in virtual and real contexts (Cuff 2003).

Transduction space
Crang and Graham develop a third spatial process described as ‘transduction
space’, within which an object, or indeed a person, can be located (what I have
described as semantic space) in ways of which they are unaware and have no
direct access to, so that the identification and decision-making part of the system
creates a ‘technological unconscious’ through what Thrift (quoted in Crang and
Graham 2007: 794) describes as

the bending of bodies-with-environments to a specific set of

addresses without the benefit of any cognitive inputs, a prepersonal
substrate of guaranteed correlations, assured encounters, and there-
fore unconsidered anticipations.

This contrasts with a world in which currently

spaces depend upon the gradual construction of complex ethologies

of bodies and objects, which are repositories of the ‘correct’ posi-
tioning and juxtapositionings that allow things to arrive and become
known … the modest but constant hum of connection and intercon-
nection that they make possible.
(Thrift 2004: 177)

Thrift describes the difference between a discursive and a non-discursive space.

Between one space in which action, space, place and programme are explicitly
linked and visible, and another in which our experiences are shaped by invisible
forces. This means that spaces which are otherwise open to tactical appro-
priation are in fact strategic. These spaces are not articulated through physical
architecture, but by systems which generate invisible associations between
places.

Shaping experience
The implications of a transduced space are wide and significant. Decisions that are
made by systems, which are non-discursive, about our behaviour and the informa-
tion to which we have access, have implications in many aspects of our lives. I
have shown that these systems exist already in the form of innocuous software
agents such as recommender systems, where these agents have the limited
power to affect our buying decisions. When integrated with computational rep-
resentations, which are pervasive in the world, they have the potential to shape
our experiences, through a whole range of contexts, for the better or worse.

Part 6: Conclusion
The architectonic system with which I started this book was defined to enable
me to look at how representations of information in computers were shaped
by the patterns we encountered and author in our environment. The case for

143
Reality becomes display

the architectonic system rests on there being a mutually reinforcing relation-

ship between our articulation of the material world and our conceptual world
through a relation between topos and topic and between category and place.
Computational technology has done its best to transcend this ‘ballast of mate-
riality’, but has not been entirely successful. However, as computers become
part of the material world, the relationship between virtual information and its
physical manifestations becomes complex. We no longer face the dichotomy
of the ‘non-space of the mind versus city lights receding’ and we don’t have a
computational world that is constrained by the geometric possibilities of space,
but rather a physical world shaped by the possibilities of computation.
In this chapter, I have defined space through the notion of compu-
tational agency and its relationship to places and actions. A central problem
for computational systems, which are pervasive and distributed in real-world
contexts, is of understanding what information a computer user needs and
where and when they need it, in other words filtering information dependent
on contexts.
The difference between direct manipulation and agent-based sys-
tems of computational information is blurred in these physical contexts, which
require context-based filtering of information, in order to become part of ‘fluid’,
directly manipulated interfaces.
The notion of contexts in computation can be considered though an
understanding of strategic space, where there is a direct relationship between
places and actions (or as I have described it: place–actions). However, we must
also take into account tactical interaction in space, where space is appropriated
and flexible. The computational relationship to these different types of spatial
practice can lead to:

1 a reinforcing of strategic space by, for example, in the case of smart archi-
tectures, associating places and objects with specific applications
2 the creation of an invisible strategic space and thus creating a programmed
space that is inscribed by the behaviour of those who move through it.

The architectonic system becomes characterized, not in the constraints of space

as a way of categorizing the world, but by computational categories as a way of
shaping our behaviour in space.

The future of the architectonic system

If the technological predictions turn out to be true, and the course of computing
creates a world in which computation is invisible and mediated by computational
agents, then ubiquitous computing has the potential to radically change our rela-
tionship to the architectonic system as it is currently understood. Information
only becomes relevant once it is linked to action. By shaping our experience
of space and influencing our behaviours, computational information acts as an
intermediary between place and action and software becomes as ‘architectural’
as its more traditional ‘hardware’ counterparts.
I suggested, in my hypothesis, that ‘the spatial patterns which we
make in our environment are a primary means of human communication’. My

144
 Reality becomes display

subsequent analysis of the way in which spatial metaphors have been used in
computation have shown this statement to be true and have shown that ‘despite
the potential created by digital technology to transcend the material organiza-
tion of information, patterns that relate to the organization of architectural space
continue to have a role in the design of digital information systems’. And yet,
when computation transcends the metaphor and actually becomes bounded by
real architectural space, the validity of the last part of this hypothesis begins to
erode. We are entering an era in which computation shapes architectural space
rather than being shaped by it. The shaping occurs not simply on a material level
through the creation of computer-generated forms, but through the process by
which we use and experience space as a functional and semantic system.

145
Conclusion

Review
This history of information space started with the emergence of categorization
as a means of articulating the conceptual world initiated by the organization of
social space and the spatial distribution of social groups. Citing Preziosi’s con-
cept of the ‘architectonic system’, I sought to understand architectural space
as a medium of communication distinct from language. Through the concept of
conceptual spaces I also showed that, as well as being shaped by our material
engagement with the architectonic world, the way we order our ideas is related
to our embodied cognition. Through Lakoff and Johnson’s theory of ‘image
schemas’, I showed that representations of abstract ideas can be constrained,
through metaphor, to embodied patterns and I isolated those schemas which
relate to space as a relational topological phenomena. I also showed how the
architectonic system evolved into distinct building typologies and I focused on
mnemonic architecture as a link between Aristotelian notions of topos (place)
and topic through the method of loci.
In Chapter 2, I extended the discussion of mnemonics to include ars
combinatoria and showed how the method of loci leads to an understanding of
information and its representation through the possibilities of combinatoriality.
Philosophers such as Ramon Lull sought to overcome the restrictions on the rep-
resentations of information fixed to places, by inventing combinatorial machines
where semantic relationships existed in the potential of the machine rather than
in fixed spatial patterns. Lull was an early logician and I traced, through him and
through the much later proposal by Vannevar Bush and the architectural work
of Paul Otlet, a shift in an understanding of both the mind and of information
representation from being fixed and patterned to being dynamic and associative.
This philosophical shift was made real by the invention of hypertext and realized
as the WWW. Despite its potential, however, implementations of interfaces with
computer systems and with the GUI of personal computers and the patterning
of hypertexts on the WWW, found themselves constrained by fixed locations
and spaces of interaction.
Chapter 3 investigated the emergent spaces of information created
by digital technology. While rejecting the increasingly elaborate architectural
metaphors applied to interacting with computational information, I proposed a
model of information space consisting of three components: semantic space,
screen space and visual space. I showed that each type of space is framed by
different spatial metaphors so that, for example, semantic space is often con-
sidered as information held within a computer and modelled as objects arranged
Conclusion

in multidimensional spaces where these spaces can be measured as geometry.

By separating visual space from interaction space, I also differentiated between
the real-estate of the screen and the process of interacting with a computer
to change the screen’s output. I then proposed a mode of interaction which
I described through an addition to Lakoff and Johnson’s image schema and
which I called the ‘navigational schema’. I suggested that acts such as navigat-
ing a website can be considered to be navigational despite the fact that they
have, either in appearance or reality, little to do with real navigation because we
attempt to use some of the same cognitive faculties in interacting with them as
we do in navigating through real spaces. The architectonic system, I suggested,
survives in our interaction with information because of the navigational schema,
but attempts to associate navigational modes of interaction with methods that
present the space of the screen as architectural are superficial and fail to relate
to why the navigational metaphor exists in the first place.
Finally, in Chapter 4, I analyzed the architectonic system in the age of
ubiquitous computing. While noting that the merging of the material and digital
worlds has reinforced the importance of computation as framed by our material
engagement with the world, this chapter looked at the challenge of agency in
computational systems, which are sensitive to context in terms of place and
action. Chapter 4 was futurist in scope and looked at emerging technologies. I
observed that, where the use of metaphors of space and place had historically
structured and, in many ways, constrained our interaction with virtual informa-
tion, computation in the context of ubiquitous computing can, in turn, shape and
constrain our interactions with space. I showed that, in the context of pervasive
distributed computing systems there are no contradictions between interaction
methods which allow computer users to have direct and embodied relationships
to information, and agent-based systems that filter the information available to
the user. I showed how, in designing such systems, the designers have a choice
between augmenting space and thus reinforcing strategic spatial practices,
which are planned and programmed, and creating new structures of behaviour
that may not change the form of the built environment but do change how we
use it through new but invisible ‘transduced’ spaces. I concluded Chapter 4 by
commenting that the architectonic system, which had given rise to the use of
our environment as a means of communication and has survived even when
challenge by a world of virtualized information objects, is now being challenged
in a world in which an object can have both a physical and a virtual presence.

To begin again …
At its heart, this book provides a description of individuals surrounded (some-
times literally and sometimes phenomenologically) with objects, the spatial
organization of which expresses meaning beyond their material articulation.
This space is one that can be read by traversing its topology, where navigation
becomes the means by which meaning is selected and countless navigations
become the method by which the entirety of the semantic space is understood.
Through the navigational schema, I have proposed a mechanism for linking cogni-
tive space with information space through the use of metaphors that structure
our perception of systems which we perceived as egocentric.

148
 Conclusion

In the discussion of early proposals from Bush for the Memex and
Englebart in his pioneering work on the personal computer, the focus was to
‘extend man’s intellect’ and resonates with a profound ambition that is, at least
in part, being met. Their vision has necessarily been superseded by thousands
of endeavours to extend and improve computing experiences, to develop new
user interfaces and find new structures through which to organize information
and tame the vast and unruly structures of networked information systems
such as the web. However, as I wade through yet another paper on a variation
of a table-top interface or slight alteration to a web browser, I am impatient for
another project that has aspirations as ambitions as to ‘augment man’s intellect’.
I believe, however, that discussions on embodied computing, which I hope to
have added to in a small way here, have contributed to an underlying discus-
sion that makes new revolutions possible. The locus of discussion is the human
interactor and, most importantly, their material interaction with the world. In
this context, I have demonstrated that separating information from its material
means does not necessarily result in a pure form of associative semantic space
and that material patterns seem to exist, even in virtual systems, once a human
interactor is involved.
Through the discussion of the architectonic system and the cogni-
tive model of the navigational schema, I have offered an approach to the future
development of computers, which is derived from the design of the built envi-
ronment as a way of representing, manipulating and experiencing information.
There are many other relevant perspectives but I am hoping that architecture,
through the further development of this discussion, may become more than a
peripheral provider of metaphors.

Yes, but is it architecture?

The first agenda I had in writing this book was expansive in that I wanted to
develop the notion of architectural space beyond the domain of the built environ-
ment so that it may be applied to the development of computational systems. In
aiming for this, however, I have questioned systems that overtly resemble archi-
tectural spaces or, at least, resemble our representation of buildings. Instead,
I have sought ‘architecture’ in the unlikely places of computer user interfaces,
hypertext networks and, ultimately, the human mind. One pervasive but not
unwarranted question I get asked once the central themes of my argument are
explained to an architectural audience is something to the effect of ‘Oh I see and
can this be applied to real architecture?’. My response is patient but firm; that
this is real architecture. As I have shown, understanding information and commu-
nication technologies involves understanding the way in which spatial cognition
shapes structures of thought and how we articulate spaces to communicate our
collective mind. This is surely a fundamental practice of architecture, even if it is
not always a practice in bricks and mortar.

Discussion
If there is a single take-home message from this book, it is that our embodied
manifestation has had significant implications for the way in which we inter-
act with even the most dematerialized of information and how we shape our

149
Conclusion

abstract experiences with reference to our physical experiences. I’m not the first
person to make this claim, but I have done so uniquely here through the lens
of architectural design and the design of digital information systems. This is, in
my view, a fruitful dialogue, both in terms of understanding the appearance of
certain digital artefacts and information structures and in providing an insight for
possible new design directions.
This view is also a warning. The architecture of cyberspace diverted
attention away from the development of the GUI and the emergence of
information architecture and, similarly, I now detect that the descendants of
cyber-architecture are focusing on the creation of novel forms (blobitecture,
non-Euclidian geometry, etc.). While not dismissing these paradigms, and while
also understanding the importance of architecture’s formalisms, it should not
be forgotten that geometric form is only one aspect of architectural meaning.
Changing forms do not necessarily equate to changing spaces. There is much
evidence of a merger between interaction design and architecture and this fruit-
ful merger can only be enhanced by the development of software tools such
as processing, and hardware platforms like Arduino that offer architects and
designers a bridge to allow them to sketch in media that have, until now, been
the sole domains of computer programmers and engineers. I have found the
field of HCI, in particular, to be rich in design content, ideas and artefacts, which
have a significant impact on our lives and have evolved to become powerful
reflections, not only of the companies that have brought them into being, but
also the people they serve. Windows or Mac OS are, in my view, as much part
of our architectonic system as any building.
I believe that the relevance of this potential collaboration is all the
more important when we look through the history of computing and the change
in our understanding of what constitutes a computer user interface. The study
of human interaction with computer systems has moved through three distinct
phases, characterized by the emergence of GUIs in the 1980s, tangible user
interfaces (TUIs) in the 1990s and 2000s and, most recently, through embedded
and spatially distributed computational systems, and through ambient, intelligent
and pervasive user interfaces. Each paradigm in human–computer interaction
is associated with particular technical challenges, cognitive models and design
paradigms. For example, GUIs have traditionally been developed with reference
to the evolution of computer graphics, gestalt models of visual perception and
design paradigms such as direct manipulation. Similarly, the development of TUIs
has brought together the technical challenges of sensing and multi-touch interac-
tion, theories on embodied and situated cognition and the design of ergonomic
and physically responsive material artefacts.
While the study and development of ambient, intelligent and perva-
sive user interfaces has been substantial and there has been growing research
effort across a range of fields (both inside and outside computer science), the
core principles that bind the technical development and design to an understand-
ing of their cognitive affects have yet to be fully understood. Where graphic
design supported the development of the GUI and product design supported
the design of tangible user interface, surely the new century will prize architec-
tural design along with relevant cognitive theories in the development of situated

150
 Conclusion

and pervasive computing interfaces. In particular, following the discussion in this

book, we can observe that, while models of design and cognition in human–
computer interaction are centred around an allocentric frame of reference, (the
manipulation of physical objects), pervasive computing requires an understanding
of egocentric frames of reference (involving whole-body immersion).
We might even call these new types of interface AUIs (architectural
user interfaces). It remains to be seen what new artefacts and design agendas
might follow their evolution.

151
Glossary

Absolute space
A space defined by absolute coordinates within 2D or 3D where an object’s
position can be related to its location in relation to x, y and z axes.

Allocentric
In cognitive science, allocentric is used to refer to an understanding or memory of a
situation (often a place) that is independent of direct sensorial embodied input. For
example, the cognitive map is an observer-independent representation of the world
in that it does not correspond with any single element of sensory experience but
is, rather, a knowledge representation composed of combined knowledge inputs.

Architectonic system
Defined by Preziosi as ‘the entire set of place-making orderings whereby individu-
als construct and communicate a conceptual world through the use of palpable
distinctions in formation … [which are] addressed to the visual channel, to be
decoded spatio-kinetically over time’ (Preziosi 1979: 4). The term is also used by
Kaplan and Multhorp (1994), to describe the visual space of a computer screen
when interacting with information. The term has been extended in this book
to refer both to the built environment and any information system with which
navigation is a primary means of interaction.

Architectonic space
This is the realization of the architectonic system as a network of meaningfully
connected places.

Cognitive collage
The term originates from Tversky (2001) as an evolution of the term ‘cognitive
map’. Cognitive collage acknowledges that our memory for our environment is
built, not from one map like representation, but from a complex mixing of mul-
tiple types of knowledge representation.

Cognitive map
The term originates from Tolman (1948) and describes the representation of
navigable space in the mind. Although still widely used, the term ‘cognitive
map’ may be misleading as spatial memory is not thought to be encoded in a
single map-like representation, but rather built through multiple different types
of knowledge and sensory input.
Glossary

Cyberspace
A term coined by William Gibson in his short story ‘Burning Chrome’ (1986) (first
published in 1982) and defined later in his novel Neuromancer as: ‘a consensual
hallucination experienced daily by billions of legitimate operators, in every nation,
by children being taught mathematical concepts …. A graphic representation of
data abstracted from banks of every computer in the human system. Unthinkable
complexity’ (Gibson 1995: 67). Cyberspace is most often used to refer to the
internet and the WWW. It is also sometimes used interchangeably with ‘virtual
reality’.

Egocentric
Refers to the point of view of an individual ‘on the ground’ through their direct
perception of an environment involving observer-dependent, direct and synchro-
nous perception.

Geometric lock
Author’s own term to describe the process by which a particular object’s position
in 2D or 3D space is triangulated by its fixed relationship to three other objects,
leaving it unable to move feely to make associations with other objects in the
system. It is used uniquely in this book to describe the problem of building a
spatial visualization of relationships between objects (most often documents)
based on measures such as the similarity of their content.

Graphical user interface (GUI)

The graphical representation of computer files and applications. It commonly
refers to the ‘desktop’ representation that is common to most PCs and to an
interaction paradigm, which includes direct manipulation incorporating the ‘what
you see is what you get’ principle.

Human–computer interaction (HCI)

A study of the way in which humans interact with computer systems. HCI unites
the disciplines of psychology and computer science as a means of both designing
and evaluating new computer systems.

Hypertext
A system that allows for the digital connection of text documents through cross-
referencing ‘links’. The term ‘hypertext’ most commonly refers to ‘chunk-based
hypertexts’ defined as ‘[Information organised] … as a network in which nodes
are text chunks (e.g. lists of items, paragraphs, pages) and links are relationships
between nodes (e.g. semantic associations, expansions, definitions, exam-
ples …)’ (Rouet 1996: 3).

Hyperspace
A term developed in mathematics but used in science fiction to denote a space
with more than three spatial dimensions.

154
 Glossary

Icon
A graphical representation of an object or application used in computing to act as
symbolic representations of system objects. Icons usually resemble some ele-
ment of the application or object represented, so for example, an icon of a text
document appears as a piece of paper with lines representing text.

Information architecture
A term introduced by Richard Saul Wurman (1996). Although there are, as yet,
no clear definitions, the Institute of information architecture defines information
architecture in the following way.

1 The structural design of shared information environments.

2 The art and science of organizing and labelling websites, intranets, online
communities and software to support usability and fundability.
3 An emerging community of practice focused on bringing principles of design
and architecture to the digital landscape. Available online at http://iainstitute.
org/ (accessed January 2006).

This book focuses on the third definition and in particular the notion of informa-
tion design as an architectural design practice.

Information object
Any object which acts as a container for information, for example a document is a
text information object. In the context of digital technology, an information object
may be considered to be a text document, image, data file, database entry, etc.

Information space
A term used to describe the ‘space’ of digital information held within a computer
system, either locally or across a network. An information space can consist of
a structured database, a hypertext system, a filing system or any other object-
oriented system of storage. The term information space has been considerably
expanded in this text to encompass semantic space, visual screen space and
interaction space.

Information visualization (InfoVis)

A discipline involved in the development of interactive, and often animated,
information and data graphics to give users insight into complex digitally encoded
information.

Interaction design
An extension of HCI, which looks at user interaction with a wide range of, mostly
digital, products and systems. Interaction design encompasses not only psychol-
ogy and computer science, but also disciplines from the fields of social sciences,
humanities and design (including architecture).

155
Glossary

Interaction space
A term developed in this book to describe the ‘place between clicks’ when
interacting with a computer. Interaction space may be conceptualized in different
ways. Commonly, however, interaction space is conceived either allocentrically
(through a process of navigation) or egocentrically (through a process of direct
manipulation).

Internet
Refers to a physical infrastructure of interconnected computer nodes that make
up a communications network whose primary (though not only) role is to store
and distribute Web-based information.

Information retrieval (IR)

The process of extracting meaningful information from large archives of digital,
usually text-based, information. Applications of IR include the development of
the WWW, digital libraries, database searches and the extraction of structures
for information visualization.

Means/material means
A component of representation that consists of the material manifestation of
a code in the process of communication. For example, the material means of
textual communication may be ink and the paper on which the text is written.

Medium/media
Used here to refer to a combination of both material means and code, com-
bined to become the channel of communication that carries a particular mode
of expression.

Mental (image) schemas

A term developed by Johnson to describe the way in which we make use of pat-
terns from our ‘physical experience to organise our more abstract understanding’
(Johnson 1990: xv).

Multi-user domain/dungeon (MUD)

Mostly text-based environments where players read descriptions of rooms,
objects, events, other characters and computer-controlled creatures or non-
player characters (NPCs) in a virtual world. Players usually interact with each
other and the surroundings by typing commands such as ‘go north’, etc.

Navigational schema
A term developed in this book to refer to a computer user’s conception that
interacting within a highly abstract information space is akin to the process
of navigation. The navigational schema is derived from the concept of mental
schemas (see above).

156
 Glossary

Planarity
The condition that exists if a topological graph can be drawn without its edges
crossing. In order to achieve this state, the graph in question must not contain
K 3,3 or K 5 subgraphs; i.e. graphs where a group of three nodes are connected
perfectly to another group of three, or where five nodes are connected perfectly
together. Graphs are considered planar if they can be converted into a combi-
natorial dual graph.

Principled space
A space in which the coordinates are given meaning. For example, in a statisti-
cal scatter graph, the axial positions of objects relate to a scale of values in 2D
or sometimes 3D.

Relative space
A space defined by the relationships of objects to each other rather than in terms
of absolute coordinates. It relates closely to the notion of topological space,
which is composed solely of objects and their associations to one another,
defined by connecting edges.

Representation
Using the OED definition: ‘A material image or figure; a reproduction in some
material or tangible form’. Representation is described in this book as the layer of
material presence that allows for the articulation and subsequent communication
of abstract concepts from a sender to a receiver. The components of representa-
tion have been described as a code (i.e. language), its expression (e.g. speech,
text or architecture) and its material means (e.g. sound waves, paper and ink,
or bricks and mortar).

Semantic space
The means by which the organization or configuration of objects represents some
aspect of their information content. The term ‘semantic space’ was defined by
Wexelblat (1991) with relation to the production of mathematical diagrams where
a principled space is used to define semantic locations, which, when occupied by
an object, imbue that object with a property or characteristic. The term ‘semantic
space’ has been extended in this book to include a range of topological as well
as topographical spaces where no pre-defined principles need to be applied but
where objects and their relative positions indicate semantic as well as physical
relationships. Semantic spaces can thus include museum environments where
objects and their relationships are perceived as a sequence of hypertexts, where
links represent spatial associations between textual information objects.

Set
A collection of objects, usually with similar properties, which are grouped
together and bounded by a container. Often used to describe topological rela-
tionships where objects are arranged together in a contained space with an
undefined volume or geometry and where this bounding communicates clas-
sificatory relationships.

157
Glossary

Single user interface (SUI)

A computational interface designed to be interacted with by a single user at a
time. Often graphical user interfaces for personal computers are considered to
be SUIs.

Space Syntax
Defined at www.spacesyntax.com (accessed January 2010) as ‘the world’s first
computer-based modelling technique to treat cities and buildings “space first”,
that is as the network of spaces we use and move through’. Space Syntax is
known for its techniques of urban analysis, which are based upon topological
representations of the built environment.

Spatial ontology
Ontology, in the context of this book, means the specification of a conceptual-
ization and is used, specifically in the context of spatial ontology, to refer to the
use of spatial organization as a means of categorizing or adding meaning to a
collection of objects.

Symbol
An object, image or icon which stands in for or represents something else.
A symbol differs from an icon in that the representation does not necessarily
resemble the object of representation but may have an agreed but arbitrary
relationship to its subject. It follows from Ferdinand Saussure’s discussion of
the sign and signified (1983).

Topography/topographic space
Using the OED definition: ‘The science or practice of describing a particular place,
city, town, manor, parish, or tract of land; the accurate and detailed delineation
and description of any locality’. Used in the context of this book in contrast with
topological space, as a representation of space that encodes information such
as metric distance and geometry.

Topology/topologic space
Using the OED definition: ‘the branch of mathematics concerned with those
properties of figures and surfaces which are independent of size and shape
and are unchanged by any deformation that is continuous, neither creating new
points nor fusing existing ones; hence, with those of abstract spaces that are
invariant under homomorphic transformations’. Topology is used here to refer to
a type of space concerned with places or objects and their association with one
another, independent of metric distance or geometry. Hypertexts, for example,
can be treated as topological structures where the links are valueless connec-
tions between document objects. Topological structures are often represented
through connected graphs containing points or nodes connected together with
links or edges. Specifically, topological descriptions of space have been used to
reveal the configurational properties of buildings through descriptions of floor
plans, and city maps through the theory known as Space Syntax.

158
 Glossary

Virtual reality (VR)

Virtual environments which attempt to simulate aspects of the real world in
high-fidelity and often through immersive technologies such as head-mounted
displays.

World Wide Web (WWW)

A system of interlinked hypertext documents distributed across, and accessed
through, the physical infrastructure provided by the internet.

159
Notes

Introduction
1 Including pre-Gibson movies such as Tron (1982) and depictions based on Gibson’s stories
such as Johnny Mnemonic (1995).
2 Although others were: see for example Woolley 1993.
3 The City of Bits (1997), E-topia (1999), Me++ (2004).
4 McCullough’s book Digital Ground (2004) was once described to me, in a positive way, by a
computer scientist, as ‘the book of 1,000 PhD topics’.
5 A term coined by C. P. Snow in The Two Cultures (1993), a Rede Lecture in 1959 to describe
what he perceived to be a growing gulf between the Arts and Sciences.
6 It is worth noting that the distinction between absolute universals in architecture and
relative cultural specific meaning is a constant subject of the battle between modernism
and postmodernism where the former strips architectural form away to its simplest purest
elements and the latter revels in the messiness of cultural context and the communicative
patterns that ensue.
7 See, for example, Steadman (1976, 1983).
8 See Steadman (1983) for an overview.
9 Notably Ratti (2004).
10 See Preece and Rogers (2002) for a recent example.

1 The architectonic system

1 Nikolaus Pevsner in An Outline of European Architecture (1943) starts his historical account
with the Parthenon dating from 447 BC.
2 Saessure’s major contribution to linguistics is published in his Course in General Linguistics
(1977) first published in 1916 after being collected together by his students Charles Bally and
Albert Sechehaye.
3 Among the many discourses that might be described as spatial philosophies are the
ubiquitously cited (both in architectural philosophy and much postmodern social theory)
Deleuze and Guattari (see for example 1988). Their constant use of spatial and geometric
metaphors makes them as quotable as they are impenetrable.
4 Languages of architecture are most often associated with particular styles or movements –
as in The Classical Language of Architecture (Summerson 1980) and The New Paradigm in
Architecture: The Language of Post Modernism (Jencks 2002).
5 As well as Space Syntax there are structuralist theories of architecture around shape
grammars (Knight 1994) and many mathematical models of architectural geometry and spatial
arrangements (see for example March 1998).
6 Aboriginal rock art for example is thought to have a territorially important significance, not
just in terms of its depictions but in its placement with relation to ritually significant sites.
7 Spiro Kostof speculates on an architecture of boundaries (associated with a plot of land) that
predates an architecture of monuments which is more naturally associated with human built
forms (Kostof 1995: 21).
Notes

8 For example, the ‘Sittard’ prehistoric settlements in the Netherlands, which date from the fifth
millennium BC, already show collections of timber long houses arranged as a village (Kostof
1995: 29).
9 Such attempts, he suggests, are ‘like trying to understand the organization of language
through the study of proper nouns’ (Preziosi 1979: 3).
10 Preziosi makes this claim with reference to the pioneering cognitive psychologist Ulric Neisser,
who states that ‘Because perception and action take place in continuous dependence upon
the environment, they cannot be understood without an understanding of the environment
itself’ (Neisser 1976: 183).
11 Amos Rapoport categorized language-based approaches in The Meaning of the Built
Environment as: 1) using models based on linguistics (of which Preziosi is a partial proponent);
2) ‘Relying on the use of symbols’; and 3) studies of non-verbal communication (Rapoport
1982: 36).
12 The paper was first published in 1903 and I am using an edition reprinted as a book from
1963.
13 One of Claude Levi-Strauss’s most famous texts is called The Raw and the Cooked (1995).
14 It is notable that Claude Levi-Strauss’s proposal of cultural distinctions devised through
apparent dual oppositions is well illustrated by Cartesian dualism.
15 Sowa (2000) provides a comprehensive study of the relationship between modern
computational knowledge representation and its background in earlier philosophy.
16 Examples include the tree of Porphyry, which divided all substances into genus and genera
and where the division of each is predicated on what Aristotle calls differentiae. A logical
operation is performed in this network by simply tracing down the trees branches until the
appropriate category is reached (Sowa 2000: 5).
17 Goldstien (2005: 286–7) uses the example of flightless birds, where flight becomes an
exempted predicate for a particular type of bird.
18 The references I have used most often are Goldstein (2005) for general overviews and
excellent examples and Goldblum (2001) for a more detailed but nevertheless readable
account of connectionism.
19 It should be noted that Gardenfors’s use of a spatial analogy here is interesting but also
potentially confusing in the context of this book since, while the analogy of the jungle is not
an attempt to describe spatial cognition (it is just an analogy), as will be seen, the conceptual
level is associated with rules which emerge from the spatial structuring of concepts.
20 Gardenfors uses the example of colour categories with reference to scientific work on
describing the geometry of the conceptual space of colour categories (2000: 71–7).
21 See for example Johnson (1990), Lakoff (1990) and Lakoff and Johnson (1980, 1999).
22 See for example Lacour (1996), Karatani (1995), Wigley (1995) and Hendrix (2003).
23 Many of these systems are found in use today for remembering abstract facts such as
the names of chemicals in the periodic table, the order of the points of the compass or
mathematical equations.
24 Consider for example, ‘The Tower of Wisdom’ (Carruthers and Ziolkowski 2002: 215),
consisting of steps of mental and corporeal meditation, leading to floors of virtue and to the
tower of wisdom at the top. This is neither building nor textual manuscript, but a hybrid of
both.

2 Between city lights receding and the non-space of the mind

1 Although the use of card catalogues essentially has the same effect, as each card represents
a representation or stand-in for the book object, and they are arranged differently from the
books in the stacks.
2 I borrow this number from Eco (1995: 54) where he gives a very clear explanation of the
combinatorial mathematics behind it.

162
 Notes

3 In fact, complex rules about the use of the system mean that there were many more possible
outcomes because a given sequence of letters could be interpreted in many ways (see Eco
1995: 59–62).
4 While Lull’s diagrams and devices are far removed from the corporeal engagement with vast
‘memory palaces’ derived from the method of loci, Lull’s proposals are, nevertheless, based
on the method of loci as a way of understanding combinatorial structured places as leading
to the construction of meaning (Yates 2001: 186).
5 Although Bush’s article is usually read in the context of the post-war years (when it was
published) it was in fact written in the years before the Second World War broke out. Wright
gives a useful account of the context of Bush’s work in Glut: Mastering Information through
the Ages (2007: 192–203).
6 Many authors have placed Bush’s description of the Memex as a hugely influential factor in
the initiation of the digital revolution, including Wright (2007) and Wardrip-Fruin and Montfort
(2003).
7 Howard Rheingold narrates an entertaining but authoritative history of what he describes as
Tools for Thought (1985), from the early history of computing through to the early pioneers
of human–computer interaction technologies.
8 Englebart divided the means to extend man’s intellect in the following classes: artefacts,
language, methodology and training. The last two refer to processes which support human
learning (training) and methods which help structure problems into manageable pieces
(methodology). Language refers more generally to the way in which the mind classifies the
world, and the manipulation of abstract concepts and artefacts belong to a world of material
things and their symbolic meanings. Engelbart further offers a computational understanding
of human cognition, focussing on symbolic knowledge representation. Artefacts become
a means of either providing human comfort or of supporting the manipulation of symbols.
Computers, for Engelbart, are examples of artefacts, invented to automate some of the
processes of symbol manipulation and furthermore to perform some of those manipulations
much more rapidly than humans are capable of.
9 Direct manipulation is a term commonly ascribed to Ben Shneiderman (1981: 143).
10 Dodge and Kitchin give a full account of many of these ‘info environments’ in the ‘Mapping
the Web’ section of An Atlas of Cyberspace (2001a: 73–152) and Johnson has described the
late nineties-era of interface metaphors as a ‘hysteria of total simulation’ (1999: 21).
11 See, for example, Anders (1998), Sparacino et al. (2000) and Tan et al. (2001).
12 Inspired by Memex, Nelson set about developing a system he called ‘hypertext’, which he
published in a 1965 paper entitled ‘The Hypertext’, delivered to the World Documentation
Federation Conference. He followed this with ‘Complex information processing: a file
structure for the complex, the changing and the indeterminate’, a paper delivered to the
Association for Computer Machinery (ACM) in the same year.
13 More details of this experiment can be found in Dade-Robertson (2007).

3 The spaces of information

1 See Anders (1994) for examples of exercises in mapping text-based MUDs.
2 See Ingram and Benford: (1995a, 1995b) and Ingram et al. (1996).
3 Chen (2004) gives an extensive account of the process of extracting meaning from text
corpuses of different sorts.
4 In addition to multidimensional scaling, alternative methods include pathfinder network
scaling (Schvaneveldt et al. 1989) for the aggregation of links in network topologies such
as hypertexts and citation networks between research documents. See Chen (1998) for an
overview. Further simplifications can be performed in the context of visualization, for example,
to strip out line crossings in graph-based visualization (Kosak et al. 1994), using forces and
physics models such as simulated annealing (Davidson and Harel 1996) or force direction

163
Notes

(Eades 1984) where the objects are forced to repel one another while being held together
by springs forcing some distance between unconnected nodes.
5 A very clear graphical representation of this can be found in the visual thesaurus at www.
visualthesaurus.com (accessed August 2010).
6 The topic of spatial frames of reference is broader than I have captured in this discussion and,
for further consultation, I have found Klatzky (1998) particularly useful.
7 See for example Downs and Stea (1974) and Kitchin and Freundschuh (2000).

4 Reality becomes display

1 Web 2.0 is seen as a social phenomenon and well-referenced examples of Web 2.0 analysis
include Chris Anderson’s The Long Tail (2009), which details the change in patterns of media
consumption and production brought about by the WWW. Other examples include Charles
Leadbeater’s We-Think (2009) and Clay Shirky’s Here Comes Everybody (2009), which both
look at the power of distributed collaboration on Web 2.0.
2 Available online at http://bokardo.com/archives/infoprefixation/ (accessed 2 September 2010).
3 The equivalent to the reading room would be the Google homepage on which considerable
design effort is expanded, keeping it simple and direct.
4 Examples of collected projects in this new domain of interactive architecture can be found in
Fox and Kemp’s Interactive Architecture (2009) and Lucy Bullivant’s edited AD issues 4Dspace
(2005), Responsive Environments (2006) and 4Dsocial (2007).
5 Possibly the best illustration of architecture as information visualization comes from the
pioneering firm NOX and specifically their project ‘Vision Machine’ for the Musée des Beaux
Arts and OEMs project for the Seattle Library.
6 Smart homes, for example, offer a vision of home appliances which will further the cause
of home automation (see Harper 2003). Criticisms of such approaches have been made by
design theorists such as Donald Norman (2007) who attacks the idea of home automation
with intelligence in his introduction to The Design of Future Things.
7 It is interesting to consider the activities of, for example, skateboarders in this regard who
often subvert highly strategic spaces by re-appropriating them for the performance of stunts
(see, for example, Borden 2001).
8 The Situationist movement developed artistic strategies to break the tyranny of strategic
space through, for example, the acts of random walking initiated by the dérive.

164
Bibliography

Aarseth, A. (1997) Cybertext: Perspectives on Ergodic Literature. London: Johns Hopkins

University Press.
Ahmed, S., Blessing, L. and Wallace, K. (1999) ‘The relationship between data, information and
knowledge based on a preliminary study of engineering designers’, ASME Design Engineering
Technical Conferences and Computers in Engineering Conference. Las Vegas, NV: pp. 7–12.
Alexander, C. (1964) Notes on the Synthesis of Form. Cambridge, MA: MIT Press.
Alexander, C. (1965) ‘A city is not a tree’, Architectural Forum, 122(1): 58–62.
Alexander, C. and Poyner, B. (1970) ‘The atoms of environmental structure’, in Moore, G. (ed.),
Emerging Methods in Environmental Design and Planning. Cambridge, MA: MIT Press,
pp. 308–21.
Anders, P. (1994) ‘The architecture of cyberspace’, Progressive Architecture, 75(10): 79–81.
Anders, P. (1998) Envisioning Cyberspace: Designing 3D Electronic Spaces. London: McGraw-Hill.
Anderson, C. (2007) The Long Tail: How Endless Choice is Creating Unlimited Demand. London:
Random House.
Bateson, G. (1988) Mind and Nature: A Necessary Unity. London: Bantam Books.
Beckmann, J. (ed.) (1998) The Virtual Dimension: Architecture, Representation and Crash Culture.
New York: Princeton Architectural Press.
Bederson, B. and Hollan, J. (1994) ‘Pad++: A zooming graphical interface for exploring alternative
interface physics’, Symposium on User Interface Software and Technology. Marina del Rey,
CA. ACM Press, pp. 17–26.
Bederson, B. and Shneiderman, B. (eds) (2002) The Craft of Information Visualization: Readings
and Reflections. San Francisco, CA: Morgan Kaufmann.
Benedikt, M. (ed.) (1991) Cyberspace First Steps. Cambridge, MA: MIT Press.
Bernstien, M. (1999) ‘Structural patterns and hypertext rhetoric’, ACM Computing Surveys, 31(4):
31–4.
Bernstien, M., Brown, M., Fisse, M., Glushko, R., Landow, G. and Zellweger, P. (1991) ‘Structure,
navigation and hypertext: The status of the navigation problem’, Hypertext. New York: ACM
Press, pp. 363–6.
Bewley, W., Roberts, T., Schroit, D. and Verplank, W. (1983) ‘Human factors in the design of
Xerox 8010 “Star” Office Workstation’, Computer Human Interaction. New York: ACM
Press, pp. 72–7.
Blackwell, A. (2006) ‘The reification of metaphor as a design tool’, ACM Transactions on Human
Computer Interaction, 13(4): 490–530.
Bolter, J. (2001) Writing Space: The Computer, Hypertext and the Remediation of Print. New
York: Erlbaum Associates.
Bolter, J. and Gromala, D. (2003) Windows and Mirrors: Interaction Design, Digital Art and the
Myth of Transparency. Cambridge, MA: MIT Press.
Bolter, J. and Grusin, R. (1999) Remediation: Understanding New Media. Cambridge, MA: MIT
Press.
Borden, I. (2001) Skateboarding, Space and the City: Architecture and the Body. Oxford: Berg
Publishers.
Borges, J. (1975) A Universal History of Infamy. London: Penguin.
Borges, J. (2000) ‘The library of Babel’. Labyrinths: Selected Stories and Other Writings. London:
Penguin, pp. 78–86.
Bibliography

Brooks, T. (1995) ‘Topical subject expertise and the semantic distance model of relevance
assessment’, Journal of Documentation, 51(4): 370–87.
Brown, P. (1989) ‘Do we need maps to navigate round hypertext documents?’, Electronic
Publishing – Organization, Dissemination and Design, 2(2): 91–100.
Bullivant, L. (ed.) (2005) 4Dspace: Interactive Architecture. London: Architectural Design.
Bullivant, L. (2006) Responsive Environments: Architecture, Art and Design. London: V&A
Publications.
Bullivant, L. (ed.) (2007) 4Dsocial: Interactive Design Environments. London: Architectural Design
Bush, V. (2003) ‘As we may think’, in Wardrip-Fruin, N. and Montfort, N. (eds) The New Media
Reader. Cambridge, MA: MIT Press, pp. 37–47.
Calvino, I. (1997) Invisible Cities. London: Vintage.
Card, S., Mackinlay, J. and Shneiderman, B. (eds) (1999) Information Visualization: Using Vision
to Think. London: Morgan Kaufmann.
Carruthers, M. and Ziolkoski, J. (eds) (2002) The Medieval Craft of Memory. Philadelphia, PA:
University of Pennsylvania Press.
Castells, M. (2000) The Rise of the Network Society. London: Blackwell Publishing.
Catledge, L. and Pitkow, J. (1995) ‘Characterizing browsing strategies in the World Wide Web’,
3rd International World Wide Web Conference. Darmstadt: pp. 1065–73.
Chabard, P. (2008) ‘Towers and globes: Architectural and epistemological differences between
Patrick Geddes’s outlook towers and Paul Otlet’s mundaneum’, in Rayward, W. B. (ed.),
European Modernism and the Information Society: Informing the Present and Understanding
the Past. London: Ashgate, pp. 105–26.
Chen, C. (1998) ‘Bridging the gap: The use of pathfinder networks in visual navigation’, Journal of
Visual Languages and Computing, 9(3): 267–86.
Chen, C. (2004) Information Vizualisation: Beyond the Horizon. London: Springer.
Collins, A. and Loftus, E. (1975) ‘A spreading-activation theory of semantic processing’,
Psychological Review, 82(6): 407–28.
Collins, A. and Quillian, M. (1969) ‘Retrieval time from semantic memory’, Journal of Verbal
Learning and Verbal Behaviour, 8(2): 240–7.
Cornell, E. and Hay, D. (1984) ‘Children’s acquisition of a route via different media’, Environment
and Behaviour, 16(5): 627–41.
Craglia, M. (2004) ‘Cogito ergo sum or non-cogito ergo digito? The Digital City Revisited’,
Environment and Planning B, 31(2): 3–4.
Crang, M. and Graham, S. (2007) ‘Sentient cities: Ambient intelligence and the politics of urban
space’, Information, Communication & Society, 10(6): 789–817.
Cuff, D. (2003) ‘Immanent domain: Pervasive computing and the public realm’, Journal of
Architectural Education, 57(1): 43–9.
Dade-Robertson, M. (2007) ‘Information Architecture in Screen Based Semantic Spaces’ (thesis).
Cambridge University.
Dade-Robertson, M. (2011) The Architecture of Information: Architecture, Interaction Design and
the Patterning of Digital Information. London: Routledge.
Davidson, R. and Harel, D. (1996) ‘Drawing graphs nicely using simulated annealing’, ACM
Transactions on Graphics, 15(4): 301–31.
de Certeau, M. (1983) The Practice of Everyday Life. Berkeley, CA: University of California Press.
Deleuze, G. and Guattari, F. (1988) A Thousand Plateaus. London: Athlone Press.
Dellarosa, D. (1988) ‘The psychological appeal of connectionism’, Behavioural and Brain Sciences,
11(1): 28–9.
Descartes, R. (2011) Discourse on Method and The Meditations, London: Penguin.
Dieberger, A. (1995) ‘Providing spatial navigation for the WWW’, Conference on Spatial Information
Theory. Semmering, Austria: pp. 21–3.
Dieberger, A. (1996) ‘Browsing the WWW by interacting with textual virtual environments:
A framework for experimenting with navigational metaphors’, Proc. of ACM Hypertext
’96. Washington, DC: pp. 170–9.

166
 Bibliography

Dieberger, A. (1997) ‘A city metaphor to support the navigation of complex information spaces’.
Proc. of COSIT’97. Laurel Highlands, PA: pp. 53–67.
Dillon, A., McKnight, C. and Richardson, J. (1993) ‘Space: The final chapter of why physical
representations are not semantic intentions’, in McKnight, C. (ed.), Hypertext and
Psychological Perspective. London: Ellis Horwood, pp. 169–91.
Dodge, M. and Kitchin, R. (2001a) Atlas of Cyberspace. London: Addison Wesley.
Dodge, M. and Kitchin, R. (2001b) Mapping Cyberspace. London: Routledge.
Dourish, P. (2004) Where the Action is: The Foundations of Embodied Interaction. Cambridge,
MA: MIT Press.
Dourish, P. (2006) ‘Re-space-ing place: “Place” and “space” ten years on’, Computer Supported
Collaborative Work. New York: ACM Press, pp. 299–308.
Downs, R. and Stea, D. (eds) (1974) Image and Environment: Cognitive Mapping and Spatial
Behaviour. London: E. Arnold.
Dunne, A. and Raby, F. (1995) ‘Fields and thresholds’, in Spiller, N. (ed.), Architects in Cyberspace.
London: Architectural Design.
Durand, D. and Kahn, P. (1998) ‘MAPA: A system for inducing and visualizing hierarchy in
websites’, Hypertext. New York: ACM Press, pp. 66–76.
Durkheim, E. and Mauss, M. (1963) Primitive Classification. Chicago: The University of Chicago Press.
Eades, P. (1984) ‘A heuristic for graph drawing’, Congressus Numerantium. Winnipeg, MB: Utilitas
Mathematica Publishing Inc., pp. 149–60.
Eco, U. (1995) The Search for the Perfect Language. London: Fontana Press.
Eco, U. (1997) ‘Function and Sign: Semiotics of Architecture’, in Leach, N. (ed.), The City and the
Sign: an Introduction to Urban Semiotics. London: Routledge, pp. 182–201.
Edwards, D. and Hardman, L. (1989) ‘‘‘Lost in hyperspace”: cognitive mapping and navigation in
a hypertext environment’, in McAleese, R. (ed.), Hypertext: Theory into Practice. London:
Intellect Books, pp. 90–106.
Engelbart, D. (1962) ‘Augmenting man’s intellect: A conceptual framework’. Summary Report
AFOSR-3233. Menlo Park, CA: Stanford Research Institute.
Floridi, L. (2010) Information: A Very Short Introduction. Oxford: Oxford University Press.
Folz, M. (1997) The JAIR Information Space. Cambridge, MA: MIT Press.
Fordor, J. (1981) Representations. Cambridge, MA: MIT Press.
Foucault, M. (1986) ‘Of other spaces’, Diacritics, 16(41): 22–7.
Foucault, M. (1997) ‘Utopias and heterotopias’, in Leach, N. (ed.), Rethinking Architecture:
A Reader in Cultural Theory. London: Routledge, pp. 350–5.
Fox, M. and Kemp, M. (2009) Interactive Architecture. Princeton, NJ: Princeton Architectural Press.
Frankl, P. (1969) Principles of Architectural History: the Four Phases of Architectural Style,
1420–1900. Cambridge, MA: MIT Press.
Franz, G., Mallot, H. and Wiener, J. (2005) ‘Graph-based models of space in architecture and
cognitive science: a comparative analysis’, 17th International Conference on Systems
Research, Informatics and Cybernetics (InterSymp). Baden Baden: pp. 30–6.
Friedenberg, J. and Silverman, G. (2005) Cognitive Science: An Introduction to the Study of Mind.
London: Sage.
Fry, B. (2008) Visualizing Data: Exploring and Explaining Data with the Processing Environment.
Sebastopol: O’Reilly.
Furnas, G. (1986) ‘Generalized fisheye views’, Computer Human Interaction. New York: ACM
Press, pp. 16–23.
Gardenfors, P. (2000) Conceptual Spaces: The Geometry of Thought. Cambridge, MA: MIT Press.
Gibson, W. (1986) Burning Chrome. London: Victor Gollancz.
Gibson, W. (1995) Neuromancer. London: Voyager.
Goldblum, N. (2001) The Brain-Shaped Mind: What the Brain Can Tell us about the Mind.
Cambridge: Cambridge University Press.
Goldstein, E. (2005) Cognitive Psychology: Connecting Mind, Research, and Everyday Experience.
Belmont: Wadsworth.

167
Bibliography

Guallart, V. (2009) Geologics: Geography Information Architecture. Barcelona: Actar.

Hanson, J. (1998) Decoding Homes and Houses. Cambridge: Cambridge University Press.
Harper, R. (ed.) (2003) Inside the Smart Home. Berlin: Springer.
Harrison, S. and Dourish, P. (1996) ‘Re-place-ing space: The roles of space and place in cooperative
systems’, Computer-Supported Collaborative Work. New York: ACM Press, pp. 67–76.
Henderson, D. and Card, S. (1987) ‘Rooms: The use of multiple virtual workspaces to reduce
space contention in a window-based graphical user interface’, ACM Transactions on Graphics,
5(3): 211–43.
Hendrix, J. (2003) Architectural Forms and Philosophical Structures. Bern: Peter Lang.
van den Heuvel, C. (2008) ‘Building society, constructing knowledge, weaving the web: Otlet’s
visualizations of a global information society and his concept of universal civilization’, in
Rayward, W. B. (ed.), European Modernism and the Information Society: Informing the
Present, Understanding the Past. London: Ashgate.
Hillier, B. (1996) Space is the Machine. Cambridge: Cambridge University Press.
Hillier, B. and Hanson, J. (1984) The Social Logic of Space. Cambridge: Cambridge University
Press.
Hillier, B. and Leaman, A. (1974) ‘How is design possible?’, Journal of Architectural and Planning
Research, 3(1): 4–11.
Hollier, D. (1992) Against Architecture: The Writings of Georges Bataille. Cambridge, MA: MIT
Press.
Hooper-Greenhill, E. (1992) Museums and the Shaping of Knowledge. London: Routledge.
Ingram, R. and Benford, S. (1995a) ‘Improving the legibility enhancement of virtual environments’,
2nd Eurographics Workshop on Virtual Environments. Monte Carlo: pp. 211–23.
Ingram, R. and Benford, S. (1995b) ‘Legibility enhancement for information visualisation’, 6th
Conference on Visualization IEEE. Washington, DC: pp. 209–16.
Ingram, R., Bowers, J. and Benford, S. (1996) ‘Building virtual cities: Applying urban planning
principles to the design of virtual environments’, Symposium on Virtual Reality Software and
Technology. Hong Kong: pp. 1–4.
Jackson, N. (2006) ‘The architectural view: Perspectives on communication’, Visual Communication
Quarterly, 13(1): 32–45.
Jencks, C. (2002) The New Paradigm in Architecture: The Language of Postmodernism. London:
Yale University Press.
Johnson, M. (1990) The Body in the Mind: The Bodily Basis of Meaning, Imagination and Reason.
Chicago: Chicago University Press.
Johnson, S. (1999) Interface Culture: How New Technology Transforms the Way We Create and
Communicate. London: Basic Books.
Kaplan, N. and Moulthorp, S. (1994) ‘Where no mind has gone before: Ontological design for
virtual spaces’, European Conference on Hypermedia Technology. Edinburgh: pp. 206–16.
Karatani, K. (1995) Architecture as Metaphor: Language, Number, Money. Cambridge, MA: MIT
Press.
Kitchin, R. and Freundschuh, S. (eds) (2000) Cognitive Mapping: Past Present and Future. London:
Routledge.
Klatzky, R. (1998) ‘Allocentric and egocentric spatial representations: definitions, distinctions, and
interconnections’, Spatial Cognition. Berlin: Springer, pp. 1–17.
Knight, T. (1994) Transformations in Design: A Formal Approach to Stylistic Change and Innovation
in the Visual Arts. Cambridge: Cambridge University Press.
Kosak, C., Marks, J. and Shieber, S. (1994) ‘Automating the layout of network diagrams with
specified visual organisation’, Transactions on Systems, Man and Cybernetics, 24(3): 440–54.
Kostof, S. (1995) A History of Architecture: Settings and Rituals. Oxford: Oxford University Press.
Kuhn, W. and Blumenthal, B. (1996) ‘Spatialization: Spatial metaphors for user interfaces’,
Computer Human Interaction. New York: ACM Press, pp. 346–7.
Kupier, B., Tecuci, D. and Stankiewicz, B. (2003) ‘The skeleton in the cognitive map’, Environment
and Behaviour, 35(1): 81–106.

168
 Bibliography

Lacour, C. B. (1996) Lines of Thought: Discourse, Architectonics and the Origin of Modern
Philosophy. Duke University Press: London.
Lakoff, G. (1990) Women, Fire and Dangerous Things: What Categories Reveal About the Mind.
Chicago: Chicago University Press.
Lakoff, G. and Johnson, M. (1980) Metaphors We Live By. Chicago: Chicago University Press.
Lakoff, G. and Johnson, M. (1999) Philosophy in the Flesh: The Embodied Mind and Its Challenge
to Western Thought. London: Basic Books.
Lan, J. (2008) ‘Smart space for office daily life: A situated life pattern approach’, International
Conference on Computer-Aided Architectural Design Research in Asia. Nanjing: pp. 168–73.
Landow, G. (1997) Hypertext 2.0: The Convergence of Contemporary Critical Theory and
Technology. Baltimore, MD: Johns Hopkins University Press.
Laurel, B. (1991) The Art of Human–Computer Interface Design. New York: Addison Wesley.
Lawrence, D. and Low, S. (1990) ‘The built environment and spatial form’, Annual Review of
Anthropology, 19: 453–505.
Leach, E. (1978) ‘Does Space Syntax really “constitute the social”?’ in Green, D. R., Haselgrove, C.
and Spriggs, M. (eds). Social Organization and Settlement: Contributions from Anthropology.
Oxford: British Archaeological Reports.
Leadbeater, C. (2009) We-Think: Mass Innovation, Not Mass Production. London: Profile Books.
Lefebvre, H. (1991) The Production of Space. Oxford: Blackwell.
Leibniz, G. (1896) New Essays Concerning Human Understanding. New York: Macmillan.
Levi-Strauss, C. (1963) Structural Anthropology. London: Basic Books.
Levi-Strauss, C. (1995) The Raw and the Cooked. London: Pimlico.
Lister, M. (2003) New Media: A Critical Introduction. London: Routledge.
Lynch, K. (1960) The Image of the City. Cambridge, MA: MIT Press.
Manovich, L. (2002) The Language of New Media. Cambridge, MA: MIT Press.
March, L. (1998) Architectonics of Humanism. London: John Wiley & Sons.
March, L. and Steadman, P. (1971) The Geometry of Environment: An Introduction to Spatial
Organisation in Design. London: RIBA Publications.
Markus, T. (1993) Buildings and Power: Freedom and Control in the Origin of Modern Building
Types. London: Routledge.
Mauss, M. (1979) Seasonal Variations of the Eskimo. London: Routledge.
McCullough, M. (1998) Abstracting Craft: The Practiced Digital Hand. Cambridge, MA: MIT Press.
McCullough, M. (2004) Digital Ground: Architecture, Pervasive Computing and Environmental
Knowing. Cambridge, MA: MIT Press.
McLuhan, M. (2001) Understanding Media. London: Routledge.
Merleau-Ponty, M. (2002) The Phenomenology of Perception. London: Routledge.
Mistry, P. and Maes, P. (2009) ‘SixthSense – A Wearable Gestural Interface’, SIGGRAPH Asia
Conference, available online at <http://poral.acm.org/citation.cfm?id=1667160&dl=ACM>
(accessed 17 March 2011).
Mitchell, W. (1995) ‘Soft cities’. Architects in Cyberspace. London: Architectural Design, pp. 8–13.
Mitchell, W. (1999) E-topia: Urban Life, Jim – But Not as We Know It. Cambridge, MA: MIT Press.
Mitchell, W. (1997) City of Bits: Space Place and the Infobahn. Cambridge, MA: MIT Press.
Mitchell, W. (2004) Me++: The Cyborg Self and the Networked City. Cambridge, MA: MIT Press.
Morgan, L. (1965) Houses and House-Life of the American Aborigines. Chicago: The University
of Chicago Press.
Morovec, H. and Elfes, A. (1985) ‘High-resolution maps from wide-angle sonar’, International
Conference on Robotics and Automation. IEEE. St Louis, MO: pp. 116–21.
Mukařovský, J. (1977) Structure, Sign and Function: Selected Essays by Jan Mukařovský. New
York: Yale University Press.
Munt, S. (2001) Technospaces: Inside the New Media. London: Continuum.
Murray, J. (1997) Hamlet on the Hollodeck. New York: Simon and Schuster.
Neisser, U. (1967) Cognitive Psychology. London: Prentice Hall.
Neisser, U. (1976) Cognition and Reality. London: W. H. Freeman.

169
Bibliography

Nelson, T. (1965) ‘A file structure for the complex, the changing and the indeterminate’,
Proceedings of the ACM National Conference. ACM, New York: pp. 84–100.
Nelson, T. (1965) ‘The hypertext’, World Document Federation Conference. Washington DC: p. 80.
Nelson, T. (1998) ‘The unfinished revolution and Xanadu’, available online at http://www.cs.brown.
edu/memex/ACM_HypertextTestbed/papers/64.html (accessed 31 January 2011).
Norman, D. (2007) The Design of Future Things. New York: Basic Books.
Novak, M. (1991) ‘Liquid architectures in cyberspace’, in Benedikt, M. (ed.). Cyberspace: First
Steps. Cambridge, MA: MIT Press, pp. 225–54.
Novak, M. (1995) ‘Transmitting architecture’, Architects in Cyberspace. London: Wiley & Sons.
O’Keefe, J. and Nadel, L. (1978) The Hippocampus as a Cognitive Map. Oxford: Oxford University
Press.
Osman, K. and Suliman, M. (1994) ‘The Space Syntax methodology: Fits and misfits’, Architecture
and Behaviour, 10(2): 189–204.
Ostler, T. (1994) ‘Architecture in cyberspace’, The Architects’ Journal, 21: 33–5.
Otlet, P. (1934) Traite de documentation. Le livre sur le livre. Theorie et pratique. Brussels: Palais
Mondial.
Pake, G. (1985) ‘Research at Xerox PARC: A founder’s assessment’, Spectrum, 22(10): 54–61.
Parunak, H. (1989) ‘Hypermedia topologies and user navigation’, Hypertext. New York: ACM
Press, pp. 43–50.
Paulini, M. and Schnabel, M. (2007) ‘Surfing the city: An architecture for context-aware urban
exploration’, MoMM: The Fifth International Conference on Advances in Mobile Computing
and Multimedia. Jakarta: pp. 31–40.
Peponis, J. (1997) ‘Geometries of Architectural Description: shape and spatial configuration’,
Space Syntax International Symposium. London: pp. 34.1–34.8.
Pérez-Gómez, A. and Pelletier, L. (2000) Architecture and the Perspective Hinge. Cambridge,
MA: MIT Press.
Pevsner, N. (1943) An Outline of European Architecture. London: Penguin.
Pirolli, J., Pitkow, J. and Rao, R. (1996) ‘Silk from a sows ear: Extracting usable structures for
the web’, Special Interest Group on Computer–Human Interaction. New York: ACM Press,
pp. 118–25.
Pountain, D. (2003) The Penguin Concise Dictionary of Computing. London: Penguin.
Preece, J. and Rogers, Y. (2002) Interaction Design: Beyond Human Computer Interaction. New
York: Wiley.
Preziosi, D. (1979) Architecture, Language and Meaning. The Hague: Mouton Publishers.
Psarra, S. (2009) Architecture and Narrative: The Formation of Space and Cultural Meaning.
London: Routledge.
Radford, G. (2003) ‘Trapped in our own discursive formations: Towards an archaeology of library
and information science’, The Library Quarterly, 73(1): 1–18.
Radin, P. (1923) ‘The Winnebago tribe’, Annual Report of the Bureau of American Ethnology.
Washington, DC: Smithsonian Institute.
Rapoport, A. (1982) The Meaning of the Built Environment. Tuscon, AZ: The University of Arizona
Press.
Raskin, J. (2000) The Human Interface: New Directions for the Design of Interactive Systems.
Boston, MA: Addison Wesley.
Ratti, C. (2004) ‘Space Syntax: Some inconsistencies’, Environment and Planning B, 31(4):
487–99.
Rheingold, H. (1985) Tools for Thought: The History and Future of Mind Expanding Technologies.
Cambridge, MA: MIT Press.
Robertson, G., Card, S. and Macinlay, J. (1993) ‘Information visualisation using 3D interactive
animation’, Communications of the ACM, 36(4): 57–71.
Robertson, G., Dantzich, M. V., Robbins, D., Czerwinski, M., Hinckley, K., Risden, K., Thiel, D.
and Gorokhovsky, V. (2000) ‘Task gallery: A 3D window manager’, Special Interest Group on
Computer–Human Interaction. New York: ACM Press, pp. 494–501.

170
 Bibliography

Rohrer, T. (2001) ‘Pragmatism, ideology and embodiment: William James and the philosophical
foundations of cognitive linguistics’, in Dirven, R., Hawkins, B. and Sandikcioglu, E. (eds), Language
and Ideology: Cognitive Theoretical Approaches. Amsterdam: John Benjamins, pp. 49–82.
Rosenfeld, L. and Morville, P. (2002) Information Architecture for the World Wide Web. Sebastopol:
O’Reilly.
Rossello, M. (1994) ‘The screener’s maps: Michel de Carteau’s “Wandersmanner” and Paul
Auster’s Hypertextual Detective’, in Landow, G. (ed.), Hyper/Text/Theory. Baltimore, MD:
Johns Hopkins University, pp. 121–58.
Rossi, P. (2000) Logic and the Art of Memory: A Quest for a Universal Language. London: The
Athlone Press.
Rouet, J. (1996) ‘An introduction to hypertext and cognition’, in Rouet, J., Levonen, J. J., Dillon, A.
and Spiro, R. J. (eds), Hypertext and Cognition. Mahwah, NJ: Lawrence Erlbaum Associates,
pp. 3–8.
Rucker, R., Sirius, R. and Mu, Q. (eds) (1993) Mondo 2000: A Users Guide to the New Edge.
London: Thames and Hudson.
Saussure, F. D. (1983) Course in General Linguistics. London: Duckworth.
Schnadelbach, H. (2009) ‘Visibility in architecture extended through audiovisual communication
technologies’, Space Syntax Symposium. Stockholm: pp. 97:1–97:13.
Schölkopf, B. and Mallot, A. (1995) ‘View-based cognitive mapping and path planning’, Adaptive
Behaviour, 3(3): 311–48.
Schvaneveldt, R., Durso, F. and Dearholt, D. (1989) ‘Network structures in proximity data’, The
Psychology of Learning and Motivation. New York, pp. 249–84.
Shirky, C. (2009) Here Comes Everybody: How Change Happens when People Come Together.
London: Penguin.
Shneiderman, B. (1981) ‘Direct manipulation: A step beyond programming languages’, SIGSOC.
New York: ACM Press, pp. 461–7.
Shneiderman, B. (2003) ‘Why not make user interfaces better than 3D virtual reality?’, Computer
Graphics and Applications, 23(6): 12–15.
Shneiderman, B. and Maes, P. (1997) ‘Direct manipulation vs. interface agents’, Interactions,
4(6): 42–61.
Siegel, A. and White, S. (1975) ‘The development of spatial representations of large-scale
environments’, in Reese, H. W. (ed.), Advances in Child Development and Behavior, vol. 10.
New York: Academic Press, pp. 9–55.
Small, J. P. (1997) Wax Tablets of the Mind: Cognitive Studies or Memory and Literacy in Classical
Antiquity. London: Routledge.
Snow, C. (1993) The Two Cultures. Cambridge: Cambridge University Press.
Snyder, I. (1997) Hypertext: The Electronic Labyrinth. New York: New York University Press.
Sorabji, R. (2004) Aristotle on Memory. London: Duckworth.
Sowa, J. (2000) Knowledge Representation: Logical, Philosophical and Computational Foundations.
Pacific Grove, CA: Brooks/Cole.
Sparacino, F., Davenport, G. and Pentland, A. (2000) ‘Wearable cinema/wearable city: bridging
physical and virtual spaces through wearable computing’, available online at http://alumni.
media.mit.edu/~flavia/Papers/flavia_imagina00.pdf (accessed 31 January 2011).
Spence, R. (2001) Information Visualisation. London: ACM Press.
Spiller, N. (ed.) (1995) Architects in Cyberspace. London: Architectural Design.
Spiller, N. (ed.) (1998a) Architects in Cyberspace II. London: Architectural Design.
Spiller, N. (1998b) Digital Dreams: Architecture and the New Alchemic Technologies. London:
Watson-Guptill.
Spiller, N. (ed.) (2002) Reflexive Architecture. London: Architectural Design.
Steadman, P. (1976) ‘Graph-theoretic representation of architectural arrangement’, in March, L.
(ed.), The Architecture of Form. Cambridge: Cambridge University Press, pp. 94–115.
Steadman, P. (1983) Architectural Morphology: An Introduction to the Geometry of Building Plans.
London: Pion.

171
Bibliography

Stephenson, N. (1993) Snow Crash. London: ROK.

Stotts, P. and Furuta, R. (1989) ‘Petri-net based hypertext: Document structure with browsing
semantics’, ACM Transactions on Office Information Systems, 7(1): 23–9.
Suchman, L. (1987) Plans and Situated Actions. Cambridge: Cambridge University Press.
Summerson, J. (1980) The Classical Language of Architecture. London: Thames and Hudson.
Talmy, L. (1983) ‘How language structures space’, in Pick, H. (ed.), Spatial Orientation: Theory,
Research and Application. New York: Plenum, pp. 225–82.
Tan, D., Syefanucci, J., Proffitt, D. and Pausch, R. (2001) ‘The infocockpit: providing location and
place to aid human memory’, Perceptual/Perceptive User Interfaces. New York: ACM Press,
pp. 1–4.
Thrift, N. (2004) ‘Remembering the technological unconscious by foregrounding the knowledges
of position’, Environment and Planning D, 22(1): 175–90.
Tolman, E. (1948) ‘Cognitive maps in rats and men’, The Psychological Review, 55(4): 189–208.
Tufte, E. (2001) The Visual Display of Quantitative Information. Cheshire: Graphics Press.
Tversky, B. (2001) ‘Structures of mental spaces’, Third International Space Syntax Symposium.
Atlanta, GA: pp. 12.1–12.5.
Tversky, B. and Lee, P. (1998) ‘How space structures language’, in Freksa, C. (ed.), Spatial
Cognition: An Interdisciplinary Approach to Representing and Processing Spatial Knowledge.
Berlin: Springer-Verlag, pp. 175–6.
Van Dijck, P. (2003) Information Architecture for Designers: A Guide to Structuring Websites for
Business Success. London: Rotovision.
Varela, F., Thompson, E. and Rosch, E. (1993) The Embodied Mind. Cambridge, MA: MIT Press.
Walker, J. (1990) ‘Through the looking glass’, in Laurel, B. (ed.), The Art of Human–Computer
Interface Design. London: Addison Wesley, pp. 439–48.
Wardrip-Fruin, N. and Montfort, N. (eds) (2003) The New Media Reader. Cambridge, MA: MIT
Press.
Ware, C. (2004) Information Visualisation: Perception for Design. London: Morgan Kaufman.
Weinberger, D. (2007) Everything is Miscellaneous: The Power of the New Digital Disorder.
London: Henry Holt and Company.
Weiser, M. (1991) ‘The computer for the 21st century’, Scientific America, 265(3): 94–104.
Wertheim, M. (1999) The Pearly Gates of Cyberspace. London: Virago Press.
Wexelblat, A. (1991) ‘Giving meaning to place: semantic spaces’, in Benedikt, M. (ed.), Cyberspace:
First Steps. Cambridge, MA: MIT Press.
Widdows, D. (2004) Geometry and Meaning. Stanford: CSLI.
Wigley, M. (1995) The Architecture of Deconstruction: Derrida’s Haunt. Cambridge, MA: MIT
Press.
Williamson, S. and Shneiderman, S. (1992) ‘The dynamic home finder: evaluating dynamic queries
in a real-estate information exploration system’, Special Interest Group on Information
Retrieval. New York: ACM Press, pp. 339–46.
Wise, J., Thomas, K., Pennock, D., Lantrip, M., Pottier, M., Schur, A. and Crow, V. (1995) ‘Visualizing
the non-visual: spatial analysis and interaction with information from text documents’,
Symposium on Information Visualisation, IEEE. Atlanta, GA: pp. 51–8.
Wodtke, C. (2003) Information Architecture: Blueprints for the Web. London: New Rides.
Woolley, B. (1993) Virtual Worlds. London: Penguin.
Wright, A. (2007) Glut: Mastering Information throughout the Ages. London: Cornell University
Press.
Wurman, R. S. (1996) Information Architects. London: Watson-Guptill Publications.
Yates, F. (2001) The Art of Memory. London: Pimlico.
Zellweger, P. (1989) ‘Scripted documents: A hypermedia path mechanism’, Hypertext ’89. New
York: ACM Press, pp. 1–14.

172
Index

Aarseth, A. 87 artificial intelligence 18, 20, 37, 47

abstract concepts 31, 38, 48, 53, 55, 67, 114, associations 18, 22–3, 44, 50, 63, 77, 83, 99,
121, 123–4 102–3, 105, 111, 119–20, 124; semantic
adjacencies 74, 87, 97, 119, 121, 133 62, 78, 82, 132–3, 138, 154
agency 124, 130–1, 135, 137, 139–40, 142,
148 ballast of materiality 6, 11, 67–8, 91, 93, 132,
agents 45, 47, 127–8, 143 144
Ahmed, S. 18, 20 Beckmann, J. 12
AI see artificial intelligence Bederson, B. 29
Alexander, C. 24, 83, 129 Benedikt, M. 9–11, 67
allocentric 111, 115 Berners-Lee, T. 76, 78, 82–3
applications 11, 17, 29, 80, 120, 127, 136, Bernstien, M. 84
140, 144 Blumenthal, B. 81
architectonic space 34, 89, 96, 115–16, Bolter, J. 15, 32–3, 87–90
130–1; system 33–65, 67, 76, 78–9, Borges, J. 3, 27, 93
89–91, 119–21, 123–4, 126, 130–2, 138, Brown, P. 84
140, 143–4, 147–50 Bush, V. 69–70, 76–8, 90, 133, 147, 149
architectural analysis 21–2, 39; approaches
21, 44; design 12–13, 15, 20, 129, 150; Camillo Delmino, G. 3–5, 59
discourse 7, 12, 15, 41, 138; elements Card, S. 29, 31, 81
23, 83; form 22, 27, 39, 41, 43, 45, 57; Cartesian dualism 47; space 88, 97
languages 27, 36, 38; metaphors 17, 37, Castells, M. 129
61, 65, 68, 90–1, 93–4, 96, 119, 138, categorization 33, 39, 45, 47–8, 53, 56, 64,
147; 15, 17, 21, 27, 30, 36, 38, 45, 57, 67, 71, 79
61–2, 74, 94, 120, 145, 149; theories 16, Certeau, M. de 87, 140
20–1, 24, 27–8, 38, 44, 46, 101 Chabard, P. 75
architecture: cognitive 36; of cyberspace change 67, 69, 81, 93, 96, 103, 126, 129,
11, 13, 150; digital 12, 133; of 148, 150; and input 16, 33; and language
information 13, 15, 96; liquid 10; 36, 39–40; mechanical 33, 74; onscreen
mnemonic 57, 62–3, 120, 147; physical 16, 30; and output 96, 107; radical 14,
57, 64, 143; virtual 10, 17; see also 144; and semantics 50; in viewpoint
information architecture 114–15
archive 1, 3–4, 6, 33, 57, 69, 75–6; universal Chen, C. 29
3, 33, 69, 75–6 Cicero 59, 64
Aristotle 42, 48, 60, 62, 64–5 classification 5–6, 21, 37, 41–3, 46, 53, 57,
ars combinatoria 60, 71–3, 78, 147 63, 65, 76, 137–8; primitive 37, 63
art of memory 38, 57, 59–60, 68, 71, 78 closeness 62, 105–6, 109–10, 121
artefacts 20, 59, 110, 141, 150 code 77, 127
articulation 21, 40–1, 45–6, 56, 98, 100, 120, cognition 28, 32, 53, 57, 67, 77, 79, 111,
133, 144 121, 151
Index

cognitive collage 32, 118–19, 126; map 30–2, design 9–10, 12, 14, 17, 21, 24, 28, 33, 83–4,
69, 98–101, 116–20; mapping research 149–51; content 150; of digital information
116–17; psychology 28, 31, 37, 46–7, 63, systems 15, 124, 145, 150; disciplines
116; structures 37, 54 115; paradigms 81, 150; process 129
Collins, A. 49 designers 7, 9, 14, 21, 35, 99, 148, 150;
combinations 14, 68, 71–3, 78, 83, 125, architectural 12, 15; graphic 13
135 determinism 12, 141
communication 11, 13, 22, 32, 35, 40–1, Dewey Decimal System 5–6
67–8, 88, 120–1, 126, 132, 148 Dieberger, A. 98–9, 106, 119
computation 15–16, 48, 68, 70–1, 78, 90, digital artefact 79, 150; information 7, 12,
124–5, 130–1, 138, 141, 144–5, 148 16–17, 70, 81, 91, 93–4, 96–7, 101–2,
computational agents 135, 139, 141–2, 144; 116, 120, 130, 132, 140, 142; information
devices 16, 115, 129, 140; information systems 15, 17, 34, 69–70, 78–80, 87,
70, 91, 101, 123, 130, 137, 144, 147; 93, 121, 123–4, 145, 150; media 28, 33,
representations 18, 123, 143; systems 70, 78–9, 87–8, 90, 97; representations
33, 47, 79, 95, 123–4, 129, 131, 139, 12, 116, 134; technologies 6, 11–12,
144, 148–9 14–16, 20, 32, 38, 65, 67–8, 71, 78, 87,
computer interfaces 28, 80, 128; systems 90, 124, 132–3
17, 69, 137, 139, 142, 147, 150; user Dillon, A. 89, 102
interfaces 17, 113, 149–50 direct manipulation 70, 113, 128, 130, 134–5,
computer-based information 16, 34, 81, 94, 137, 144, 150
128, 138 discourse 1, 4, 13, 17, 20–1, 23, 44–5, 60;
computing 12, 16, 20, 28, 32, 34, 69–70, 78, architectural 7, 12, 15–16, 32, 38, 40–1,
80–1, 128–9, 138–9, 144; embodied 28– 45, 70, 138; on cyberspace 7, 13; design
9, 96, 149; personal 70, 80–1; pervasive 139; and knowledge 6; mathematical 24;
12, 16, 124, 126, 130–1, 151 philosophical 61, 64, 71; speculative 16
conceptual level 36, 52–4, 56; representation districts 99–100
38, 47, 51, 53; spaces 38, 46, 48, 53–4, Dodge, M. 9, 30, 97
56–7, 63, 67, 147 domains 15, 22, 37, 47, 53, 55, 63, 149–50
conceptualization 93, 101–2, 105–7, 120 Dourish, P. 130, 140–2
configuration 4, 21–3, 36, 39, 80, 119, 133 dual organizations 43, 45
connectivity 26, 101, 105–6, 115 Dunne, A. 11
constraints 8, 55–6, 60, 62–3, 70, 100, 116, Durkheim, E. 37, 41–2, 46, 63
123, 139; combinatorial 73, 82
containers 55–6, 62, 102, 105–6, 109, 114 Eco, U. 22, 71–2, 74
containment 101, 105–6, 110, 115; physical edges 26, 97–101
55 embodied relationships 53, 109–10, 148;
context 11–12, 17–18, 20–2, 27–8, 38–9, schemas 121
70–1, 82–3, 96–7, 99, 126–7, 129–35, encoded information 27
137–44, 148–9 Engelbart, D. 79–80, 149
Cornell, E. 117 environment: large-scale 26, 32; physical 12,
Coyne, R. 20 15–16, 38, 46, 59, 89; real 27, 97, 130;
Crang, M. 131, 142–3 screen-based 30, 32, 134
cyberspace 6–7, 9–13, 15, 29–30, 68, 93, environmental structures 24, 39–40, 100
101, 128–9, 131, 150; architectural 12;
architects 10 files 17, 79–80, 82, 109, 135
Foucault, M. 3–4
data 7, 10–11, 14, 18–20, 29–31, 67, 77, 84, Fry, B. 103
95, 99, 101–3, 105, 142 functions 21–3, 27, 35, 47, 53, 57, 63, 78, 80,
databases 29, 101–2, 105, 134 101, 125, 129, 132, 137, 139, 141
Dellarosa, D. 50
dematerialized information 6–7, 15, 68, 70 Gardenfors, P. 32, 38, 47–8, 50–4, 56, 63
Descartes, R. 47, 61, 65, 111 geometric lock 106–7

174
 Index

geometry 22, 44, 83, 89, 102, 106, 115, 118, information space 9, 14, 17, 30, 34, 70,
148 82–3, 91, 93–103, 105–21, 126–7, 132,
Gibson, William 7, 9–10, 68 147–8; complex 103; formless 138;
Gibsonian cyberspace 7–11 multidimensional 89; structureless 94;
Graham, S. 131, 142–3 visual 135
graphical user interface 16–17, 30, 59, 70, information systems 72, 77, 100, 120;
79–82, 84, 90, 93, 96–7, 128, 134, 150 complex 115; computational 33, 79, 123;
Gromala, D. 32–3, 87 digital 15, 17, 34, 69, 87, 121, 123–4,
Grusin, R. 15 126, 150; emergent 103, 126; intuitive
Guallart, V. 138 17; networked 149; screen-based 33;
GUI see graphical user interface topological 116
interaction design 20–1, 28, 96, 150; human
Hanson, J. 25, 45 68–9, 79, 90, 93, 150; human-computer
Harrison, S. 140 16, 21, 28, 31, 70, 80, 93, 150–1; mode
Hay, D. 117 of 114, 127–8, 148; physical 27, 38,
HCI (human-computer interaction) see 55–6, 110; process of 110, 123, 134;
interaction, human-computer screen-based 16, 96, 111; space 34, 81,
head-mounted displays (HMDs) 9, 16, 115, 94–6, 110, 115–16, 120, 123, 132, 148;
131 web-based 126–7
hierarchy 42, 48–50, 79, 83–5, 88, 102, 106 interactor 111, 116, 124, 128, 137, 149
Hillier, B. 24–5, 41, 44–6, 129 interfaces 17, 28, 68, 79–80, 90, 97, 138,
hypercubes 84–5 142, 147 see also user interfaces
hypermedia 33–4, 89 internet 79, 82–3, 128
hyperspace 13, 82, 97 IR see information retrieval
hypertext 16–17, 70, 79–80, 82–5, 87–8,
90, 95, 97–101, 103, 114–16, 119–21; James 2, 111–12
documents 17, 83–4, 90, 115; links 95, Johnson, M. 31–2, 38, 54–6, 64, 97, 114,
103, 116; page 101, 115, 119; patterns 147–8
84–5, 95; research 101, 119; systems 70,
84, 87, 98; topologies 70, 84, 100 Kaplan, N. 89, 95
hypertexts: chunk-based 82; tree-structured Karatani, K. 37, 61
84 keyboard 16, 77, 80, 96, 107, 138
Kierkergaard, S. 61
IA see information architecture kitchen 30, 138
icons 80–1, 91, 138 Kitchin, R. 9, 30, 97
image schemas 55–7, 114, 147 knowledge 1, 3–4, 6, 18, 20, 22, 37–8, 46–7,
images 10, 29–32, 58–60, 62, 88, 94, 96, 57, 63, 69, 71–2, 76–7, 116–17, 120;
98–9, 142 conceptual level of 53, 56; representation
inference 27, 47–9, 51, 78, 142 18, 33, 37–8, 46–7, 53–4, 56, 63, 67, 70,
information architects 13–14, 75, 84, 126–7; 77, 120; symbolic 51
architecture 13–15, 33, 38, 76, 119–20, Kuhn, W. 81
123, 126–7, 133, 150; content 22, 36; Kupier, B. 118
distribution of 79, 90; environments 17,
99, 101; filtering 139, 141; objects 7, 18, Lakoff, G. 31–2, 38, 54–5, 64, 147–8
20, 71–2, 79, 81–4, 90, 95, 97, 101–3, Lan, J. 139–40
105–10, 123, 128, 130; representation landmarks 99–100, 117
of 29, 72, 74, 78, 81, 83, 90, 101, 120, Landow, G. 87
147; retrieval 14, 29, 96, 103; rretrieval language 22, 27, 33, 35–41, 45–6, 53–4, 56,
algorithms 108; separation of 6, 34, 63–5, 71–2, 74–5, 88, 114, 118–19, 140;
101; society 75–6; virtual 101, 144, analogy 27, 36, 38–9
148; structures 17–18, 101, 115, 150; Le Corbusier 75–6, 90
visualization 28–32, 96, 98, 100–1, 110, Leaman, A. 24
115, 119, 134–5, 138 Lee, P. 119

175
Index

Lefebvre, H. 23 metaphor of 97, 101, 114, 119; process

Leibniz, G. 71 of 62, 97, 114–15, 121
Levi-Strauss C. 37, 41, 43–4, 46 navigational schema 115–16, 121, 123,
library 1, 3–4, 6, 14, 20, 30, 34, 38, 57, 65, 68, 128–30, 132, 148–9
71–2, 77, 127, 133 Neisser, U. 31
Library of Babel 1–3, 27, 34, 72 Nelson, T. 81–2
linguistic system 33, 39–41, 45, 53, 63, 65, network 25, 46, 51, 82, 88, 101, 119, 128,
123–4 131–3, 138; neural 50–1; semantic 48,
linguistics 28, 38, 63 50, 56, 62, 102, 104
Lister, M. 32 neurons 51, 63
locations 1, 17, 30, 62, 97–8, 102, 106, 108, new media 32, 87, 89–90
111, 115, 133–4, 138–40, 142; physical nodes 25, 48, 50, 62, 64, 82, 84–5, 98–101
64, 83, 101, 130; virtual 130 non-space 7, 10, 13, 144
logic 47, 60, 68, 70–1, 80, 90 Novak, M. 10–11
Lull, R. 33, 68, 72–4, 77, 90, 123, 147
Lynch, K. 32, 94, 98–101, 115, 117, 119 O’Keefe, J. 111, 117
ontology 60, 71
McCullough, M. 7, 12, 20, 29, 67 organization 3–6, 9, 14–15, 17, 20, 22, 37, 40,
McLuhan, M. 32 43, 56–7, 60, 67, 69, 75, 123; fixed 71–2;
Maes, P. 134–5, 137 physical 4–6, 15, 68, 124, 138
mapping 55, 64, 93, 106–11, 113, 131, 142 orientation 44, 63, 112, 117, 142
March, L. 24 Otlet, P. 33, 69, 74–6, 90, 147
Markus, T. 4, 21–3, 27, 71, 102
Marx, K. 6 paradigms 16, 29, 47, 70, 79, 81, 90, 113–14,
materiality 6, 11, 67–8, 91, 119, 132 126, 135, 139, 150
Mauss, M. 37, 41–2, 46, 63 paths 15, 51–3, 62, 84, 88, 98–100
Memex 69–70, 77–8, 90, 133, 149 patterning 6, 38–9, 90, 120, 138, 147
memory 1, 3–5, 18, 32–4, 38, 47, 57–60, patterns 15, 21, 23, 36–7, 39, 43, 50–1,
69, 72, 99, 116, 118–20; environmental 54–5, 57, 60–1, 84, 90, 123–4, 133;
117–18; structuring of 57, 59 combinatorial 27, 71–2, 123; structural
metaphorical projections 55, 62, 97 43
metaphors 7, 9, 12–13, 15, 17, 34, 53–7, PCs see personal computers
64–5, 74–5, 96–7, 114–17, 120–1, 138, perception 65, 99, 111, 114, 118, 130–2, 148
147–9; navigational 97, 121, 148; visual permutations 33, 68, 71–3, 77–8
70, 81, 87 personal computers 6, 16–17, 68, 81, 93,
method of loci 33, 38, 58–60, 62, 64, 74, 77, 110, 115, 128–9, 147, 149
81, 89–90, 147 physical manifestations 29, 101, 124, 144
Mitchell, W. 6, 11, 14, 20 physical objects 3, 16, 40, 64, 70, 74, 99, 120,
models 37, 46–9, 53, 56, 68, 147, 151; 123, 130, 132, 134, 138, 142, 151
cognitive 116, 149–50; connectionist 51; physical spaces 9, 11–12, 14–15, 37–8, 43,
semantic distance 104–5 57, 59–60, 62, 68, 71, 105, 111, 129,
Morville, P. 13–14 132, 134; structure 18, 82; world 6, 37,
Moulthorp, S. 89, 95 40, 43, 56–7, 64–5, 67, 91, 96–7, 105,
mouse 16, 78, 80, 96, 107, 110, 115, 128, 127, 129–30, 138–9, 144
135 planarity 62–3
MUDs (multi–user domains/dungeons) 11, 98 Plato 42, 61
Mukarovský, J. 35 Poincare, H. 111
museums 4, 6, 20, 38, 57, 65, 68, 81, 120, Popper, K. 67
139 Porter, J. 127
Preziosi, D. 40–1, 45, 63, 120, 147
Nadel, L. 111, 117 programming 18, 80, 129–30, 141; languages
navigation 1, 3, 14, 30–1, 34, 84–5, 89, 80
94, 96–8, 114–18, 121, 126–8, 148; Psarra, S. 27, 57

176
 Index

query 103, 108, 126–8 multidimensional 120, 148; neutral 33,

Quillian, M. 49 87, 125; perception of 27, 111; principled
108; real 7, 93, 97–8, 116, 130, 148;
Raby, F. 11 relative 108–10, 115; topographical 88,
Radin, P. 43 110; topological descriptions of 24, 46;
recommender systems 135, 142–3 topological representations of 44, 110,
reference 11, 17, 31, 38, 42–3, 58, 60, 83, 119; transduced 131, 142–3, 148
112, 128, 133, 139, 150; allocentric Space Syntax 23–5, 27, 39, 44
frame of 112–13, 151; egocentric frames spatial cells 23, 26–7, 46; environment 33, 36;
of 114–15, 151 memory 32, 119; metaphors 31–2, 67,
relationships: meaningful 33, 108, 123; 93, 116, 123, 138, 145, 147; organization
pattern of 39, 84; social 24, 83; space 38, 44, 58, 79, 98, 116, 148; practices
action 141; structural 75 87, 141–2, 144; relationships 31, 45,
relative positions 62, 110, 112–13 62–4, 74, 105, 114; representations 30,
relative relationships 109–10 32, 97, 108, 110, 114, 120; structure
remediation 15, 32–3, 70, 87–8, 90 89, 102
representations 12–15, 17, 25–7, 31–2, 38, spatiality 67, 88, 90
51, 53, 56, 67, 70–2, 80–1, 96, 119–21, spatialization 29, 81
126, 147; of information 71, 143, 147; Spence, R. 29
internal 30, 102; observer-independent Spiller, N. 11–12
116; onscreen 30, 89; schematic 53, 60; Steadman, P. 24, 63
sub-conceptual 50, 53, 63; textual 87, 89; Stephenson, N. 7–8
virtual 32, 98; visual 10, 89, 97, 111 strategic spaces 141–2, 144
restrictions 4, 33, 62, 68, 72, 74, 81, 121, 147 structures 1, 4–5, 18, 24, 36–7, 39, 48–57,
Rosenfeld, L. 13–14 83, 94–5, 98–9, 101–3, 120, 148–9
Rossello, M. 87 Suchman, L. 28
SUIs see single-user interfaces
Saussure, F. 36 symbolic representations 48–51, 53–4, 56, 63
scale 22, 40, 71 symbols 1, 9, 45, 89
schemas 31, 55, 60, 62, 114, 121, 147 syntax 24, 27, 36, 72, 103, 140
screen 14, 16, 28, 30, 32, 68, 76–7, 88–9, 95,
97, 100, 107–8, 111, 115, 131–2, 148 Talmy, L. 118–19
screen space 30–1, 34, 81, 88, 94–6, 106, tangible user interfaces 150
110, 115–16, 128, 132, 147 technologies 6–7, 11, 16, 21, 28, 39–40, 68,
search engine 103, 107–8, 126 78, 80, 83, 126, 130–1, 135, 139, 141–2
semantic space 34, 83, 89, 93–6, 101–3, text 18, 21, 29, 32–3, 38, 60, 63, 70–1, 87–8,
105–8, 110, 115, 119–21, 127–8, 130, 90, 107, 133; documents 105
132–4, 147–9 textual virtual environments 98
semantics 27, 38, 89, 108–9, 114, 123, 140 Theatre of Memory 1, 3–5, 34, 59, 61, 69, 72
shaping 4, 37, 141, 144–5 Tolman, E. 32, 116–17
Shneiderman, B. 29, 97, 108, 134–5 tools 21, 32, 40–1, 46, 63, 80, 103, 110, 124
single-user interfaces 17 topic 16, 21, 28, 33, 38, 57, 60–2, 64, 69–70,
smart space 130, 139 77, 85, 88–9, 120, 128
Snyder, I. 87 topography 88, 101
social space 37, 42, 63, 65, 142, 147; topological relationships 27, 101, 105–6, 119,
structure 23–4, 27, 36, 41, 43–4, 46 126; representations 24–5, 120; space
software 11, 18, 113, 130, 135, 142, 144; 73, 89, 102, 111, 114–15, 119; structures
agents 130, 134–5, 138 27, 82; systems 114–15
space: absolute 108–11, 113; articulation of topology 25, 27, 34, 44, 62–3, 84, 99–101,
37, 45, 65, 120, 125; augmented 131, 103, 105, 118–19, 140, 148
142, 148; bounded 55, 114; constraints topos 38, 57, 60–2, 64, 70, 77, 85, 88–9, 120,
of 4, 70, 144; experience of 139, 142, 144, 147
144; metaphor of 68, 97, 123, 148; tree 65, 82–3, 98, 111–13, 133

177
Index

TUIs see tangible user interfaces Walker, J. 9, 62

Turing, A. 78 Ware, C. 29–30
TVEs see textual virtual environments web 13, 30, 70, 82–3, 97–101, 126–8, 131,
Tversky, B. 32, 114, 118–19 138, 149
typologies 4, 6, 35, 38; building 23, 38, 57, Web 2.0 127–8
65, 68 web-based information 30, 126–7
web pages 79, 84, 96–7, 115–16, 126
ubiquitous computing 16, 34, 126, 129–31, websites 13, 17, 84–6, 97, 101, 107, 119,
138, 144, 148 148
user interfaces 9, 12, 81, 114, 150–1 see also Weiser, M. 130–1
interfaces Weinberger, D. 5–6
Wise, J. 29
virtual environments 7, 16, 28–30; abstract World Wide Web 6, 13, 16–17, 70, 76, 78–9,
32; immersive 9; real–time 98 82–3, 88, 91, 93–4, 101, 147
virtual reality 7, 16, 120, 129, 131 Wurman, R. 13
virtual spaces 7, 68, 88; implied continuous WWW see World Wide Web
98; undeclared 34
visual display 107, 110–11, 115, 129; space Xerox 80–1, 97
82, 88, 108, 111, 114, 116, 119–20, 123,
127, 134, 147–8 Yates, F. 1, 3–4, 38, 57–60
visualizations 15, 17, 29–30, 48, 68, 109, 133
VR see virtual reality Zellweger, P. 84

178