Seeing Like an Interface

Paul Dourish
Donald Bren School of Information and Computer Sciences
University of California, Irvine
Irvine, CA 92697-3440, USA
jpd@ics.uci.edu
ABSTRACT attempting to uncover the ways in which aspects of social life may
Mobile and ubiquitous computing systems are increasingly of be transformed by the arrival of new forms of technology and
interest to HCI researchers. Often, this has meant considering the practice. While technologies clearly do have impacts, this
ways in which we might migrate desktop applications and approach tends to regard technology as inevitable and to fail to
everyday usage scenarios to mobile devices and mobile contexts. acknowledge the range of relationships between technology and
However, we do not just experience technologies in situ – we also social life. A second approach, by contrast, is to examine the
experience everyday settings through the technologies we have at “social shaping” of technologies, looking at the ways in which
our disposal. Drawing on anthropological research, I outline an technological and design decisions are made within social,
alternative way of thinking about the relationship between organizational, and economic circumstances that collectively
technology and “seeing” everyday life and everyday space. shape the technologies that emerge. Subsuming both of these, to
some extent, my concern is with information technologies as sites
of social and cultural production. Social and cultural realities are
Categories and Subject Descriptors no simply givens; they are performed, enacted, and reproduced in
H.5.0 [Information Systems and Presentation]: General the course of everyday life. I want to examine information systems
here as one particular site where these activities take place. I find
General Terms this position particularly instructive for design, because it
Measurement, Documentation, Design, Economics, provides not only an analytic perspective but one that encourages
Experimentation, Security, Human Factors, Standardization, intervention and transformation.
Theory, Legal Aspects.
1.1 The Case of Mobile Computing
As a particular case to ground these concerns, consider the current
1. INTRODUCTION interest in mobile and ubiquitous computing.
This is a paper about legibility. Where legibility occurs as a topic
As a relatively recent technical development, mobile computing
in HCI, we are usually concerned with the ways in which
has arisen in the context of a dominant model of conventional
information might be readable on a screen. Instead, though, I want
desktop computing. As such, then, the primary concern of mobile
to address a different sort of legibility – how do computer systems
computing designers has been to resolve the problems imposed by
help us read the world?
the new context of mobile computing, that is, the ways in which
Computers, as we all know, are tools. They are mechanisms for mobile systems fail to match a series of expectations associated
getting things done – not only in working domains, but also in with desktop settings. There are three particular problems to
domestic, leisure, and cultural settings. HCI has, since its which mobile computing applications typically respond [9]. First,
inception, been concerned primarily with the effectiveness of the problem of disconnection refers to the ways in which mobile
interactive systems when viewed from this instrumental work might be separated from resources – artifacts, people,
perspective. How well does a system match its needs, and how documents, and activities – that would be accessible in desktop
well does it get the job done? settings. So, the interest in “anytime, anywhere” access to
I want to argue here for a different perspective, one that focuses information figures mobile computing as inherently deficient in
on interactive systems as a means for encountering the world this regard, and frames disconnection as a problem towards which
around us. I am interested here not simply in interactive systems mobile computing must be oriented. Second, the problem of
as tools, but as cultural objects – objects that structure and dislocation addresses the fact that the computer and its user might
organize the world around us in meaningful ways. find themselves in unfamiliar surroundings that cannot be
navigated easily, and so suggests that a significant opportunity for
My concern here is with the relationship between social and mobile computing is wayfinding and resource identification –
technical, which can be understood in various ways. One common finding a path from A to B, and locating points of interest along
approach is to examine the “social impacts” of technologies, that path (including friends, places, and objects of interest). Third,
the problem of disruption refers to the ways in which, while
desktop systems are used in stable and known contexts, mobile
OzCHI 2007, 28-30 November 2007, Adelaide, Australia. Copyright the systems are used in settings that vary, so that the systems
author(s) and CHISIG. Additional copies are available at the ACM
themselves might disrupt the social contexts in which they are
Digital Library (http://portal.acm.org/dl.cfm) or can be ordered from
CHISIG(secretary@chisig.org) deployed. Efforts need to be made, then, to ensure that mobile
systems react appropriately in different settings, and much work in
OzCHI 2007 Proceedings, ISBN 978-1-59593-872-5 context-aware systems is directed towards this end.

1
The emergence of mobile and ubiquitous computing as areas of All the state simplifications we have examined have the
research inquiry has opened up new questions of interaction character of maps. That is, they are designed to summarize
design and directed our attention towards the connection between precisely those aspects of a complex world that are of
technology and the world in which it is deployed, in a variety of interest to the map-maker and to ignore the rest. … In
ways. One of these has been a concern with new forms of case after case, however, we have remarked on the
interaction that go beyond traditional keyboards and mice, and see apparent power of maps to transform as well as merely to
physical objects as potential sites for interaction with summarize the facts that they portray. (p. 87).
computational systems [13]. However, a second and potentially In other words, once maps – or similar representations – become
more significant transformation is to look at the interactive system tools of governance, they become the basis of actions which tend
as, itself, a means by which people encounter and understand the to bring the world into line with the map. Maps are inherently
world around them, both the built environment and the social partial, though, and so the simplifications of representation tend to
setting in which they find themselves. However, in order to do re-impose themselves upon the world.
this effectively, we need to find new conceptual tools and lenses
for understanding the role that computational and physical To the extent that the process of developing information systems
representations might play, which is my concern here. I want to is one of abstracting and representing the world for the purposes
suggest some ways in which we might think about the relationship of making it computationally tractable, these are arguments that
between representations and interaction design. However, are very familiar to those working in HCI, CSCW, and related
although my overall concern is with design, that is not where I areas. Research in Participatory Design, in particular, has
want to start. I will take a rather meandering path towards emphasized the importance of a sensitivity towards the ways in
information systems design by starting somewhere much further which practices of representation and abstraction are enmeshed
afield. within broader power relationships. These are not simply
arguments against the practices of abstraction and representation
2. SEEING LIKE A STATE at work; they are, after all, necessary. As Scott notes:
In his book “Seeing Like a State: How Certain Schemes for To complain that a map lacks nuance and detail makes no
Improving the Human Condition Have Failed,” anthropologist sense unless it omits information necessary to its function.
James Scott [18] sets out a broad account of the recurring patterns A city map that aspired to represent every traffic light,
of failure of central planning and social engineering. Scott’s every pot-hole, every building, and every bush and tree
primary target for investigation is the ideology of high modernist would threaten to become as large and complex as the city
statehood – the application of scientific rationalism to the it depicted. And it would certainly defeat the purpose of
governance of peoples. Two ideas are central to Scott’s mapping, which is to abstract and summarize. A map is an
investigation. instrument designed for purpose. We may judge that
The first is the idea of simplification. Modern statehood – by purpose noble or morally offensive, but the map itself
which he means a state as an institution that takes responsibility either serves or fails to serve its intended purpose. (p. 87)
for the planning and management of resources, including human So it is not partiality that is at issue here. Instead, Scott wants to
resources – relies first on accounting, and that accounting rests in explore how these abstractions are put to use in service of
turn upon a system for organizing the complexity of everyday statehood and governance, and what happens when abstraction
experience into calculable abstractions: and reality butt up against each other.
Officials of the modern state are, of necessity, at least one Scott draws effectively on a wide range of examples of the
step – and often several steps – removed from the society practices and consequences of high modern statehood. A number
they are charged with governing. They assess the life of of examples, for instance, discuss the impact of modernist
their society by a series of typifications that are always approaches on the spatial organization of everyday life, looking
some distance from the full reality these abstracted are for instance at villagization programs in Tanzania and Ethiopia
meant to capture… The functionary of any large and at modernist cities such as Brasilia or Chandigarh. The
organization ‘sees’ the human activity that is of interest to uniform layout of layout of villages, based on straight lines and
him largely through the simplified approximations of exact grids, simplifies planning and management but at the cost of
documents and statistics… These typifications are ignoring the local conditions and natural toplogy that dictate
indispensable to statecraft. The invention, elaboration and patterns of movement and habitation. In Brasilia, the utopian city,
deployment of these abstractions represent… an enormous with vast formal plazas, identical apartment buildings, and streets
leap in state capacity – a move from tribute and indirect designed for cars rather than people, is surrounded by an informal,
rule to taxation and direct rule. (p. 76-77) unplanned, peripheral jumble of residential neighborhoods that,
Scott is concerned, then, not just with the forms of abstraction and ironically, prospered and grew much faster than the city itself.
represention that are introduced, but with the ways in which these A second area that occupies Scott’s attention is the central
come to constitute the legibility of everyday life to the practice of planning of agricultural projects. Large-scale industrial agriculture
statehood; they are the forms of “seeing” in which states can is characterized by the same sorts of simplification and
engage. rationalization that Scott has outlined elsewhere. An emphasis on
Scott’s second insight is that these forms of simplification are not yield, on consistency, and on large-scale production results in a
simply the means by which states understand everyday action, but highly regulated approach to agriculture based on monocropping,
also the means by which they act upon it. In other words, in high control of genetic diversity, etc. However, while successful in the
modernism, these are also the means by which action is effected: industrialist West, these processes have generally failed when
imported (or exported) to Africa. Again, Scott argues that part of

2
the problem is the failure to adapt to local conditions and to resource planning and management that he ascribes to high
realize the complexity of the local environment. Invoking the modern statehood. As Agar [1] has extensively demonstrated, the
example of African polycropping, for instance, he notes that what history of the development of computing machinery – from
seems to industrialized scientific agriculture as unsystematic and mechanical tabulators to digital computers – has been intertwined
inefficient is, in fact, a careful and intricate response to local soil with the efforts of states and governments to manage their own
conditions, patterns of erosion, flooding and other location processes. If high modern statehood is fundamentally data-
particularities that are erased in the large-scale models associated centered, then computer systems have become the essential tools
with most centrally planned agricultural efforts. of state governance. It is hard to conceive how a contemporary
Scott’s primary concern, throughout these examples, is with the state could function without them.
ways in which centralized decision-making depends on systems of While, as Agar notes, this is a long-standing relationship, it is one
standardization and regulation that, by definition and necessity, that is being reconfigured and reinforced by the contemporary
erase local particularities, even though those particularities may transformations of state-hood that go under the broad umbrella of
be critical to effective planning and resource management. It is “neoliberalism.” Neoliberalism is characterized by the deliberate
important to recognize that there are two simultaneous issues at reduction in the areas of state responsibility, which manifests
work here. The first is a pragmatic and empirical one – simply to itself in the privatization of state-owned industry (e.g.
note that conditions “on the ground” are inevitably more complex transportation), the application of market models to state
than most views from afar will reflect, and that approaches that functions and the outsourcing of government functions to private
value the local will frequently yield more effective solutions. The industry (e.g. private prison services) and the delegation of state
second is an epistemological one – that normalization and functions to non-governmental regulatory agencies.
abstraction are part of the ideological apparatus of modernist Organizationally, then, neoliberalism manifests itself in part
rationalism, and, where they are called into question, the through the breakdown of monolithic state enterprises into
modernist model is itself undermined. Scott’s argument, then, constellations or networks of entities, some public and some
does not simply concern the failures of particular forms of private, linked together by market-based exchanges. Critically,
scientific rationalism as a tool for large-scale governance, but they are also linked by extensive information infrastructures.
rather questions the very institutional practices of scientific Arguably, data networking is a necessary precursor to the
rationalism. effectiveness of these kinds of networked organizational forms.
Digital information systems play a critical role in the distribution
2.1 High Modernism in Information Systems of governance roles between the state and private firms:
There are two reasons why Scott’s argument is particularly
relevant to the development of interactive software systems. Under the contemporary "mode of information" or
alternatively "cybernetic capitalism", administrative
The first is that that practices of software development are
control of social life by the state is facilitated by the
themselves often very much in line with the systems rationalism
systematic surveillance of consumption by private capital,
of high modernity that Scott describes. Information systems traffic
while the administrative control of consumption by
in ideal types in the form of data structures and abstractions that
corporations is facilitated by the systematic surveillance of
represent people and activities. The processes of representing
social life by government. [12]
elements of the everyday world in order to make them available
for processing, ordering, and managing through computer systems So, Scott’s account of high modernism is particularly relevant to
is bound up in the same modernist enterprise that Scott describes. the design of interactive information systems both because of the
Bowker and Star [3] have extensively analysed the processes parallels between the roles of systems rationalism in state
surrounding standardization and categorization in information governance and in information systems development, and because
systems, and Suchman [21] has explored the politics of of recent developments that give information systems an
computerized formalisms when put to work in workflow systems. increasing prominence in the governance of everyday life. The
Scott’s analysis highlights the problems that emerge when essence of my argument here, building up on the arguments of
representations begin to substitute for the phenomena that are Sengers and others, is that Scott’s argument motivates particular
represented, and where the process of abstraction and sorts of design interventions that refigure interactive systems as
homogenization is erased or disappears from view. It highlights points of reflection and engagement with the social structures
the ways in which practices of information representations in within which they are embedded – not so much “seeing like a
interactive systems are part of the broader modernist enterprise. state,” but “seeing like (or through) an interface.”
Given that software is inherently representational, interactive 2.2 Mobile Computing and Legibility
systems can never be built without engaging in these kinds of These issues show that Scott’s analysis is relevant for information
processes. However, in contrast to the prevailing modernist systems, but can we go further than that? I would argue that it is
tradition, a variety of authors have, over the last ten years or more, relevant not only for the analysis of information systems and their
begun to articulate an alternative position in which the use, but also for design. I will outline some specific examples
relationship between representation and action is refigured, and towards the end of the paper, but in order to set some context for
computational representations are seen as themselves embedded the discussion, let me note here that there are two ways in which
within socially-organized encounters between people that give we can see this position as being generative for design practice.
them meaning [2, 6, 19, 22].
The first is that, while mobile computing systems inherently
The second particular relevance of Scott’s argument is that depend upon representations of both space and activities, Scott’s
information technology has long been implicated in the kinds of account of the relationship between the practice of abstraction and

3
the power relations at work opens up the opportunity not simply threats to their sovereignty against which states must defend, their
to use such representations but to make them the focus or topic of borders are marked by considerably less fanfare and visibility. In
the system. That is, we can look for ways in which mobile systems consequence, unity is a less prounced feature of urban ordering.
might expose and present the representational systems at work as So, while municipalities are certainly subject to the same
a focus for examination and reflection. In this mode, mobile rationalizing tendencies as modernist states, the municipal
systems become devices through which the representational experience develops differently and in particular is characterized
practices that govern everyday space can be encountered, brought by a fragmentation of jurisdictions and responsibilities. Cities are
into view, understood, and questioned. patchworks of districts and regions reflecting different –
Of course, we might be able to do this through any sort of neighborhoods, zip codes, postal routes, school districts, tax
interactive system, but there are some particular resonances that assessment zones, conservation, electoral wards, catchment
make mobile systems especially interesting sites for these sorts of regions, police precincts, and more.
encounters. Mobile and ubiquitous systems allow us to create If the high modernism of state governance resulted in a series of
spatially-situated collisions between the abstract and process of homeogenization and standardization, urban
decontextualized rationality of rationalist representations and the governance is characterized instead by plural and heterogeneous
messy detail of the everyday world. orderings – but orderings none the less. As Valverde notes, the
The second link to design focuses on the ways in which various issues with which municipalities have cocnerend
interaction with the world through mobile systems is part of the themselves – public hygiene, transportation, crime, etc – are
way in which the social reality of the world is constituted. The irreducible both to each other and to some other fundamental
widespread adoption of mobile systems, perhaps most especially order. Fragmentation is a matter here, then, of scale, of
mobile telephony but also wireless internet services and mobile perceptions of risk, and of the nature of the municipality’s affairs.
computers, has been accompanied by a series of transformations Valverde’s goal is not to set up the notion of municipal as
in our understanding of the spaces that we occupy each day. Such conceptual dual to Scott’s state ordering, but rather to highlight
phenomena as people working from WiFi hotspots in cafes, the different forms of administrative gaze and the ways in which
“silent” coaches on trains where mobile phone calls are they reflect political, pragmatic, and historical considerations. To
disallowed, and hypercoordination achieved through SMS think of the gaze as plural and heterogeneous is to start thinking
messaging all speak to a complex series of mutual of the many different simultaneous perspectives and
accommodations between emerging technologies and evolving representations at work. What is perhaps particularly valuable,
understandings of space [8]. Our experience of everyday space is from the perspective of mobile and ubiquitous computing, is the
not pre-given, but continually being negotiated and produced as a ways in which one might, at any given moment, be situated within
consequence of individual and collective action From this an overlapping set of infrastructures, systems, domains, and
perspective, the relationship between technology and spatial representations. One is positioned, at any moment, in relationship
representation is much more complicated; representations, explicit to different organizations, infrastructures, and regulatory regimes
and implicit, are means by which spaces are not only encountered arises both from “above” (regimes of governmental authority) and
and managed but produced. For instance, geographers Rob from “below” (the actions of self and others).
Kitchin and Martin Dodge have recently suggested that
cartography be rethought as processual, rather than As a legal scholar, Valverde is interested too in the ways in which
representational; that is, that we think about maps as things that these kinds of regulative systems are invoked. She points out that
emerge through practice rather than as capturing and documenting the forms of urban regulation are quite different from state
underlying realities [15]. regulation; urban law enforcement is largely (although not
entirely) a response to citizen complaints rather than a
This suggests a second avenue for design practice, then, which consequence of vigilant surveillance. Regulative schemes, then,
focuses on the ways in which spatial representations arise out of vary in the extent and circumstances in which they are brought to
patterns of inhabitation. Mobile technologies can act as sites for bear upon the world; the processes of alignment between
the production of new forms of spatial experience, connecting representation and reality are themselves heterogeneous.
people to each other in space and creating representations and
“mappings” that reflect emergent patterns of activity rather than a 4. SEEING LIKE AN OIL COMPANY
surveillant gaze. Finally, here, we will consider another exploration of Scott’s ideas
I will present specific design examples towards the end of the in a different context. Ferguson’s brief article “Seeing Like an Oil
paper that illustrate these approaches. First, though, it’s useful to Company” [11] sets out explicitly to explore the relevance of
examine two critiques of Scott’s argument that elaborate his Scott’s arguments for neoliberal global capitalism, drawing in
arguments around the homogenizing practices of high modernism. particular on his fieldwork on the operations of multinational
mineral extractive industries in Africa. As Ferguson notes, Scott
3. SEEING LIKE A CITY himself comments that the monolithic forms of statehood that he
Legal scholar Mariana Valverde has used the term “seeing like a describes might be seen as somewhat archaic relics from the
city” to describe an alternative to the homogenizing strategies that perspective of the early twenty-first century, but suggests that
Scott reports [23]. Valverde’s concern is with “the tools of urban global capitalism depends every bit as much upon
ordering.” Concerned particularly with urban legalities, she “homogenization, universality, grids, and heroic simplification.”
observes that cities have different concerns than states. She notes It is this latter claim that Ferguson contests. Ferguson raises
in particular that municipalities and their ordering tools do not several issues, but I will focus on just two here –extractive
suffer the same sort of “territorial anxieties” as states; with the enclaving, and the relationship between production and violence.

4
By “extractive enclaving,” Ferguson refers to the common pattern potential design interventions when we take these ideas and apply
of presence of multinational extractive industries in Africa, where them in the context of interactive systems.
they are located within but separated from the countries in which
they operate. Offshore drilling is perhaps the most clear-cut 5. SEEING “SEEING”: TOOLS FOR
example; in extreme cases, neither the oil nor the money that it LEGIBILITY
generates even touch the soil of the country in question. However, As I have suggested, information systems are implicated in the
even in cases where oil or other minerals need to be mined on kinds of projects that Scott (and Valverde, and Ferguson) describe
land, it is quite common for the mines to be located on private in two significant ways. The first is that the representational
land managed entirely by the company, policed and secured by practices that support much of information systems development
private firms, with foreign workers accommodated on site, bussed are strongly related to the systems rationalism that underlies the
in directly from the airport or transported by helicopter so that phenomena that Scott analyses. The second is that information
they scarcely set foot outside of the protective regime. As systems have always been strongly involved in the development of
Ferguson notes, these territorial enclaves are often tightly these kinds of rationalist projects on a large scale. Contemporary
connected to the headquarters of the relevant multinational developments, especially in the form of networked organizations
corporation while largely disconnected from the nation within and distributed governance, have only deepened the connection
which they are located. The homogenizing grid has not been between information and interactive systems and forms of
extended to incorporate the developing world, then, as much as it administrative control.
has been overlaid, passing over those areas that it cannot control
or in which it is not interested. If we accept the deep link between forms of information system
development and the homogenizing character of high modernism
We can see this in particular in the second issue of Ferguson’s I in its various manifestations, what might we want to do? Clearly,
want to raise here, the relationship between mineral production there is a range of responses. One is simply to develop a set of
and endemic violence. During the major period of African conceptual resources for the analysis of information systems and
decolonization and reconstruction, the dominant ideology their use; arguably, the emergence of social informatics [16] is
suggested that the interests of global capital and of nation-states precisely this, an attempt to understand the linkage between social
overlapped; foreign economic investment would provide emergent systems and information systems. A second response is that
nations with the resources they needed to build civic and material represented by the Participatory Design movement, in its classic
infrastructure, and in turn the stability and increasing economic Scandinavian tradition, which looks upon information systems
well-being of these nations would make them attractive partners design, implementation, and deployment as sites for democractic
for global enterprise. As suggested above, however, action.
multinationals have been able to operate quite effectively while
by-passing the nation-state altogether for a range of their needs, In recent work, colleagues and I have been concerned with a third
and indeed one could argue that not only do they not depend on response, in line with Agre’s concern for “critical technical
national stability for their effective operation, but they seem to practices” and Senger et al.’s “reflective design.” [2, 19] At the
thrive particularly in conditions of endemic violence, from low- heart of the Critical Technical Practice agenda is to see
level criminality and insecurity to outright civil war. The information system design as, itself, a critical practice and to see
dependence upon private security firms rather than civil police or interactive systems as a tool for critical inquiry. In her account of
armies means that multinational firms may, in fact, operate most reflective design, Sengers discusses the ways in which systems
effectively where states are least effective. As Ferguson reports, it design can provide a point of critical reflection on underlying
is precisely the countries that are, from the perspective of the IMF assumptions for both system designers and users. She comments,
or the World Bank, the biggest “failures” as states that are the “reflection is not a purely cognitive activity, but is folded into all
most successful at developing these transnational extractive our ways of seeing and experiencing the world.” It is just this
enclaves. issue – seeing in particular – that motivates the exploration here,
both in conceptual terms and in design terms.
Like Valverde’s discussion, Ferguson’s analysis opens up a series
of interesting questions around the centrality of the grid and As a design strategy, we have bee interested in the ways in which
practices of homogenization that Scott discusses. Scott talks about we might be able to use information technologies to turn attention
the ways in which systems rationalism fails when the grid is towards modes of seeing – strategic modes, tactical modes,
rigidly imposed upon local realities; Ferguson highlights surveillant modes, practical modes, and more. Our concern is to
alternatives for bringing grid and world into alignment. As he highlight the forms of representation and standardization at work,
notes, capital does not simply “flow” uniformly in the systems he to turn attention to the processes through which they emerge and
is exploring; it hops from place to place. Similarly, the imposition the sites at which they are negotiated, and so to make visible the
of uniformity may not, itself, be uniform. Where Valverde points various grids in which people are enmeshed and which they enact
to systems of urban ordering as plural and heterogeneous, in the course of daily life. I will provide two examples here from
Ferguson notes that the ordering systems of global capital may not ongoing work.
be strongly aligned with those of nation states which it comes into Datascape [14] is tool for exploring urban space and its
contact. Both recognize the power of Scott’s analysis, but provide representational analogues. Datascape provides a vehicle-mounted
us with new examples that suggest that we need to pay attention to display that can be used, much like a periscope, to examine the
the many ways in which standards and grids might manifest surroundings while moving through them. The system, then,
themselves. Indeed, it is just these issues – the complex, multi- combines two visual experiences of urban space. One is the sight
faceted, and multilayer nature of grids, regulation, and of the surroundings as viewed from the vehicle. The second,
representation in contemporary space – that motivate a series of presented on the display, is a virtual world with a topography

5
generated on the basis of geo-referenced data sets such as census relationship between above ground and below ground, and the
records, block-by-block geodemographic marketing data, disease non-instrumental aspects of public transportation. In studies of
incidence records, etc. This is Valverde’s urban patchwork public transit in both California and London, a strong concern
mapped into a virtual world; what is particularly interesting here emerged with those aspects of transit that went beyond “getting
is the superimposition of the representational space with the from A to B” and focused on the elements by which people would
physical space, so that one can explore the landscapes of daily life understand trips as good or bad, including the skill of execution,
in the terms in which they are organized for marketers, the experiences or people encountered along the way, and more.
epidemiologists, and others. By linking different databases for the This turned our attention to the aesthetics of the journey, broadly
same space, spatial correlations can be made; while this can also construed, and to the collective rather than purely the individual
be achieved with conventional geographical information systems, elements of riding together on public transit. Undersound is a
a radically different experience is achieved when these can be music sharing system for mobile phones that allows Tube riders to
grounded in the space through which one is moving. Datascape is browse and exchange music. Music files, though, are strongly tied
at is most powerful when it used not as a tool for exploration but to the geography of the Tube itself; they originate, terminate, and
as a tool for reflection, and when the spaces being examined are are transferred in particular places. So, from another perspective,
the familiar uses of daily experience. Initial, static versions of Undersound is a system by which music hitch-hikes around the
Datascape have already been exhibited; the mobile version is Tube on people’s mobile phones. Music circulates, much as
currently in the prototype stage. people do. What we are interested in here is the ways in which the
It is clear that Datascape responds to the considerations presented system can provide people with an understanding of two
in this paper by exposing representational grids. However, otherwise obscured elements of life in public transit. One is the
Datascape attempts to go beyond this, in at least three ways. connection between places and regional or ethic styles, as
captured by the forms of locally-produced and uploaded music
First, by drawing together multiple different forms of information that characterize stations or regions of the city. The second is the
and different databases, Datascape is designed to expose the flow of bodies within which one is enmeshed as a traveler along
relationship between multiple grids and multiple practices of particular routes at particular times of day.
homogenization and representation. What becomes particularly
interesting as a matter for exploration is the ways in which those Here again, then, we are concerned with questions of legibility,
grids do and do not line up, the elements that they have in although they are of quite different sorts. In his writings on the
common or separately, the boundaries that are shared and those practices of urban life Michel de Certeau distinguished between
that are unique. Datascape provides a view of this complex of strategic modes of encounter with urban forms, and tactical
representations as much as it does any single one. modes. The strategic modes are the modes of design; they are
concerned with the creation of urban forms in order to control,
Second, Datascape doesn’t simply allow one to explore manage, and regulate activities within them (shaping urban
georeferenced data via a digital system (e.g. through a web growth, creating points of assembly, routes, etc.) By contrast, the
browser), but instead situates people within the spaces that are tactical modes are the modes of use and appropriation; the ways in
represented, as a means to explore them. Critically, it uses these which people’s movements through space create new forms of
familiar spaces as ways of contextualizing the data. In other local meaning, on the individual level (e.g. personal choices
words, it is not simply demonstrating the legibility of everyday between different routes taken to work that reflect different
spaces through the use of large-scale data sets; it is also moods) or on the collective level (e.g. the emergent and transitory
attempting to make those data sets legible through the use of associations between particular ethnic or subcultural groups and
everyday space. Using familiar space as a lens through which to regions of the city and the impacts that they have.) Our concern is
view and understand things like scientific, environmental, and very much here with the tactical, then. Undersound makes visible,
epidemiological data invites new stakeholders into conversations as a site of examination and reflection, the urban flows within
for whom the source of legibility is the inhabited space itself which people and artifacts are enmeshed, making them
rather than the abstract representations of graphs, charts, and meaningful by placing them in the contexts of journeys that
scientific or mathematical formalism. themselves are suggestive of particular purposes, needs, and
Third, when used as an authoring tool rather than purely as a characters. This is legibility from below rather than above.
visualization tool, Datascape offers the opportunity for people to
create tours and routes simultaneously through physical space and 6. INFRASTRUCTURE AND
data space, offering a very different form of legibility, one based INTERACTION
on communities and peer communication. Here, again, the data One of the considerations motivating this work is to reconsider
sets that are not being used to make space legible, but rather are infrastructure as a site of interaction. Much of what we do in
the objects of representation and critique. creating interactive systems is to create infrastructures – data
While Datascape is primarily concerned with the sorts of data sets representations, storage mechanisms, networks, frameworks, and
that make everyday space legible to organizations, bureaucracies other technical elements that are designed to stabilize elements of
and states, our second example is one that draws in particular on the world and make them accessible and manageable. The
the forms of emergent legibility that arise out of habitation and essential features of infrastructures is that the lie “beneath” the
movement in everyday space. level of conventional interaction or, indeed, of conventional
interest; they are relied upon for everyday task, but largely
Undersound [4] is a music-sharing system design for the London unremarked and invisible in use.
Underground. It responds to a number of aspects of the urban
experience within which the London Underground is enmeshed, Leigh Star makes a compelling case for infrastructure as a site of
including the social and ethnic diversity of London, the complex ethnographic inquiry, noting that the processes by which

6
infrastructures become infrastructures – that is, the processes by and summarization is to overlook the critical and analytic
which complex and unruly objects are stabilized and subdued so engagement that is the basis of its approach. In turn, I would
that they can withdraw into the background of everyday use – are argue that it is useful to turn to social science and related
important ones that deserve attention when we look at materials not merely for their empirical richness but for their
technological systems [20]. These are the processes by which analytic insight and for the ways in which they encourage us to
standards are established, services configured and offered, and ask a new set of questions as the basis of design experience.
conventions developed that, through their ubiquity and The third concern is a recognition of the transformation of
uniformity, start to be taken for granted.1 Infrastructures are not interactive systems as sites of inquiry. The discipline of HCI has
purely technological; they are political, legal, and economic evolved considerably over several decades, but so too have
entities, and the processes by which they are established, computer systems themselves. What I want to draw attention to
transmitted, used and overturned need to be examined also in here is not simply the fact that computers have become faster,
those terms. smaller, and more powerful as technological artifacts, but that
However, when we think of technological systems as tools they have emerged as cultural objects in a radically different way
through which we experience the everyday world, then the than they did before. They are elements of the landscape of daily
interactional consequences of these kinds of infrastructural life in many different forms. Digital devices are embedded in our
processes become even more pressing. Infrastructures are tools of cultural and social imagination in very different ways than they
interactive experience; they become manifest in interaction, and were when HCI was emerging. To the extent that our discipline
they structure those interactions. Despite claims of an emerging thinks not simply about user interface design but about
world of “ubiquitous” computing, it is precisely the disjunctures, interactions between humans and computers, these
boundaries, and seams that infrastructures are designed to erase transformations suggest that we need to look more broadly for
that are of interest to us here [10]. We find this a compelling and theoretical perspectives that help us understand how computation
fruitful area for design. manifests itself as a cultural object [e.g. 17].
In this paper, I have attempted to draw on these positions in order
7. CONCLUSIONS to argue, first, that legibility is a critical issue for information
The work presented here forms part of three larger and related
technology in two ways – in the forms of ordering that digital
programs.
representations support, and in the opportunities for making these
The first is the burgeoning interest in a reflective approach to HCI visible in the course of interaction. I have argued, second, that
– one that is itself concerned with the critical dimensions of mobile and ubiquitous computing is a particularly fruitful area for
design, and is also oriented towards promoting reflection and exploring these questions, because the movement “off the
critical inquiry into technological devices on the part of “users.” desktop” (and out of the office) inherently raises questions about
One of the underlying considerations here is that computers, as the nature and orgnization of the spaces and settings into which
cultural artifacts, participate in, reflect, enact, and reinforce computation has moved. As illustrated by Datascape and
cultural values and assumptions, and so that, first, computer undersound, the ways in which ubicomp technologies are situated
system design is an effective site at which to engage in these kinds in space, in the “here and now,” is a powerful element in building
of critical engagements with everyday practice, and, second, compelling reflective systems.
critical approaches to technology design are both appropriate and
necessary. Reflective HCI suggests an approach to interaction 8. ACKNOWLEDGMENTS
design in which cultural assumptions and values play as important This work was supported in part by the US National Science
a role as traditional usability metrics both as measures of success Foundation under awards 0133749, 0205724, 0326105, 0527729,
and as elements of the design process. To the extent that my 0524033, and 0712890. I am grateful to Johanna (undersound)
considerations of legibility here are ones oriented towards a Brewer, Eric (Datascape) Kabisch, Phoebe Sengers, Amanda
“user’s” interaction with the people and institutional structures Williams, and the students enrolled in the Winter 2007 offering of
that also occupy everyday space, it is very much in line with these my seminar “Information, Technology, and Space,” for
considerations. contributions, corrections, conversations and comments on drafts.
The second broader concern is the relationship between
technology design and social science, most particularly the 9. REFERENCES
theoretical and analytic considerations in social science research. [1] Agar, J. 2003. The Government Machine: A Revolutionary
HCI has a long history of drawing upon the empirical components History of the Computer. Cambridge, MA: MIT Press.
of sociology, anthropology, and related areas through, for [2] Agre, P. 1997. Towards a Critical Technical Practice:
example, the adoption of ethnographic research methods as means Lessons Learned in Tryng to Reform AI. In Bowker, Gasser,
for requirements analysis and capture. However, as argued Star, and Turner, eds, Bridging the Great Divide: Social
elsewhere [6], a purely empirical perspective neglects many of the Science, Technical Systems and Collaborative Work.
contributions to be gleaned from social science research, even that Laurence Erlbaum.
with an empirical basis. To reduce ethnography to data collection
[3] Bowker, G. and Star, S. 2000. Sorting Things Out:
Classification and its Consequences. Cambridge, MA: MIT
1
The oft-repeated story that, through a series of infrastructural Press.
constraints and attempts to achieve backward compatibility, US [4] Brewer, J., Bassoli, A., Martin, K., Dourish, P., and
trains run on tracks that are the width of a Roman war chariot is Mainwaring, S. In press. Aesthetic Journeys: Rethinking
illustrative, if apocryphal. Urban Computing. IEEE Pervasive Computing.

7
[5] De Certeau, M. 1984. The Practice of Everyday Life. [14] Kabisch, E. In press. From Virtual to Hybrid Worlds. IEEE
Berkeley: University of California. Pervasive Computing.
[6] Dourish, P. 2001. Where the Action is: The Foundations of [15] Kitchin, R. and Dodge, M. 2007. Rethinking Maps. Progress
Embodied Interaction. Cambridge, MA: MIT Press. in Human Geography, 31(3), 1-14.
[7] Dourish, P. 2006. Implications for Design. Proc. ACM Conf. [16] Kling, R. 2000. Learning about information technologies and
Human Factors in Computing Systems CHI 2006 (Montreal, social change: The contribution of social informatics. The
Quebec), 541-550. New York: ACM. Information Society, 16(3), 217-232.
[8] Dourish, P. 2006. Re-Space-ing Place: Place and Space Ten [17] Satchell, C. 2006. Cultural Theory: From Armchair Critic to
Years On. Proc. ACM Conf. Computer-Supported Star Performer. Proc. OzCHI 2006 (Sydney, Australia), 199-
Cooperative Work CSCW 2006 (Banff, Alberta), 299-308. 204.
[9] Dourish, P., Anderson, K., and Nafus, D. 2007. Cultural [18] Scott, J. 1998. Seeing Like A State: How Certain Schemes to
Mobilities: Diversity and Agency in Urban Computing. Proc. Improve the Human Condition have Failed. New Haven, CT:
IFIP Conf. Human-Computer Interaction INTERACT 2007 Yale.
(Rio de Janeiro, Brazil). [19] Sengers, P., Boehner, K., David, S., and Kaye, J. 2005.
[10] Dourish, P. and Bell, G. 2007. The Infrastructure of Reflective Design. Proc. Fourth Decennial Aarhus
Experience and the Experience of Infrastructure: Meaning Conference on Critical Computing.
and Structure in Everyday Encounters with Space. [20] Star, S.L. 1999. The Ethnography of Infrastructure.
Environment and Planning B: Planning and Design, 34, 3, American Behavioral Scientist, 43(3), 377-391.
414-430.
[21] Suchman, L. 1993. Do Categories have Politics? The
[11] Ferguson, J. 2005. Seeing Like An Oil Company: Space, Language/Action Perspective Reconsidered. Computer-
Security, and Global Capital in Neoliberal Africa. American Supported Cooperative Work, 2, 177-190.
Anthropologist, 107(3), 377-382.
[22] Suchman, L. 2007. Human-Machine Reconfigurations: Plans
[12] Goss, J. 1995. We Know Who You Are And We Know and Situated Actions. Cambridge.
Where You Live: The Instrumental Rationality of
Geodemographic Systems. Economic Geography, 71(2), [23] Valverde, M. 2004. Seeing Like A City: Legal Tools of
171-198. Urban Ordering. Paper presented at Symposium on
“Genealogies of the Legal,” UC Irvine, Irvine, CA.
[13] Ishii, H. and Ullmer, B. 1997. Tangible Bits: Towards
Seamless Interfaces between People, Bits, and Atoms. Proc.
ACM Conf. Human Factors in Computing Systems CHI’97
(Atlanta, GA), 234-241.