Re-place-ing space: the roles of place and space in collaborative systems

atece

tu-nde”.e

banink a

than at

ntiteddms

hasverthen

asn-agese],

rtu-onsa”, asre-tedora-l

su-

hed byur-

Re-Place-ing Space: The Roles of Place and Space in Collaborative Systems

Steve Harrison* and Paul Dourish†

*Xerox Palo Alto Research Center†Rank Xerox Research Centre, Cambridge Lab (EuroPARC)

[email protected], [email protected]

Abstract

Many collaborative and communicative environments usenotions of “space” and spatial organisation to facilitate andstructure interaction. We argue that a focus on spatial modelsis misplaced. Drawing on understandings from architectureand urban design, as well as from our own research findings,we highlight the critical distinction between “space” and“place”. While designers use spatial models to support inter-action, we show how it is actually a notion of “place” whichframes interactive behaviour. This leads us to re-evaluatespatial systems, and discuss how “place”, rather than“space”, can support CSCW design.

Keywords: space, place, media space, virtual reality,MUDs, metaphor.

1 IntroductionWe live in a three-dimensional world. The structure of thespace around us moulds and guides our actions and interac-tions. With years of experience, we are all highly skilled atstructuring and interpreting space for our individual or inter-active purposes. For instance:

• The objects we work with most often are generallyarranged closest to us. Computer keyboards, currentdocuments, common reference materials and favouritepieces of music might immediately surround us in anoffice, while other materials are kept further away (in fil-ing cabinets, cupboards or libraries).

• Physical spaces are structured according to uses andneeds for interaction. An office door can be closed togive independence from the space outside, or left open tolet us see passers-by. People’s offices are more likely tobe sited near to the offices of their colleagues.

Observing the way that space structures actions and interac-tions—the “affordances” of space [Gaver, 1992]—manydesigners have used spatial models and metaphors in collab-orative systems. The desktop metaphor of single-usersystems has been extended to a metaphor of desks, offices,hallways and cities. These systems all facilitate natural col-laboration by exploiting our understandings of space—theproperties of the three-dimensional world in which we liveand interact every day.

In this paper, we will critically explore the use of space as abasis for CSCW design. We will argue that the critical prop-

erty which designers are seeking, which we call appropribehavioural framing, is not rooted in the properties of spaat all. Instead, it is rooted in sets of mutually-held, and mually available, cultural understandings about behaviour aaction. In contrast to “space”, we call this a sense of “placOur principle is: “Space is the opportunity; place is thunderstood reality”.

Place is a fundamental concept in architecture and urdesign, and we can learn from those disciplines how to thabout place in collaborative systems. Place derives fromtension between connectedness and distinction, rather from three-dimensional structure, and we can see thiswork in a variety of collaborative systems.

We will begin, in the next section, by looking at the curreuse of space in collaborative systems, and how it is exploto structure interaction. Next, we will introduce the relatenotion of place, and compare their roles in existing systeand consequences for future designs.

2 Space in Collaborative SystemsThe use of spatial metaphors and spatial organisation become increasingly popular in a collaborative systems othe past few years. We will describe some systems, and look at the properties they exhibit.

2.1 Spatially-based Systems Collaborative Virtual Reality. Most demonstrably, experi-ments with collaborative virtual reality systems, such DIVE [Carlsson and Hagsand, 1993] and MASSIVE [Greehalgh and Benford, 1995], use virtual spaces to mandistributed multi-user interaction. Both of these systems ua “spatial model of interaction” [Benford and Fahlen, 1993in which participants’ awareness of each other, and opponities for interaction, are managed through spatial extensiof their presence, attention and influence called “aur“focus” and “nimbus”. These mechanisms are designedcomputational equivalents of real-world patterns of awaness and interaction in these virtual spaces. Relamechanisms extend these interactional spaces for collabtive work, such as collaborative information retrieva[Sawyer and Mariani, 1995], using spatial metaphors to vialise users in an information landscape.

MUDs. At the other end of the technology spectrum, texplosion of interest in the Internet has been accompaniea huge increase in the popularity of MUDs and MOOs [C

Draft submission for CSCW’96—not for distribution. Last modified: 3/12/96

d aity,e is

c-to

s-onof

y-ithnceurend,ntvi- of

eyurto

m).tialus

ulithtial of

els

ssa-ntur

erly-

It put

e”thehtiveose

tis and Nichols, 1994]; text-based (or simple graphical)interactive environments. Collaborative systems based onthese technologies [e.g. Curtis et al., 1995] have also empha-sised the use of “real-world” spatial metaphors to supportcollaboration. MUDs structure their virtual worlds into sep-arate locations (“rooms”), and allow participants to movefrom location to location, selectively participating in events,activities and conversations. The spatial metaphor runs allthe way through the MUD model of action and interaction.

Multimedia Communications. The same principles have alsobeen at work in other communicative systems. Like collabo-rative virtual realities, these have drawn upon analogies withthe spatial organisation of the everyday physical world tostructure aspects of multi-user interaction. For example, theoriginal design of the Cruiser media space system [Root,1988] used a metaphor of “virtual hallways” as an organisingprinciple for interaction and participation in an AV-mediatedcommunication system. The Vrooms system at EuroPARC[Borning and Travers, 1991] used spatial proximity in aninterface to control connections; and recent work on the Tor-onto Telepresence project has introduced multiple camerasinto a single media space node to reflect the notion of differ-ent views from the office doorway or across the desk of acolleague [Buxton, forthcoming].

The basic premise which lies behind these varied uses of spa-tial models and metaphors is that, in collaborative settings,designers can exploit our familiarity with the spatial organi-sation of our everyday physical environments. In particular,they wish to exploit the ways that space structures and orga-nises activity and interaction.

2.2 The Features of SpaceThere are many aspects of the “real world” which can beexploited as part of a spatial model for collaboration:

Relational orientation and reciprocity. The spatial organisa-tion of the world is the same for all of us. “Down” is towardsthe center of the earth, and “up” is towards the sky; we rec-ognise “front” and “back”, and understand what that impliesfor our field-of-view. Our common orientation to the physi-cal world is an invaluable resource in presenting andinterpreting activity and behaviour. Since we know that theworld is physically structured for others in just the same wayas it is for ourselves, we can use this understanding to orientour own behaviour for other people’s use. This is what letsus point to objects, or use spatial descriptions to establish ref-erence. Referring to “the document on top of that pile” or“the person standing by the bookcase” relies on mutual spa-tial orientation. Reference can also depend on our sharedexperience of what it’s like to be in a space, such as when wetell someone, “the door is on the left just as you come aroundthe corner”.

Proximity and action. In the everyday world, we act (more orless) where we are. We pick up objects that are near us, notat a distance; we talk to people around us, because our voicesonly travel a short distance; we carry things with us; and weget closer to things to view them clearly. Similar propertiesare exploited in collaborative virtual spaces. Understandingsof proximity help us to relate people to activities and to each

other. When we see a group of people gathered arounmeeting table, we understand something about their activand we know that another person standing off to one sidlikely to be less involved in their activity.

Partitioning. Following on from the notion of proximity andactivity is a notion of partitioning. Since actions and interations fall off with distance, so distance can be used partition activities and the extent of interaction. MUD sytems, for example, use rooms or locations to partitiactivity. MUD rooms provide a restricted view into the set interactions currently in progress in the system overall.

Presence and awareness. As we move around the everydaworld, it is filled not only with the artifacts, tools and representations of our work, but also with other people and wsigns of their activity. The sense of other people’s preseand the ongoing awareness of activity allows us to structour own activity, seamlessly integrating communication acollaboration ongoingly and unproblematically. Similarlyspatially-organised collaborative environments preseviews of other people and their actions within the same enronment which represents activity and holds the artifactswork.

3 Appropriate Behavioural FramingThe real-world value of the features listed above is that thgive critical cues which allow us to organise our behavioappropriately (such as moving towards people to talk them, or referring to objects so that others can find theCollaborative virtual spaces exploit aspects of space (spamechanisms, such as providing identity, orientation, a locfor activity, and a mode of control) which can be powerftools for the design. But these spatial metaphors carry wthem some decidedly non-metaphoric aspects—spabehaviours—that emerge from our everyday experiencethe physical world.

So, what is being supported by spatial collaborative modis a way of ongoingly managing activity in collaborative set-tings. We call this appropriate behavioural framing. Theimplied rationale is that if we design collaborative systemaround notions of space which mimic the spatial organition of the real world, then we can support the emergepatterns of human behaviour and interaction which oeveryday actions in the physical world exhibit. In othwords, spatially-organised systems will support spatialmanaged behaviours.

Our argument here is that this model is too simplistic.needs to be examined and studied further before it can beto use in systems design.

4 From Space to PlaceThe properties outlined above define the notion of “spacwhich we will use in this paper. Space is the structure of world; it is the three-dimensional environment, in whicobjects and events occur, and in which they have relaposition and direction. The properties of space are thwhich derive from that definition, as we showed above.

2

rive

thanndcts.s to

nsn-

calshalent

uchelf;in

ilaranig-

anrmset-d, of inbutcta-

for88] a ofcesle’sal

gse,ys-e

plybut

r-gd-rns

We argued that features of space have been exploited bysystem developers in the attempt to regain the sense ofappropriate behavioural framing which we observe andencounter in the real world. However, in everyday action,this appropriate behavioural framing comes not from a senseof space, but from a sense of place. Our key principledescribes the relationship between the two: Space is theopportunity; place is the understood reality.

So, what is place and what does it do for us?

4.1 Place and Behavioural FramingPhysically, a place is a space which is invested with under-standings of behavioural appropriateness, culturalexpectations, and so forth. We are located in “space”, but weact in “place”. Furthermore, “places” are spaces that are val-ued. The distinction is rather like that between a “house” anda “home”; a house might keep out the wind and the rain, buta home is where we live.

A conference hall and a theatre share many similar spatialfeatures (such as lighting and orientation); and yet we rarelysing or dance when presenting conference papers, and to doso would be regarded as at least slightly odd (or would needto be explained). We wouldn’t describe this behaviour as“out of space”; but it would most certainly be “out of place”;and this feeling is so strong that we might try quite hard tointerpret a song or a dance as part of a presentation, if facedwith it suddenly. It is a sense of place, not space, whichmakes it appropriate to dance at a Grateful Dead concert, butnot at a Cambridge college high table; to be naked in the bed-room, but not in the street; and to sit at our windows, peeringout, rather than at other people’s windows, peering in. Place,not space, frames appropriate behaviour.

Conversely, the same location—with no changes in its spa-tial organisation or layout—may function as different placesat different times. An office might act, at different times, asa place for contemplation, meetings, intimate conversationand sleep. So a place may be more specific than a space. Aspace is always what it is, but a place is how it’s used.

4.2 Place is In SpaceOne reason that it can be hard to see the separation betweenplace and space is that, in our everyday experience, placeslargely exist within spaces. (Later, however, we will describesome space-less places.) A place is generally a space withsomething added—social meaning, convention, culturalunderstandings about role, function and nature and so on.The sense of place transforms the space. As a space, the brickporch outside EuroPARC where smokers gather is uninvit-ing; but it is valued as a place for relaxation and gossip. It’sstill a space, even though place is what matters.

Since our world is spatial and three-dimensional, notions ofspace pervade our everyday experience. Everything in ourworld is located in space, and so “place” is tied up with it too.It is part of the very metaphoric structure of our language.Tuan [1977] points out that even spatial relations are loadedwith meaning, with “high” being good and “low” being bad.Lakoff and Johnson [1980] label this an “orientational meta-phor” and give a long list of examples. Spatiality runs

throughout our experience and our thought. Places demuch of their meaning, then, from their spatiality.

However, the sense of place is dependent on much more simply the spatial organisation of our surroundings, amore than the three-dimensional arrangement of artefaPlaces also call up cultural understandings which help uframe our behaviour.

4.3 Place in Social AnalysisAnalysts of social action have been concerned with notioof place, and with the settings which convey cultural meaing and frame behaviour. Goffman [1959] uses a theatrimetaphor, where “frontstage” and “backstage” distinguidifferent modes of behaviour and action in interpersoninteraction. He points explicitly to “regions” as one of thelements which contributes to the framing of these differestyles of action. However, behaviour can be framed as mby the presence of other individuals as by the location itsin other words, the “place” is more than simply a point space.

Giddens [1984] adopts the term “locales” to capture a simsense of behavioural framing. Again, these are more thsimply spaces; he observes, “it is usually possible to desnate locales in terms of their physical properties... but it iserror to suppose that locales can be described in those tealone.” For Giddens, again, the critical feature of these stings is the way in which “features of settings are [...] usein a routine manner, to constitute the meaningful contentinteraction”. In other words, what these analysts point tohuman action is the how it is framed not only by spaces, by the pattern of understandings, associations and expetions with which they are infused.

4.4 Place in the Built EnvironmentPlace, as we have described it here, is a central concernarchitects and urban designers. For example, Whyte [19provides detailed descriptions of the life of the street inmodern city. His comprehensive descriptions of the usethe street-side plazas highlight the issues between plawhich “work” and those which do not; whether or not peopwant to be there. Similarly, while Christopher Alexander[1987] “patterns” ostensibly describe principles of physicdesign, the focus is less on the structure of buildings and cit-ies, and more on the living which goes on in them. Hecomments, “Those of us who are concerned with buildintend to forget too easily that all the life and soul of a placall of our experiences there, depend not simply on the phical environment, but on the pattern of events which wexperience there.” [Alexander, 1979]

So, architects and urban designers are concerned not simwith designing three-dimensional structures (spaces), with places for people to be. For them, the idea of placederives from a tension between connectedness anddistinction.

Connectedness is the degree to which a place fits with its suroundings, maintaining a pattern in the surroundinenvironment (such as color, material or form)—or responing to those patterns, even if it does not maintain the patte

3

nec-eirmishtheho

p.

rypthadla-onmhecanentible,s.m- by

een

o-ese

rs; itrtu-acee,

om-erdintter-s].nlyre

inis

—aon-is

ingw

cesnvi-esvaryhechtheal

explicitly. It is when these relationships are broken downthat we say that something is “out of place”.

One measure of placeness is the degree to which a place rein-forces—or even defines—the pattern of its context. But to bea place is also to be distinct from its context. How is it pos-sible for a place to be both “part of” and “apart from” itscontext? The tension is addressed by defining the distinctive-ness of a place in terms of the surrounding context— andvice versa.

This model of place, in the tension between connectednessand distinction, will turn out to be a valuable way to thinkabout and design places in computational space as well asphysical space. First, though, we will present work frommedia spaces, to illustrate the distinction between place andspace, and to ask, how do we make spaces into places?

5 Making a Place in Media SpaceWe have experimented with these ideas of space and place inresearch over the last ten years into “media spaces” [Gaveret al., 1992; Bly et al., 1993]. Media spaces integrate audio,video and computer technology to provide a rich, malleableinfrastructure for workgroup communication across time andspace. This work—especially recent reports bringing a long-term perspective (e.g. Harrison et al., forthcoming; Dourishet al., in press), presented in examples below—vividly illus-trate the distinction between place and space.

It’s no accident that these experimental audio-video environ-ments are called Media Spaces, not Media Places. Placenessis created and sustained by patterns of use; it’s not somethingwe can design in. On the other hand, placeness is what wewant to support; we can design for it. Media spaces wereintended to provide the structure from which placeness couldarise, just as places arise out of the space around us. Theywere not designed as places themselves, but for people tomake places in them.

To understand how this works, we need to spend some timethinking about how people turn a space into a place.

5.1 Adaptation and AppropriationOne critical element in the emergence of a sense of place andappropriate behaviour is support for adaptation and appro-priation of the technology by user communities. This appliesto physical places as well as technological ones. We make ahouse into a home by arranging it to suit our lives, and put-ting things there which reflect ourselves. People make placesin media spaces with just the same ideas of adaptation andappropriation. Like tacking pictures to the walls, rearrangingthe furniture or placing personal artifacts around a room,these are the ways that people can turn a space into a place.

5.1.1 Example: Linking Public SpacesAs an example, consider the very different experiences oftwo seemingly similar uses of video to link public spaces,one at Xerox (reported by Olson and Bly [1991]), and one atBellcore (reported by Fish et al. [1990]).

Both experiments linked public spaces in R&D office envi-ronments with audio and video, to foster informalcommunication. However, the groups had very different

experiences of the successes and failures of their contions. The Bellcore researchers were disappointed with thresults, concluding that “the current VideoWindow systelacks something due to factors we do not understand” [Fet al., 1990, p.9]; the PARC researchers concluded “media space offered something wonderful to those of us wexperienced the Palo Alto-Portland link” [Bly et al., 1993, 45].

We believe that one critical factor contributing to the vedifferent patterns of use is this ability to participate, adaand appropriate. In these experiments, the differences their roots in the technology used. The Xerox link used retively inexpensive cameras, which were mounted wheeled tripods. Anyone could pick them up, move thearound and play with them—and many people did. On tother hand, the Bellcore system used a prototype wide-scamera array and video projection system. The equipmwas designed to simulate copresence as closely as posswith high-quality video and audio, and life-size imageHowever, the result was that the forbidding equipment, coplex and delicate to configure, could not be appropriatedits users. It wasn’t theirs, and they could not make it theirs.This separation between users and technology could be sto inhibit the community’s adoption of the technology.

It is only over time, and with active participation and apprpriation, that a sense of place begins to permeate thsystems. The sense of place must be forged by the usecannot be inherent in the system itself. Space is the opponity, and place is the understood reality. Just as spprovides the underlying opportunity for a media spacplace-making provides its realities.

Since the sense of place takes time to develop within a cmunity, we look for it in studies of long-term use. A numbof studies of “virtual shared offices” linked by video anaudio over the long-term (periods of two or three years) poto the emergence of place-centric behaviours and characistics [Adler and Henderson, 1994; Dourish et al., in presWe found that new patterns of behaviour emerged, not obetween the “direct” participants—those whose offices welinked by the media space—but also, critically, by othersphysical or organisational proximity. We will discuss thfurther in our section on “Hybrid Spaces”

6 Place as a Cultural PhenomenonWe have been developing the idea of a sense of placecommunally-held sense of appropriate behaviour, and a ctext for engaging in and interpreting action. This essentially a cultural phenomenon.

These understandings develop within cultures, and learnthem is part of our assimilation and socialisation. Like nemembers of any culture, new arrivals in our media spalearn the cultural norms and mores of the media space eronments, as part of their enculturation into the workplacand organisations where they are situated. These norms from place to place. For example, Dourish [1993] details tvarying views of media space activity in different researenvironments, and shows how these have influenced development of the technologies. Similarly, as cultur

4

dns In

ne-e iti-ts)byal

d inri- in

cesro-nd ofan

e-ults-hend

the

uteepha-

ues.

ndcial

ech-

newrva-erch

ort-lly

’t itndwn

iniprac-

ost of

understandings drift, change and are forged anew, we see thesense of appropriate behaviour change as well.

One illustrative example of cultural understandings ofappropriate behaviour in media spaces lies in the varyingapproaches which have arisen around concerns of privacyand control.

6.1 Privacy Concerns in Media SpacesQuite rightly, much time and effort, including much of ourown, has been devoted to these concerns (e.g. [Dourish,1991]; [Bellotti and Sellen, 1993]; [Clement, 1994]). It’simportant to note, though, that the solutions to these prob-lems, which arise in situ to address local concerns, aredifferent in every setting [Dourish, 1993]. Concerns aboutprivacy, about the balance between control and availability,and potential approaches to tackling the problems, are notabsolute. They arise in a mesh of cultural, personal andorganisational issues in different locations, and so we see thenature and scope of solutions vary widely to reflect this.

We don’t raise this to dismiss concerns with privacy in mul-timedia environments—far from it—or even to classify themas purely “local” problems. Instead, we want to draw atten-tion to the relationship between ways of acting and behavingand the patterns of cultural associations.

The kinds of ideas generally raised in discussions of privacyin media spaces are, themselves, cultural understandings.Privacy has been a major concern for critics of media spaceand designers of related projects. However, in general, thedebate has not been conducted with a rich view of the mean-ing of privacy. A place-centric view emphasizes importantdistinctions: privacy is not the same as private events, nor isit a direct consequence of private places; and in contrast,being seen or even heard is not absolutely and inevitablypublic.

6.1.1 The Social Construction of PrivacyThere are many dimensions to the notion of “privacy”—con-venience, turf, control of embarrassment, and control ofinformation. But let us start with the kind of privacy thatmost people think of first, a relatively recent invention inEuropean cultures. Consider the bedroom. Today, we com-monly think of the bedroom as a private place. We believethat the activities we associate with it should be visually andacoustically segregated from other people and other activi-ties. The bedroom is a place of intimacy, and is emblematicof the concept of privacy.1 This was not always so. The bed-room dates from the end of the Middle Ages when Europewas in a mini-ice age. Up until then, people ate and slepttogether in large groups in a single room. Then someoneinvented the bed. It raised bodies off the cold drafty floor andseparated people from one another. Enclosing canopies wereadded that made tent-like rooms and created separate placesfor the now separated sleepers; these tent-like rooms evolvedinto bedrooms.

It is difficult to say whether the concept that is now calleprivacy was an unfulfilled desire waiting for these inventioor the by-product of keeping warm and displaying status.any case, the roots of privacy are in physical form, not in anabstract notion of the control of others’ visual access to oself and one's property. (Of course, this is not how we setoday.) Privacy is relative, not a set of psychological primtives. Technology (such as walls, doors or permission lisis not the only way to create privacy, nor is it enough itself. Social convention gives meaning to the act of visuseparation. For example, the PARC Media Space is rootethe open studio of the architectural office, a place where pvacy has a different meaning and is created and useddifferent ways than in the closed spaces of research offior the bedrooms of today. The RAVE media space at EuPARC emerged with a different set of technological aorganisational aesthetics, and manifests a different viewprivacy again, rooted in a “service” model rather than “open access” one [Dourish, 1993].

6.2 Cultural and Technological StructureThe identification of “placeness” as a cultural phenomnon—or, at least, one rooted in human social action—resin a critical implication for the design of collaborative systems and technologies. It shifts our focus away from ttechnology of place, since that technology—doors, walls aspatial distance—only gives rise to “placeness” through way in which it is given social meaning.

Office doors in our workplaces are typically left open, bsome doors carry signs to explain that they’re closed to kout noise, not visitors. The presence of these signs empsises the relationship between technology and social cThey reinforce the social meaning (availability) even in thepresence of conflicting physical configurations (the closeddoor). Technological configurations of private places aquiet ones are the same; they are distinguished by soaction, not spatial structure.

The relationship between space and place is social, not tnological. CSCW tools and technologies create new socialplaces, based on the ways in which their users ascribe social meanings to new technological features. This obsetion raises important questions for design. Carrying ovtechnological arrangements which ape the real world, suas spatial organisation, might give us a convenient shhand for establishing shared social meaning; but is it reathe most appropriate means? And, furthermore, doesnlimit the ways in which individuals and groups can adopt acreatively appropriate the technologies to create their onew meanings?

In the next two sections, we will explore these questionsmore detail. In particular, we will explore the relationshbetween space and place—their dependence and intetion—by looking at two complex forms of places: space-lessplaces and hybrid physical/virtual spaces.

7 Complex Forms: Space-less PlacesThe distinction between “space” and “place” is perhaps mstrongly demonstrated by examples of the emergence

1. Lerup [1987] notes that television “soap operas” often set action in bedrooms when characters share intimate thoughts. The bed-room is an icon for the private and the personal.

5

edia thee-ichti-

ar-

s-er vir-ehateec-

mensdialy

ep-chesSold

n,

inga-en forheonsn

of

gstr-yal

ldarttwoandhere

n-ing

place without notions of space. Earlier, we introduced theconcept of place as space invested with social meaning. Thespaces can be computational, as well as physical. Whatremains is the tension between connectedness and distinc-tion which leads to placefulness. As our first “complexform”, we present two examples in this section of placefulcollaborative action without a model of physical space.

7.1 Placeful Discussion without Physical SpaceOne obvious source of such examples are USENET newsgroups and Internet mailing lists. The technology of eachUSENET group is exactly the same, and yet the resultantgroups exhibit very different notions of place. It’s not simplythat they separate discussion into topics, making certainpostings appropriate to one group or another; but that theyalso make distinctions between styles of posting. Neophytequeries may be more or less appropriate, depending on theculture of the group; so are flames. These styles are relativelyindependent of topic. Complaints about spelling or grammat-ical errors are acceptable (or even encouraged) inalt.peeves , but they would be inappropriate incomp.protocols.tcp-ip .

The different groups serve different purposes to overlappingconstituencies and communities; and they exhibit differentsocial norms. They’re different places. This placeness buildsupon the tension between connectedness and distinctionwhich we raised earlier; but, critically, it emerges without anunderlying notion of space.

7.2 Placeful Navigation without Physical SpaceDourish and Chalmers [1994] discuss various models of nav-igation through information, and draw distinctions between“spatial”, “semantic” and “social” navigation. Social naviga-tion is navigation through information collections on thebasis of information derived from the activity of others. Thisis a particular case which spatially-based models aim to sup-port [Sawyer and Mariani, 1995]; drawing on therelationship between proximity and activity discussed in sec-tion two, these systems allow users to move to areas whereothers are clustered, to join the crowd and see what’s goingon. However, as Dourish and Chalmers illustrate, similarpatterns of social navigation also occur through “personalhotlists” on the World Wide Web, as well as through inter-est-matching systems such as Ringo/HOMR and GroupLens[Shardanand and Maes, 1995; Resnick et al., 1994]. Again,this demonstrates that place-based behaviour doesn’t needspace to underpin it.

The behaviours exhibited here—varieties in conversationalstructure, and navigation according to others’ interests—arethe same sorts of behaviours which spatial models try to sup-port. However, as in the examples from our media spaceexperiences, we find that these are not spatially-organisedafter all; they show people responding to places, not spaces.

8 Complex Forms: Hybrid SpacesOur second complex form is the hybrids of physical and vir-tual space which technology can create, and the places whichemerge.

When we observe the emergence of a sense of place in mspace, a distinction arises between “spatial” features thattechnology might provide—visual access, proximity, movment—and the place-oriented aspects of interaction whmight arise there—formal and informal discussion, inmacy, a sense of ownership, and so forth.

A key feature of interactions in media spaces (or, more pticularly here, interaction over particular connectionsestablished in media spaces) is that they take place in hybridspaces. A hybrid space is one which is comprises both phyical and virtual space. These tend to be less common in othrelated systems. For example, when my avatar enters atual collaborative environment, then not only is thenvironment (the space the system creates) virtual, but wis projected into that space (my avatar) is virtual too. On thother hand, in a media space, while the “space” (the conntion between two people) is virtual, the projections are not.What I project into a media space connection is a view of (the real me) and my office (a physical space). My actioand behaviour in my real space are visible in the mespace; but in the virtual system, I act only by remotemanipulating my representation.

The reason that this distinction between projection and rresentation works is that the media space connection reaout to encompass everything in front of the camera. there’s more in the connection than simply the “virtuaspace” of the two monitors. When two offices are linketogether in a media space, then a hybrid space is created; itinvolves not only the virtual space of the media connectiobut also the real physical space of the two offices.

8.1 Acting in Hybrid SpacesDourish et al. [in press] detail a range of experiences arisout of their experiences with very long-term, semi-permnent audio and video connections (“office-shares”) betweparticular offices. Since these connections were in placea long time—at least two years, day-in and day-out—tusers could observe transformations which the connectiintroduced, not just in their own behaviour but, critically, ithe behaviour of other colleagues in organisational andphysical proximity too.

Two examples particularly illustrate the importance hybrid space in these connections.

Shared Office Etiquette. In the first, two office-share partic-ipants observed a “shared office etiquette” arise amonvisitors to their offices. When someone arrived in the dooway or office of one participant to talk to him or her, thewould begin their interaction by greeting not only the locparticipant, but also their remote partner, “present” acrossthe audio and video link. In other words, visitors woubehave in either office—a physical space—as if it were pof a shared office. Neither physical space was shared by persons, but the shared place which they occupied, which was acknowledged by visitors, was formed from thybrid of physical and virtual space in the office-shaconfiguration.

Seeing Out the Door. The second example involves a recofiguration of physical space for the purposes of manag

6

seonir

andessn’tich

sy

d,tor.y two

by Atly

iseour

ions

slo-ual

gyur wece

ual; it

ns;icaltheion

ints for

s.s;

e benon-cialties;es.

communication in the hybrid space. When one of the con-nections was first set up, the cameras were pointed directlyat the office-sharers, so each participant could see the otherand their immediate working area. However, because of theway one office was laid out, one of the cameras was placedon the end of a desk between the office’s occupant and thedoor. As a result, the view of that office available in themedia space was rather like the view into the office from thedoorway. While this gave the remote participant could seelocal office inhabitant, it did not provide a view of the door-way, or into the space beyond which meant that the remoteparticipant could not see people who arrived in the office, orwho passed by outside. This turned out to be an importantfacility.

After a few months, the local inhabitant of this office volun-tarily re-organised it, turning it around 180°. This allowedthe remote participant to see not only of the office’s occu-pant, but also the doorway (and people standing in it) as wellas the public space outside the office. This hugely improvedthese vital hybrid interactions with office visitors and pass-ers-by (raising interesting questions of ownership andcontrol).

8.2 The Structure of Hybrid SpacesThe hybrid nature of media space connections is a criticalaspect of their use. This is why, for instance, media spacesand collaborative virtual environments do fundamentallydifferent things. Whereas I must enter a virtual environment(be it a 3-D rendered virtual reality, or a text-based MUD), Iuse the media space to create a new, hybrid space whichincludes real, physical me.

One of the curious properties of media space is that a placecan be made of hybrid spaces. Two people can be what theythink of as the same place (like an electronically sharedoffice), but will not be in the same physical space, nor evenwill they be the same hybrid space. My image on your mon-itor does not interact with your image on my monitor,although we can say the we each of us interact with the other.Each of us is in a separate space; linked, but not shared. Inthis section, we will talk in more detail about the structure ofhybrid spaces, and how to decompose it.

It is not only the structure of the space which we have todecompose. It’s very easy to blindly talk about “audio andvideo” in media spaces as if they were equivalent media, per-forming the same sort of function. However, when we takethe place-centric view—and as we have seen, it’s the place-centric view which affects how people communicate andbehave—then we can see that audio and video actually pro-vide very different sorts of functions.

8.2.1 Virtual Acoustic SpacesThe critical feature of the “virtual acoustic space” which amedia space can create is that it is all-pervasive. It fills thephysical space in a way which an image cannot. There aretwo aspects to this.

First, audio reaches out to encompass the participants; notjust those connected, but those around and passing through.The “open audio” aspect of the long-term media space con-

nections described by Dourish et al. was critical to their ufor just this reason, and was a highly significant contributito what they refer to as “communal” aspects of theconnection.

Second, the audio space is truly shared; we each speakhear in the same audio space. The sound of my voice carriover the audio connection and invades your space; it doestay in a fixed place until you attend to it. The space whthe audio channel creates is one which we share.

8.2.2 Virtual Visual SpacesIn contrast, visual space is not shared, but simply madeavailable. The image of my office which my camera sendout, and which is displayed on your monitor, remains mimage. It carries with it the context in which it was captureand at your end, it is framed and bounded by the moniYou can’t be in it, or walk around in it. You see me, in mspace, and I see you, in yours; but neither of us sees theof us, together, in a shared visual space.

This non-shared aspect of visual space is reflected Gaver’s [1992] “affordance” analysis of media spaces.number of the points illustrated there (and subsequendeveloped in later design work [Gaver et al., 1995]) arbecause it is not your space which enters mine, but yimage.

The value and interesting use of media space connectwhich we have observed lie in the balance between thesefeatures—the shared nature of the audio space, the trancated nature of the visual space, and the melding of virtand physical space which the media space affords.

We disrupt this balance at our peril. For instance, technolois available to let us digitally process the images from ocameras, and reconstruct them to make it appear thatwere sitting across from each other at a virtual “conferentable”. We would be being presented in a shared visspace. However, this space would belong to neither of uswould convey nothing about our current settings and actioand would cut out anyone else who happened by in physspace. It would be a very different sort of experience. In search for realism, the practical everyday value of interactin the media space’s hybrid space would have been lost.

9 Designing Around Space and PlaceAs we have gone through this discussion, a number of pohave arisen which are worth collecting together as rubricsdesign.

Spaces are not places. Spaces and places are different thingWe can all think of lifeless spaces in our buildings and citiespaces that “don’t work”, that have no sense of plac2.Spaces are part of the material out of which places canbuilt. Dealing with physical structure, topology, orientatioand connectedness, spaces offer opportunities and cstraints. Places, on the other hand, reflect cultural and sounderstandings. Places can also have temporal properthe same space can be different places at different tim

2. One reviewer commented, “Yes! The Stanford quad!”

7

t inace

ofithse

ts.eeple

serer-nes-

le)nt;om the

sta-h,xitsofhan

ln

allyntnt,-the

ri-ce,en-

toeirheitout

e

,er,and

l.,eder

cts,

While spaces have up and down, left and right, places haveyesterday and tomorrow, good and bad.

Places, not spaces, frame appropriate behaviour. Much ofthe motivation for spatially-organised collaborative environ-ments is that they can provide the cues which frame andorganise appropriate social behaviour in the real world.However, that behaviour is not associated with space, butwith place. A community hall might, on different evenings ina week, be used as a rock venue, a sports arena, and a placeof worship. On these different occasions, it’s not the struc-ture of the space which frames people’s behaviour, but theplace where they find themselves.

Places have social meaning. The meanings which placescarry are social meanings; they are rooted in the practicesand understandings of communities. They arise over timethrough as practices emerge and are transformed within thegroups. This has two consequences. The first is that differentgroups will have different understandings of similar placesand similar concepts, and these will change over time. Thesecond is that places have to be created, through practice andappropriation, to fit into the culture of the group. Placenesscan be designed for, but it can’t be designed in.

Different media have different spatial properties. The vari-ous media which we might use in creating collaborativeenvironments—streams of text or graphics, high-quality 3Dimages, audio, video, etc.—exhibit very different properties,and these properties in turn strongly influence patterns ofuse, adoption and adaptation to media environments. Forinstance, in the case of media spaces, we illustrated howaudio and video embody very different notions of “reach”and of “sharing”.

Our basic principle, stated earlier, captures all of thesetogether, and relates the distinction between place and spaceto our everyday experience: Space is the opportunity; placeis the understood reality.

9.1 The Utility of Space and PlaceThe question we must always ask in thinking about design isnot simply what aspects of space are being exploited in thesesystems, but to what end are they being used? By way of con-trast, one significant area of research interest around spatialmodels in interactive systems is to visualise large bodies ofinformation. Information visualisation techniques shift partof the information retrieval task from the cognitive to theperceptual system. Although the popularity of spatial meta-phors in collaborative environments follows their popularityin single-user interactive systems, the purposes are very dif-ferent. But we can still ask analogously, “What is it that isbeing made perceptible in collaborative spatial models?”

Place-making, then, would appear to be a complex enter-prise. It reflects the conscious arrangement of elements tocreate a space that accommodates activity, and (here is thehard part) the interplay of reflective design and happenstanceto give expression to the values of the occupants and theirwider community. In other words, as we have observed, aspace can only be made a place by its occupants. The bestthat the designers can do is to put the tools into their hands.Trying to do more—trying to build places—is not our job.

Let us now take a look at how this is and is not carried ouone kind of collaborative system, and how space and plinteract.

9.2 MUDs: Designing with Space and PlaceThrough this paper, we have been talking largely in termsmedia spaces. Not only are these the environments wwhich we are most familiar, but they have also been in ufor longer than many other collaborative environmenHowever, it is instructive to look at another example, to show the ideas we have been discussing apply. The examwhich we will look at, loosely, is MUDs.

MUDs employ a strong spatial analogy to manage multi-uinteraction. Connected regions (or “rooms”) serve as a filting mechanism; my view of the activities currently iprogress in the MUD is largely restricted to those activitiin the same room. Different MUDs (not just different implementations, but different services, run by different peopexploit this “real-world” analogy to a greater or lesser extefor instance, in some systems, characters can “teleport” frone room to another, whereas in others, one must walk todestination, passing through the points in between.

However, it’s a curious sense of geography which MUDexhibit. They have topology (connectedness) but no oriention; there is generally no real notion of up, down, nortsouth, back and forward (except in the names of the ewhich link rooms). Meantime, most MUDs have no notion space within a room; I can’t be closer to one character tanother, or hide behind the sofa.

So, in fact, MUDs do not exhibit the spatially-based controwhich might seem central to them, and which is ofteappealed to by developers. The spatial metaphor is actuof much less value in controlling interaction, engagemeand so forth than might be imagined. Where presecommon MUD facilities like teleporting or inter-room messaging undermine it even more. The spatial metaphor—connectedness of space, and the geography of the MUD—breaks down, and only the places remain.

Every MUD has its places—general gathering places, pvate places, homes, etc.—where the sense of plaestablished over time, within a community, is used to framappropriate action. The inappropriate action of new, uneculturated arrivals (“clueless newbies”) only serves reinforce how much the regular characters know about thsocial norms. But this has almost nothing to do with t“geography” and the familiar spatial metaphor—why, if did, then the clueless newbies should be able to work it right away!

MUDs, like more traditional virtual reality systems, arimmersive environments; users enter the MUD, and interactin the MUD, rather than using the MUD to link their ownphysical worlds in the way that media spaces do. Howevin recent experiments, researchers have added audio video conferencing facilities to MUDs. Jupiter [Curtis et a1995] is a multi-media extension to a traditional text-basMUD. It retains the traditional spatial analogy of thMUD—objects and users located in rooms, which filteactivity as a whole—and then adds shared graphical obje

8

lyendoust usthat ofv-ors,

se,n-toy isddar-

es.le,selingry

nti-to

ib-anding,

intin

h-

r-,

art

-o,

as well as audio and video streams. Like typed text, audioand video are available on room-by-room basis; each usercan see and hear others in the same room.

However, the result can be quite confusing. Following fromthe observations we made in section eight, Jupiter exhibits avery mixed metaphor. My virtual presence is in one of Jupi-ter’s virtual rooms, where I share with some other usersaccess to a virtual whiteboard and an acoustic space whichextends into all of our real offices. But at the same time, I canalso see representations of these people in their offices,typing at their computers. They seem to be in two places atonce. The information which their video signal projects intothe system denies the metaphor which the system is con-structed to provide. Since that metaphor is what unifies andunderpins all activity in the MUD, the result is confusing.

The analysis of Jupiter in supporting interaction and collab-oration in a large research facility is only just beginning.However, from our own experiences, we would suggest thatthe mixed metaphor results in Jupiter being used much moreas a media space than as a MUD, by those who have audioand video connectivity through it. The dissonance of virtualreality and multimedia representations is just too much tobear.

This reiterates a point we made earlier; that, when we thinkabout the difference between space and place, we can seethat the various media which might be involved in a commu-nication technology play very different roles. When audioand video are added to a textual interface, the result is defi-nitely not the same sort of system with more bandwidth.Instead, the result is a new kind of medium, and so the sortsof behaviours which people will exhibit are changedradically.

10 ConclusionsWhat we have pointed to here is the distinction betweenspace and place; and this distinction has important implica-tions for the design and evaluation of collaborativecomputational environments.

Spatial models are increasing popularity in the design of allsorts of collaborative environments. These designs are basedon the assumption that, since many aspects of our behaviourseem to be organised around spatial elements of the everydayworld, then we can carry over these patterns of behaviour tovirtual environments by designing them around the sameaffordances for action and interaction that the everydayworld exhibits—doors, windows, walls, distance, proximityor whatever.

However, we have argued that, in everyday experience andinteraction, it is a sense of place, rather than the structure ofspace, which frames our behaviour. Our sense of place is acultural or communally-held understanding of the appropri-ateness of styles of behaviour and interaction, which may beorganised around spatial features but is, nonetheless, quiteseparate from them. After all, as we have seen, non-spatialenvironments exhibit placeness, too.

The placeness which operates in these non-spatial environ-ments—and, we would argue, in spatial ones too—is an

evolved set of behaviours rooted in our ability to creativeappropriate aspects of the world, to organise it, and to bit to our needs. From this, then, we argue that it is dangerto confuse the notions space and place. By all means, ledesign interfaces based on spatial organisation and all comes with it; but at the same time, we must be waryclaims that this will support place-based “real world” behaiours. In fact, by embedding placeness in spatial metaphwe can accidentally undermine the very thing that makesplace work—the shared understandings of appropriate uand the social interpretation of cues in the physical enviroment. When my “virtual door” absolutely controls access my virtual presence in a media space, then the opportunitlost for an appropriate social interpretation of a “closedoor”. My ability to appropriate elements of the world anturn the into cues for availability disappears. This is the padox of design around spatial metaphors.

After all, a virtual world filled with virtual offices and virtualdesks isn’t populated by virtual people, but by real onDrawing contrasts and analogies between, for exampmedia spaces and the “real world” is unhelpful, becaumedia spaces are the real world. Their inhabitants are reapeople, engaged in real interactions in the course of dotheir real work. And, as such, they will engage in the vereal creation of forms of activity and work, just as they do itheir everyday physical environments. This is what it’s crical to design for; and this is what is lost when we fail support the duality of space and place.

Acknowledgments

We would like to thank the many people who have contruted to the development of these ideas over the years, have given us valuable feedback on this paper, includAnnette Adler, Victoria Bellotti, Sara Bly, Tom EricksonBill Gaver, Beki Grinter, Austin Henderson, AllanMacLean, Scott Minneman, Vicki O’Day and Bob Stults.

References

[Adler and Henderson, 1994] Annette Adler and AustHenderson, “A Room of our Own: Experiences from a DirecOffice-Share”, Proc. ACM Conference on Human Factors Computing Systems CHI’94, Boston, Mass., April 1994.

[Alexander, 1979] Christopher Alexander, “The TimelessWay of Building”, Oxford University Press, New York,1979.

[Alexander et al., 1977] Christopher Alexander, Sara Isikawa and Murray Silverstein, “A Pattern Language”,Oxford University Press, New York, 1977.

[Bellotti and Sellen, 1993] Victoria Bellotti and AbigailSellen, “Design for Privacy in Ubiquitous Computing Envi-ronments”, Proc. Third European Conference on ComputeSupported Cooperative Work ECSCW’93, Milano, ItalySeptember 13–17, 1993.

[Benford and Fahlen, 1993] Steve Benford and LenFahlen, “A Spatial Model of Interaction in Large VirtualEnvironments”, Proc. Third European Conference on Computer-Supported Cooperative Work ECSCW’93, MilanItaly, September 13–17, 1993.

9

ber

er-

ms

nd

y,

d

rk

u,

fdr

.

ni,or.

and

er,

-

[Borning and Travers, 1991] Alan Borning and MichaelTravers, “Two Approaches to Casual Interaction Over Com-puter and Video Networks”, Proc. ACM Conf. HumanFactors in Computing Systems CHI’91, New Orleans, 1991.

[Bly et al., 1993] Sara Bly, Steve Harrison and Susan Irwin,“Media Spaces: Bringing People Together in a Video, Audioand Computing Environment”, Communications of theACM, 36(1), January 1993.

[Buxton, forthcoming] Bill Buxton, “Living in AugmentedReality: Ubiquitous Media and Reactive Environments”, toappear in Finn, Sellen and Wilbur (eds.), “Video MediatedCommunication”, Lawrence Erlbaum.

[Carlsson and Hagsand,1993] Christer Carlsson and OlofHagsand, “DIVE: A Platform for Multi-User Virtual Envi-ronments”, Computer Graphics, 17(6), pp. 663–669, 1993.

[Clement, 1994] Andrew Clement, “Considering Privacy inthe Development of Multimedia Communications”, Com-puter Supported Cooperative Work, 2(1–2), 1994.

[Curtis and Nichols, 1994] Pavel Curtis and David Nichols,“MUDs Grow Up: Social Virtual Reality in the RealWorld” , Proc. IEEE Computer Conference, pp. 139–200,January 1994.

[Curtis et al., 1995] Pavel Curtis Michael Dixon, Ron Fred-erick, and David Nichols, “The Jupitert audio/videoarchitecture: secure multimedia in network places”, Proc. ofthe Third ACM International Conference on Multimedia, pp.79-90, November 1995.

[Dourish, 1991] Paul Dourish, “Godard: An Architecture forA/V Services in a Media Space”, Technical Report EPC-91-134, Rank Xerox EuroPARC, Cambridge, UK, 1991.

[Dourish, 1993] Paul Dourish, “Culture and Control in aMedia Space”, in Proc. Third European Conference onComputer-Supported Cooperative Work ECSCW93, Mil-ano, Italy, September 1993.

[Dourish and Chalmers, 1994] Paul Dourish and MatthewChalmers, “Running out of Space: Navigating in Informa-tion Spaces”, Adjunct Proceedings (Short Papers), HCI’94,Glasgow, August 1994.

[Dourish et al., in press] Paul Dourish, Annette Adler, Victo-ria Bellotti and Austin Henderson, “Your Place or Mine?Learning from Long-Term Use of Video Communication”, toappear in Computer-Supported Cooperative Work: An Inter-national Journal.

[Fish et al., 1990] Robert Fish, Robert Kraut and BarbaraChalfonte, “The VideoWindow System in Informal Commu-nication”, in Proc. ACM Conf. Computer-SupportedCooperative Work CSCW’90, Los Angeles, Ca., October1990.

[Gaver et al., 1992] Bill Gaver, Tom Moran, Allan MacLean,Lennart Lövstrand, Paul Dourish, Kathleen Carter and BillBuxton, “Realising a Video Environment: EuroPARC’sRAVE System”, in Proc. ACM Conference on Human Fac-tors in Computing Systems CHI ‘92, Monterey, Ca., May1992.

[Gaver, 1992] Bill Gaver, “The Affordances of MediaSpaces for Collaboration”, Proc. ACM Conf. Computer-Supported Cooperative Work, Toronto, Canada, Novem1992.

[Gaver et al., 1995] Bill Gaver, Gerda Smets and Kees Ovbeeke, “A Virtual Window on Media Space”, Proc. ACMConference on Human Factors in Computing SysteCHI’95, Denver, Co., May 1995.

[Giddens, 1984] Anthony Giddens, “The Constitution ofSociety”, Polity Press, Cambridge, 1984.

[Goffman, 1959] Erving Goffman, “The Presentation of Selfin Everyday Life”, Penguin, New York, 1959.

[Greenhaugh and Benford, 1995] Chris Greenhalgh aSteve Benford, “MASSIVE: A Collaborative VirtualEnvoronment for Teleconferencing”, ACM Transactions onComputer-Human Interaction, 2(3), pp. 239–261, 1995.

[Harrison et al., forthcoming] Steve Harrison, Sara BlScott Minneman and Susan Irwin, “The Media Space”, toappear in Finn, Sellen and Wilbur (eds.) “Video MediateCommunication”, Lauwrence Erlbaum.

[Lakoff and Johnson, 1980] George Lakoff and MaJohnson, “Metaphors We Live By”, University of ChicagoPress, Chicago, 1980.

[Lerup, 1987] Lars Lerup, “Planned Assaults”, MIT Press,Cambridge, MA., 1987.

[Resnick et al., 1994] Paul Resnick, Neophytos IacovoMitesh Suchak, Peter Bergstrom and John Riedl, “Grou-pLens: An Open Architecture for Collaborative Filtering oNetnews”, Proc. ACM Conference on Computer-SupporteCooperative Work CSCW'94, Chapel Hill, NC., Octobe1994.

[Root, 1988] Robert Root, “Design of a Multi-Media Vehiclefor Social Browsing”, Proc. ACM Conf. Computer Supportfor Cooperative Work CSCW ‘88, Portland, Oregon, 1988

[Sawyer and Mariani, 1995] Peter Sawyer and John Maria”Database Systems: Challenges and Opportunities fGraphical HCI”, Interacting With Computers, 7(1), pp273–303, September 1995.

[Shardanand and Maes, 1995] Upendra Shardanand Pattie Maes,” Social Information Filtering: Algorithms forAutomating `Word of Mouth”, Proc. ACM Conference onHuman Factors in Computing Systems CHI'95, DenvCO., May 1995.

[Stoner, 1991] Jill Stoner, “The House of Montisi”, Places,7(2), 1991.

[Tuan, 1977] Yi-Fu Tuan, “Space and Place. The Perspective of Experience”, University of Minnesota Press,Minneapolis, MN, 1977

[Whyte, 1988] William Whyte, “City: Rediscovering theCenter”, Doubleday, New York, 1988.

10