From Interface to Interplace:
The Spatial Environment as a Medium for Interaction

Thomas Erickson

Human Interface
Advanced Technology Group
Apple Computer, Inc., Cupertino, CA 95014

(now at) snowfall@acm.org

0. Abstract

Today's human-computer interfaces are cumbersome, sterile, and uninviting; they stand in stark contrast to the richness and depth of the everyday world. The thesis of this paper is that spatial environments have great potential as interface metaphors, particularly as computers begin to serve as a medium through which human-human interaction occurs. One section of the paper focuses on ways in which MUDs--text-based, multi-user dialog systems--use spatial metaphors to support social interaction. Then the paper examines how real spaces structure and enrich human interaction, drawing on observations from the literature on urban design, landscape architecture, and related disciplines. Ultimately, we hope that a better understanding of these issues can lead to the development of spatially-based interfaces which support human-human interaction.

1. Introduction

I believe people are beginning to use computers in a radically new way. As computing and communications merge, and as networks increase in their bandwidth and pervasiveness, we will witness the transformation of computers from simple work processors into a networked medium which supports communication, cooperative work, and social interaction. By simultaneously providing a framework for structuring social interaction, and a collection of props for facilitating and enriching interaction between individuals, computing as a social medium will have a far reaching effect on our daily lives.

The thesis of this paper is that spatial information theory can not only provide the underpinnings for effective Geographic Information Systems (GIS), but can also provide the foundation for this new use of computing technology. Just as spatially based interfaces are natural candidates for supporting the retrieval, manipulation, and understanding of spatial information, so are they likely candidates as interfaces for computing as a social medium. However, such an application requires that we focus on different properties of space. In particular, we need to understand the properties of space which are entwined with human interaction. Although this investigation initially leads us a away from the theoretical terrain typically explored by designers and implementors of GIS, ultimately we will end up in the same place. For users do not retrieve and manipulate geographic information in a vacuum--ultimately their purpose is share it with others so that common goals might be achieved.

This paper makes a start at investigating the properties of spaces that enable them to serve as a frameworks for communication, cooperative work, and social interaction. To avoid getting lost in abstraction, I begin by giving an example of how a physical space functioned as an effective interface for coordinating the activity of a large number of people engaged in a single task. Next I turn to the realm of human computer interaction, and discuss an example in which space is used--in a purely metaphorical way--as an interface. Having established that space--both real and metaphorical--can play a positive role in supporting and structuring human interaction, we examine ways in which real spaces structure and enrich interaction. This is done by drawing upon observations from urban design, landscape architecture, and the sociology and anthropology of situated behavior. I conclude by suggesting that designers of interfaces for GIS may want to keep in mind that the interfaces they are designing for spatial information might ultimately evolve into interfaces of a much more general nature.

1.1 Space as an Interface

In December of 1989 I had an striking experience which, for me, came to foreshadow the way in which a user interface of the future might function. Oddly enough, no computer was involved.

I was participating in a workshop aimed at producing a book. About 30 authors had gathered to mutually critique papers for a period of three days. At the end of the workshop the organizers decided to try something unusual: all 30 authors would meet and jointly create an organization for the book from scratch--the authors would decide on what the book sections should be, and how the chapters should be ordered within them.

We gathered in a room, each author with a copy of his or her chapter. To start the process, someone had spread some pieces of paper on the floor, with (possible) book section names written on them, and authors were asked to put their chapters near an appropriate section. After this, the procedure was simple: anyone could pick up any chapter and move it elsewhere; anyone could change the name of a book section; anyone could propose a new section by writing a name on a new piece of paper.

Although it seemed like a recipe for chaos, and was in fact characterized by a lot of milling about and simultaneous conversations, the process was exceptionally effective. In about 30 minutes, 30 people had come up with an mutually agreed upon organization for a book of 30 chapters, with everyone participating in the discussion. It seemed to me to be a human analog of the long predicted agent-based computing systems, in which distributed agents, each possessing incomplete knowledge, cooperatively interact to accomplish a task.

The key to the success of this process lay in its spatial nature. Several phenomena seemed to be important:

First, it seemed important that the process was carried out in a bounded space, dedicated to the task. We could see activity as happened: it was clear when disagreements arose; it was evident when someone was making a major change; and we could tell when the organizational process settled down because the general level of conversation and movement settled down.

Second, the participants had assigned meanings to parts of the space. This was done by writing the names of proposed book sections on paper, and by positioning chapters relative to the proposed sections. This method of assigning meaning also allowed participants to convey ambiguity and novelty. Sometimes participants would position a chapter halfway in between two sections, indicating that they weren't sure to which it belonged. Sometimes participants would put chapters in an area of the room without a section name, suggesting that those chapters belonged together but fit no existing sections.

A consequence of having a shared, meaningful space was that actions in that space were meaningful, and often triggered discussions and explanations. For example, when someone went to move a chapter to another area of the room (i.e. assign it to another book section), there would be one or more people around (the audience) with whom the mover would discuss the rationale for the move. The consequence of this discussion was that either:
a) mover and audience would agree on the move,
b) mover and audience would decide to change the name and definition of the category so that the chapter better fit where it was, or
c) the audience would convince the mover that the chapter was indeed in the right place. In all of these cases, a result was that there was a greater shared understanding of the section names and definitions, and of the gist of each chapter.

Finally, physical constraints shaped the way in which people could participate in the organizational process. The fact that the chapters and section names were spread all over the floor had an important impact: it meant that no one could dominate the organization of the book. Those who had strong opinions about where their chapters belonged, tended to hover near their chapters, ready to 'defend' them from would-be reorganizers; those who had ideas about the arrangement of the book as a whole had to flit about from section to section, thus giving up any strong control over where their chapter (or any single chapter) was positioned.

There were a variety of other elements of the situation which influenced the success of the interaction, but these are sufficient to establish the point that properties of space can usefully structure and facilitate interactions.

Now, this is very nice, but the skeptical reader may wonder to what extent this applies to computer interfaces. Is this another paper talking about how wonderful virtual reality will be? No. While virtual reality may be an important interface medium of the future, I want to focus on interfaces in general. I claim that a better understanding of how properties of space support human interaction has direct applications to the more mundane, two-dimensional interfaces with which we interact today. To make this point, we turn to an example of a spatial interface which is entirely metaphorical.

2 MUDs: Virtual Places as Frameworks for Interaction

In this section we examine MUDs. MUDs are examples of systems which use space as a way of organizing information and coordinating activity. They are particularly interesting because they do this entirely through metaphor: no spatial representations of space (either 3-D or 2-D) are used; the entire interface is textual, and yet MUDs manage to draw upon some properties of space of facilitating interaction.

2.1 What are MUDs?

MUD is an acronym which stands for Multiple User Dungeon, though the D is mapped to many words, including Dimension, and Dialog. In technology-centered terms, a MUD is a multi-user database typically used as an infrastructure for gaming or chatting; MUDs may also be thought of as text-based virtual realities, or multi-user interactive fiction programs. But these descriptions all fall a bit short of the mark. The key attribute of MUDs is that they are environments which support social interaction: people can converse, chance meetings can occur, 'rooms' can be constructed, 'objects' can be created and interacted with, and events of all sorts can be organized and conducted. MUDs act as social media, providing frameworks for social interaction, and collections of objects which catalyze, facilitate and enrich interactions. Although MUDs have primitive interfaces, and allow interaction only in the most cumbersome way, I believe they are the precursors of a new way of using computers.

2.2 The Origin of MUDs

MUDs grew out a genre of text-based adventure games, popular in the 1970s on time-shared minicomputers. These were text-only games in which the player explored a dungeon, hunting for weapons and treasure and battling monsters, all by typing in simple command strings. Success at these tasks increased the player's score, which translated into increased rank and power. A typical dialog between the user and the adventure game might look like this:

User: Look
Game: You are in a large room; there is an arched doorway to the north, and a sword lying on the floor.
User: Pick up sword
Game: You pick up the sword. You feel a tingling in your arm. The sword appears to be enchanted.
User: Go north
Game: You are in a round room. There are doors to your left and right.

And so on.

MUDs evolved from this type of game, the first being implemented in 1979. In such adventuring MUDs, in addition to obtaining weapons, treasure, and killing monsters, one could encounter other players--potentially logged on from other machines on the Internet--and converse with them, go on adventures with them, or kill them, as inclination dictated. Often, it was the latter. In 1989 James Aspnes of Carnegie Mellon University constructed a MUD without monsters or weapons or scoring, a place where people could gather to talk and extend the MUDs structure by using a built in programming language (see [3] for a more extensive history).

2.3 MUDs Today

Recently, MUDs have captured the interest of researchers in computer science, media, communications, and human computer interaction. In particular, much interest has been stimulated by the creation of MUDs at Xerox PARC [4] and the MIT Media Lab [2]. Although there are a variety of types of MUDs in existence, I will focus on conversation-oriented MUDs, rather than those oriented towards adventuring; most of my examples will be drawn from MediaMOO, the MUD at MIT.

MUDs have become quite popular, in spite of the fact that their interfaces are cumbersome. For example, Curtis [4] estimates that about 750 players connect to the PARC MUD every week, with 12 to 35 being connected simultaneously at any time of day. In the first year and a half, over 3500 different players connected from over a dozen countries. More generally, there are 200 to 250 MUDs active on the Internet; Curtis calculates that with a conservative estimate of 100 players per MUD per week, there are at least 20,000 people using MUDs every week.

For a more complete description of MUDs and related programs, the reader should see [4] and [3]; Bruckman and Resnick [2] provide a description of MediaMOO. For our purposes, it will be sufficient to describe some basic attributes of MUDs, and show what a typical interaction would look like.

2.4 Characteristics of MUDs

. MUDs oriented towards supporting conversation have three characteristics:

MUDs have Geographies :
MUDs use geographical metaphors. LamdaMOO, the mud at Xerox PARC, is described as a rambling old house with many rooms. Other MUDs may be based on the metaphor of a medieval town, a dungeon, or of a fictional world if that MUD is oriented towards role-playing. MediaMOO, the MUD at MIT, is unusual in that its geography is partially based on the real world. Its core is a model of the Media Lab building, and--through connections over a virtual Internet--other areas are based on the layout of other Internet sites, Apple's R&D building being one example.

MUDs are Extensible by their Participants :
Most conversational MUDs allow their participants to add to the MUD's structure. In these MUDs, players can build their own rooms, create objects that other players could interact with, and even program autonomous (ELIZA-like) agents which roam around the MUD, accosting or otherwise interacting with other players. Note, again, that this is all done via text. 'Building' a room or an object or an agent simply means writing a program that gives appropriate textual responses to textual actions: 'looking' at a room, agent, or object displays its description; doing some other sort of action produces some textual description of what happens.

MUDs Support Complex Social Interaction
MUDs have social hierarchies, which are usually based on knowledge: more experienced participants know the structure of the mud, and can help newer participants find places, program things, and so on. Participants often demonstrate their prowess by creating neat spaces and objects. A variety of complex social behaviors have been reported on MUDs. These range from the formation of friendships and romances, to cases in which participants pretend to have different genders, appearances, and personalities. These, and other interesting social behaviors, are described in Bruckman [2] and Curtis [4].

An Example of an Interaction on a MUD

The text below is a simulated interaction between a user, 'Tom' and the MediaMOO MUD program. The text resembles the script of a play, with actions and comments by Tom preceded by "You", and actions by other participants preceded by their names; the rest of the text (unlabeled) is generated by the MUD program: it usually consists of descriptions of spaces or objects in the MUD. Text in curly brackets {} are commands typed by Tom, and are not seen by other players.

Amy is here.
Amy says, "now you can go to com, and then to apple."
Amy says, "there's also a router that'll jump you around."
Amy says, "type 'help internet' for more info."

{Say Thanks much! See you later.}
You say, "Thanks much! See you later."

Amy says, "see you around..."
Amy steps into The Root Lounge.

In the above script, Amy, another participant, has just appeared. Knowing that Tom is new, she's telling Tom how to travel from MIT to Apple, by using the MUD's virtual Internet. Tom uses the "Say" command to reply. (This script would look different on Amy's display: instead of "Amy says", she would see "You say", and so on.) In addition, anyone in the current room or space, can 'hear' the conversation.

The next fragment of script shows text generated by the MUD that describes what the participant, Tom, sees, and what he may do. Then it shows Tom specifying an action ("out"), and the MUD responding with a new description of what is happening to Tom (what another participant, if present, would see), and what Tom sees:

You see the back of a computer screen here.
Obvious exits: down to com and out to Apple Computer R&D Atrium

{out}
Your packets gather in a glob, and then flow into the screen! You feel yourself rematerializing.
Apple Computer R&D Atrium
You are in a glass atrium, four stories tall.
Offices look out from the walls.
Obvious exits: jeremy to Jeremias' Office

In this fragment of script Tom has traveled through the Internet, and rematerialized in the Apple Computer R&D Atrium (this description matches the appearance of the real Apple R&D Atrium). After describing what the participant sees, the MUD describes possible actions (above, typing "jeremy" will go to Jeremias' Office), as well as describing objects which may be interacted with, as shown in the next bit of script:

You see an ancient Apple I board hooked up to a monochrome monitor here. You see a sign here.

{look sign}
If you want to connect your office to the atrium, contact jeremias (Jeremy Bornstein).
[You finish reading]

Here Tom has read the sign, which contains text that another participant (Jeremias) has previously entered. Note that just as the R&D Atrium in the MUD has a real world counterpart, so 'jeremias corresponds to Jeremy Bornstein, the person who has control over what can be added to this part of the MUD).

This is enough of a description to get across the most basic things that happen in a MUD: the user can converse with other participants; the user can move through the MUD, seeing new parts of it; and the user can interact with objects in the MUD.

2.5 Ways in Which MUDs Encourage Interaction

MUDs use several interesting techniques to encourage interaction. One is to sponsor social events. For example, MediaMOO was officially opened with an inaugural costume ball on January 20, 1993, timed to coincide with the Clinton inaugural ball in Washington D. C.. Over 60 people from 5 countries logged onto MediaMOO during the ball, and participated in the festivities.

In this case, a second method for encouraging social interaction was used: spaces tailored to particular kinds of interactions were constructed (that is, the participants understood the meanings of the spaces). For example, a ballroom was constructed, with a dance floor and orchestra. When on the dance floor, participants could issue a "dance" command, and the MUD would generate a description of them dancing (often a description that was amusing or unusual). Other spaces were a bar, a lounge, a dressing room, all of which suggest particular types of action or interaction.

Another interaction generator is the participant's need for information. MUDs work in spite of their interfaces, not because of them. Although there is a help system, the easiest way to find out how to do something is to ask someone else. For example, it was possible for participants to teach the orchestra to play a particular song; the easiest way to learn this was by asking others who already knew how.

I hope that this conveys a bit of the flavor of what a MUD is like. There are two things which are difficult to convey through this sort description: First, although having a textual conversation with someone is cumbersome, it feels very realistic--much more so than an email conversation. Perhaps this is because it is occurring in real time. A second factor is the overlap between MediaMOO and the real world: entering an area of the MUD where the text descriptions resonate with your knowledge of the real place is surprisingly engaging.

Regardless of your impression of what the experience of participating in a MUD is like, the key point is that even in this purely textual interface, properties of spaces have been used to structure, facilitate, and generate social interactions. It is quite easy to imagine other types of interfaces--using two or three dimensional representations of spatial environments--which could also benefit from a greater understanding of ways in which spaces can support interactions.

3 Properties of Space Which Affect Interaction

The goal of this section is to examine ways in which real spaces function to structure and enrich human interaction. While there is little consensus on design principles for real (or virtual) spaces, my hope is that this examination will alert us to some of the untapped possibilities of spaces, and to suggest some productive directions for design explorations.

3.1 Objects can Generate and Catalyze Interactions

One of the ways in which spaces can be enriched is by what I call evocative objects. An evocative object is one that encourages people to interact with it. My favorite example is a sculpture in Seattle titled Waiting for the Interurban. The piece consists of five life size figures representing ordinary individuals (e.g., a workman with a lunch pail, a mother and child, a businessman, etc.). The figures are in various attitudes of waiting, and are, in fact, located beside the tracks at a train station. What is remarkable is the way in which people interact with the sculpture:

the public response has been hearty enthusiasm. People have climbed on the statues, placed flowers in the woman's hand, kissed the workman's cheek (as revealed by a smear of lipstick) and strapped a leash to the dog as required by city law. In cold weather, the sculptures receive scarves and warm hats; on holidays like Halloween and New Year's they are decked out in masks and party outfits; and when Mt. St. Hellens erupted, they were given dust masks. [6]

The ability of an object to capture people's attention, and to encourage interaction with it is remarkable. It would be fascinating to see a study of pieces of public art, and an analysis of why some seem to evoke activity, and others seem to be sterile. Understanding more about this power of objects could have multiple applications. Museums could use it to design better exhibits; schools could create better demonstrations; retailers could design better displays; conference exhibitors could design more effective booths.

Notice also that, in this case, people are not just interacting with the object. They are also interacting indirectly with one another through the medium of the object: they are making comments on their shared experience, telling jokes, editorializing. As one Seattle resident says about Waiting for the Interurban: 'Whenever I go by the sculpture I smile: I go by the sculpture to see what's happening.' [6]

Objects can do more than just encourage interaction with themselves. Objects can also catalyze direct interactions between people. Another example from the domain of public art is a piece called "Dancers' Series: Steps." This is a series of eight bronze panels etched with diagrams of dance steps embedded in the sidewalk along a street. Each panel uses arrows and numbered outlines of men's and women's shoes to illustrate the steps for a particular dance.

It is not difficult to guess the reaction this sculpture elicits:

Each series of dancer's steps traces the process of human motion, [eliciting] from some viewers an irresistible urge to pause and mentally reenact the dance combinations, and from others a spontaneous spurt of dance fever! Those congregating around the steps or actually dancing attract other passersby, creating a chain reaction of participation and shared amusement. [6]

William Whyte, a sociologist who has studied how people interact in spaces, uses the term "triangulation" for this tendency of certain objects to catalyze interactions between people [18]. It is interesting to reflect upon the vast range of objects, spaces, and events in our culture whose basic purpose is to promote interaction between people. Everything from ice-breaking party games, to dinner parties, to scientific conferences are directed towards that end. And as demonstrated by MUDs, most of them seem perfectly appropriate for implementation in a virtual environment.

3.2 Spatial Constraints can Generate Activity

In an article on urban design, Christopher Alexander writes of the corner of Hearst and Euclid, in Berkeley, California [1]. Hearst and Euclid is an ordinary corner with sidewalks, a stoplight, a drugstore, and a newspaper machine in the store's entrance. As pedestrians wait for the light to change, they study the headlines and perhaps decide to buy a paper. When the traffic is moving, people wait and tend to buy papers; when the light turns red and traffic stops, pedestrians hurry across the street, and are less likely to buy papers. In a real sense, the traffic light is helping to sell papers by making people pause.

What makes the news rack-stoplight system function effectively is a variety of constraints. Obviously, physical constraints are operating: pedestrians don't want to get run over crossing against the light, and most would rather not have to run across the street. In addition, there are social constraints. While flagrant violations of traffic lights by pedestrians are not uncommon, neither is it uncommon to watch a pedestrian wait patiently for a light when no cars are in sight. Such social constraints are the reason, although it is a subtler effect, that we may say that the news rack helps people obey the traffic light: if there is something of interest, people are less likely to violate the relatively weak social constraints on obeying traffic lights. More generally, if an evocative or catalytic object is present, the stoplight increases the likelihood that the object will be noticed and will have its effect. For example, a particularly outrageous newspaper headline may cause one pedestrian to exclaim in disgust, prompting a bystander to agree.

Note that these constraints may have non-local effects. Traffic lights, because they cause pedestrians to bunch up, create a rhythm in the flow of people which may have an impact at some distance away. For example, Whyte [18] has observed that best predictor of whether people will stop to look in a store window is if other people are looking in it; thus, a traffic light acting to bunch up a sparse stream of pedestrians may act to increase the incidence of window shopping in the next block.

Spatial constraints can be powerful ways of generating activity. Designers of virtual environments might very well want to build in constraints. Although such behavior runs against the grain (someone who has just spent considerable effort making certain actions possible is not typically inclined to turn around and impose artificial limits), it may nevertheless be an important way of making a virtual space a more productive environment. Urban designers recognize that putting a high speed freeway through the center of town may transform it into a barren, impoverished place; interface designers may need to come to a similar recognition.

3.3 Spatial Elements may be used to Structure Activity

Even minor features of the physical environment can structure behavior in subtle ways. In a study of automated teller station (ATM) use, Marine [10] observed that people waiting to use an automated teller station typically left a gap between the head of the line and the person using the machine. This in itself isn't surprising: entering a secret code to withdraw cash is regarded as private activity. What is surprising is that the lines of users usually formed behind a crack in the pavement, which happened to be at a reasonable distance from the ATM. An obviously accidental environmental feature served to structure the behavior of ATM users.

This type of phenomenon is apparently well known to building contractors. Don Norman, a cognitive scientist who has done extensive work on human interface design, reports that he recently had a section of his driveway re-poured, and that the contractor doing the work suggested putting in a colored border between the old and new sections. The contractor explained that it would act as a natural boundary that people who used the driveway to turn around in would not venture beyond. Norman lives on a dead-end, popular beach street, and gets about ten turnarounds a day: he reports that it works. On telling this story to his neighbor across the street, his neighbor reported being told the same thing by his contractor, and pointed to the cement entryway to his brick driveway [12].

In addition to these apparently unconscious uses of environmental features, people quite commonly use features of the environment to their own ends. For example, in a study of ways in which children play in cities, Dargan & Zeitlin [5] cite many examples of how children have re-purposed spaces and objects: "There were dozens of games going on at any given time. Whatever was handy would be a base--a dead pigeon, a car, or your little brother could be second base."

The moral here is clear. People use features of spaces in unexpected ways. It is the profusion of objects, the apparent haphazardness and disorder of spaces, that makes them adaptable and usable.

3.4 Place: Space with Meaning

An important attribute of space is that people understand a lot about particular types of space--they see meaning in space. I like to use the word "place" to refer to space plus meaning. At the most basic level, people understand that particular types of places have very generic functions. Closets, drawers, and cabinets are for storing things; libraries are places for storing, exhibiting, and browsing information and artifacts; living rooms, ballrooms, and atriums are public spaces; bedrooms and offices are typically private spaces for individuals. MUDs use this type of knowledge of space to help participants know what to expect.

Other types of place meaning have not been exploited. Places often reflect their history. Urban designer Kevin Lynch describes a multitude of ways in which cities reflect their history, from deliberately constructed monuments, to traces of vanished city walls reflected in the city's grid of streets [9]. Similarly, objects, buildings, cities and landscapes can reflect their physical context: the forces which shape them, the local materials out of which they are constructed, the situated activities to which they are responses--this is called vernacular architecture and landscape. Such places feel as if they have grown out of their surroundings, and convey a unique sense of place (see [8]). Entwined with all this are meanings which the inhabitants project onto a place. A field where a famous battle was fought, the birthplace of a celebrity, or simply the first house built in a town become invested with meaning, and serve as concrete, shared symbols to those who live there.

3.5 Ritual and Place: Places can Suggest Types of Interaction

Closely associated with the fact that places have meanings, is that places often have activities associated with them. One way of capturing this is through the concept of ritual. Ritual is useful because it connects three important elements of human interaction: participants, repeated sets of actions, and artifacts or spaces. Various theories of ritual and its forms and functions have been developed by anthropologists studying religious behavior (e.g. [17]). More recently, the concept has been applied to every day behavior such as consumption [15], tourism [16], and leisure activities [11].

A good example of the importance of place and ritual can be found in a description of the redevelopment of the town of Manteo, South Carolina [8]. The designers approached the project by interviewing community members, and asking them to describe places at which they frequently spent time. They developed a map of the 'sacred structures' of the community that should be preserved. These sacred structures were not actually sacred (usually), but were places like the corner store, the local post office, even a parking lot. What made these places special were the interactions that took place: chatting with neighbors at the post office, negotiating business in the drug store. These were rituals in the sense that they were sets of particular actions tied to particular places, and carried out at particular times. Because of the rituals tied to these places, the community had come to identify itself with them.

As in the material world, ritual can support interaction in a spatial interface by providing frameworks which encourage strangers to chat, act in particular ways, and share information with one another.

4 Conclusion

Today's computer interfaces, whether they be to spreadsheets, spatial data, or MUDs, are primitive, cumbersome, sterile, and uninviting. They stand in stark contrast to the richness and depth of our everyday world. While there has been a little work on applying some of these ideas to human computer interaction (see [7] and [19], in addition to the work on MUDs already cited), we have barely scratched the surface. Understanding more about space, and the properties which enable it to generate, catalyze, and structure human interaction, is a promising direction for further work.

I'd like to conclude with one last perspective from urban design. Stuart Perez has written about what he calls incidental architecture, about "the fences, flowerpots and checkered table cloths; window curtains, mailboxes and trash; cars and vendors and the sounds of each," all the things which make up the unnoticed fabric of our everyday existence. He suggests that perhaps architects and urban designers ought to change the way they think about what they do.

What we plan and design--the facades, the bulk, the surfaces and detail--is the armature on which the sensory life of the city is built. Like an armature, our architecture succeeds or fails on its ability to support this stuff of life, not by cleaning it up and putting it away but by knowing what it is, understanding how it works and setting a place that makes room for it to happen. [14]

As those of us in computer science, human computer interaction, and related disciplines work with spatial information theory to realize our near term goals, it is important to recognize that the embodiment of spatial information in computing systems may ultimately serve as an armature that will support rich and engaging forms of human interaction.

5 Acknowledgments

Thanks to Amy Bruckman and her colleagues at the MIT Media Lab for creating MediaMOO and opening it to outside researchers. Thanks to members of the Apple Human Interface community, particularly the erstwhile urban design reading group, for many discussions of architecture, urban and landscape design, and its application to interface design. Thanks to Amy Bruckman and Jonathan Cohen for comments on a draft of this paper.

References

1. Alexander, C. A City is Not a Tree. Design After Modernism (ed. J. Thackara). New York: Thames and Hudson, Inc., 1988.

2. Bruckman, A & Resnick, M. Virtual Professional Community: Results from the MediaMOO Project. To be presented at the Third International Conference on Cyberspace, May 1993. (Available via anonymous FTP from media-lab.mit.edu: /pub/MediaMOO/Papers/MediaMOO-3cyber-conf.)

3. Bruckman, A. Identity Workshop: Emergent Social and Psychological Phenomena in Text-Based Virtual Reality. Unpublished manuscript, 1992. (Available via anonymous FTP from media-lab.mit.edu, /pub/ MediaMOO/Papers/ identity-workshop.)

4. Curtis, P. Mudding: Social Phenomena in Text-Based Virtual Realities. Proceedings of DIAC '92. Available via anonymous FTP from parcftp.xerox.com, pub/MOO/papers /DIAC92.

5. Dargan, A. & Zeitlin, S. City Play. New Brunswick and London: Rutgers University Press, 1990.

6. Fleming, R. L., Von Tscharner, R. PlaceMakers: Creating Public Art that Tells You Where You Are. Boston: Harcourt, Brace, Jovanovich, 1987

7. Hill, W. C., Hollan, J. D., Wroblewski, D. A., & McCandless, T. P. "Edit Wear and Read Wear." In Proceedings of the ACM Human Computer Interaction Conference. Austin, TX: ACM Press, pp. 3-9, 1991.

8. Hough, M. Out of Place. New Haven: Yale University Press, 1990.

9. Lynch, K., What Time is this Place? Cambridge: The MIT Press, 1972.

10. Marine, L., Study of the Opportunistic Use of Environmental Structure to Organize Behavior, Unpublished manuscript, 1990.

11. Moore, Alexander. Walt Disney World: Bounded Ritual Space and the Playful Pilgrimage Center. Anthropological Quarterly, Vol. 53, #4, 207-218, 1980.

12. Norman, D. A., Personal communication, 1991.

13. O'Guinn, T. C. & Belk, R. W. Heaven on Earth: Consumption at Heritage Village, USA. Journal of Consumer Research, Vol. 16, September 1989.
14. Perez, Stuart. Incidental Architecture. Places, Vol. 7, #1, Fall 1990.

15. Rook, D. W. Ritual Behavior and Consumer Symbolism. Advances in Consumer Research, Vol 11 (ed. Thomas C. Kinnear). Provo, UT: Association for Consumer Research, 279-284, 1984.

16. Theilmann, J. M. Medieval Pilgrims and the Origins of Tourism. Journal of Popular Culture, Vol. 20, #4, 1987.

17. Turner, V. The Ritual Process: Structure and Anti-Structure. Chicago: Aldine, 1969.

18. Whyte, W. H., City: Return to the Center. New York: Doubleday, 1988.

19. Wroblewski, D. A., McCandless, T. P. & Hill, W. C. Advertisements, Proxies, and Wear: Three Methods for Feedback in Interactive Systems. Natural Dialogue and Interactive Student Modeling[tentative title]. Heidelberg: Springer-Verlag, in preparation.