Position paper at the Workshop on Personalized and Social Navigation in Information Space, Stockholm 3/1998
This paper looks at how people's understanding of spatial metaphors can influence the social activities taking place in virtual interaction spaces. In particular we are interested in how spatial metaphors influence "social navigation."
Many shared information spaces and virtual communities use spatial terms to either structure the environment or to convey a specific sense of place to their users. In the physical world architecture and spatial design have a strong influence on how people interact. Similarly spatial metaphors used to describe virtual spaces show "social connotations" that influence how people behave in these spaces and what meaning people attach to these spaces. Space with meaning can become a "place".
Places have a specific feel to them and a community has a shared understanding of what behavior is appropriate in this place and what is not. Although social connotations don't always determine what kind of social behavior happens in a certain place, they at least have a strong influence on the type of community forming in a place and therefore on any social activity, including social navigation.
In this paper we describe our view of social navigation and why we think it is closely related to spatial metaphors. We point out the importance of social connotations of space to support social navigation and describe a pilot study on social connotations. We further outline an early attempt at a classification scheme for places according to social connotations.
Over the past few years the popularity of systems that allow people to interact with each other over the Internet has grown rapidly and a large number of tools like chat rooms, news groups, email lists, MUDs and MOOs are in widespread use.
Most of these tools use at least some concept of spatiality in they way they describe themselves and how they structure interactions. The reason for this design is that humans live and act in space and many social interactions are based on or influenced by spatial concepts. Our language shows the importance of spatial concepts and we frequently use spatial metaphors to describe non-spatial concepts (Lakoff and Johnson 1980).
Many of these systems have a very distinct feel to them based both on the spatial descriptions, as well as on the type of community forming in them. Community members learn what is appropriate and inappropriate behavior and develop a common understanding of how to behave and what to expect. These systems have gone beyond simply spaces -- they have become "places".
This common understanding influences all social activities in the place. We are especially interested in how it influences collaborative (social) navigation. Social navigation can be either indirect, when navigation is indirectly influenced by other users activities, or direct, when users actively guide one another or when a user points out information to another user. The social connotations of a place have an influence on navigation processes like these. Because of this influence we need to gain a deeper understanding of the connection between social connotations and spatial metaphors.
Social navigation encompasses activities where (at least) two users of an information system collaborate directly or indirectly in a navigational task. In this discussion we limit our observations to finding information in shared information spaces, like the World-Wide Web.
Social navigation evolved out of a necessity. The Web shows a lack of perceivable structure even today. Many Web users therefore started to create navigational structure in the form of pointer pages or "favorite links" pages. Such tools form small personalized information spaces that are generally available on the Web. When other users started to use such pages as navigation tools ("I need information on graphics tools so I should check our graphics designer's pointer list first") these pages became social navigation tools.
Other examples are email or news groups messages containing pointers to information on the Web. Until recently most mail and news clients did not support embedding URLs into messages. People therefore developed informal protocols to point out information to others. For example they placed URLs in a separate, intended line to make them easier to copy and paste into a Web browser.
Dieberger described several early examples of social navigation in (Dieberger 1997). Many of these examples of social navigation will soon disappear as newer communication tools make the sharing of Web pointers much easier. Today it is not only possible to directly click on Web pointers in emails, but many chat rooms allow users to share Web pointers, or even give guided tours. These tools provide real support for "direct social navigation".
The difference between indirect and direct social navigation stems from how people interact with each other in the social navigation process. Direct social navigation requires a form of community, because people have to be willing and interested in sharing information. This is not always the case (see for example Mark Guzdial's note on "Virtual other than communities" at http://guzdial.cc.gatech.edu/Commentary/noncom.html).
In competitive or time-pressed situations users probably won't bother to point out information or give a guided tour. On the other hand an information seeker who is time-pressured herself might rather try to ask somebody than to find information in the environment if this is appropriate behavior in the place she is in.
While direct social navigation involves direct communication between users, not all social navigation requires it. Indirect social navigation instead uses only navigational cues found in the environment. These cues are created or modified by navigational activities of an entire user population and there is no direct interaction between the originators and users of navigational cues.
A primitive example for indirect social navigation is a Web page hit counter. It indicates how many people found certain information interesting enough to go to a particular page. Obviously a hit counter does not define a community, but it is an anonymous trace a user population leaves.
A more elaborate example is the concept of read wear (Hill and Hollan 1992). Read wear causes information to show signs of wear when it is used a lot. Examples for read wear in the physical world are library books that look more used than others or a path through a meadow where the grass is shorter or trampled down.
Read wear was used very successfully in a virtual conference this author designed. The objective of read wear was to guide a sparse user population in a large environment towards places where it was more likely to meet other people. Indeed the read wear worked so well that most users clustered in only a few locations most of the time (Dieberger 1997).
Other examples of indirect social navigation systems are Alan Wexelblat's Footprints system (http://lcs.www.media.mit.edu/people/wex/Footprints1.1/), many recommender and voting systems - see for example (Resnick and Varian 1997; Terveen, Hill et al. 1997) or the Alexa plug-in for Web browsers. Alexa monitors usage paths of the user community and gives navigation recommendations based on this data.
Indirect social navigation is no guarantee for finding relevant material, except if the user population forms a closed community with common interests. In that case what is deemed interesting for the overall community is likely to be of interest to every member of that community.
Social connotations have a strong influence also on how users perceive and use indirect social navigation cues, because the feel of a place determines if users will think of looking for indirect social navigation cues and if they perceive them as accurate or misleading.
This discussion shows that we define social navigation as a relatively general concept. Dourish and Chalmers make the distinction between "spatial", "semantic" and "social" navigation and define social navigation as "navigation through information collections on the basis of information derived from the activities of others" (Harrison and Dourish 1996) see also (Dourish and Chalmers 1994). We understand this definition to encompass mainly what we call indirect social navigation and not guided tours or a direct pointing out of information.
We are interested in social connotations because all human social activities are influenced by the environment they take place in. Note that we consciously avoid the word "space" in the previous sentence. It is not the space itself that influences social activities, but people's shared understanding of the space. When people share an understanding of a space, then this space becomes a "place".
Erickson defines place as "space with meaning" (Erickson 1993), and Harrison and Dourish call a space "a place which is invested with understandings of behavioral appropriateness, cultural expectations, and so forth. We are therefore located in space but we act in place." (Harrison and Dourish 1996)
The common understanding of a place defines what types of social interactions can take place and which activities would be "out of place."
Especially direct social navigation happens in the context of a "place". Remember, that the difference between indirect and direct social navigation is in the existence of a community in the latter case, but that both types of social navigation are strongly influenced by the type of "place" they occur in.
For the design of social navigation systems this means we have to be aware of what shared understanding is required to make a space into a place. An essential part of this shared understanding are the social connotations spaces commonly carry with them.
Social connotations can be both culture and language specific but the understanding a community has of a place can be stronger than these cultural connotations. Social connotations determine what is considered appropriate and inappropriate behavior in a certain place and situation and what initial assumptions people have when entering a place for the first time.
As an aside we would like to point out a problem with most current virtual "places". Amy Bruckman argues (personal communication) that one of the shortcomings of current virtual communities and places is that it is impossible to determine the feel of a place without being "in" it, whereas a real world place often allows people to get a general feeling of a place already from the outside. For example it is very easy to see the difference between a biker hangout and a fancy restaurant even from the other side of the street. To achieve a similar effect in a virtual environment we would need to find a way to convey social connotations already from the outside of a virtual place.
In 1996 this author designed a virtual conference on Romanticism for Emory university. The conference was based on a MOO system and users used a client called "Juggler" to interact with each other (Dieberger 1996; Dieberger 1997). MOO systems (like MUDs) are virtual environments entirely described by text and they are commonly used for collaborative work (Curtis and Nichols 1993; Bruckman and Resnick 1995).
The Juggler system supported indirect social navigation through read wear on both all the connections between locations in the conference, as well as on messages posted to message boards. In addition the client software provided strong support for direct social navigation through a one-click tool to point out Web pages to another participant.
Our conference site consisted of two major areas, one for "serious work", and the other for "socializing". In the socializing area various toys and virtual gadgets (a paranoid android, a bookshelf, virtual muffins and tea cups,...) were available to entertain people, and to help them learn how to use the environment. The underlying assumption for this separation was that most people would not want to see such toys in the areas dedicated to scholarly discussion.
The serious area consisted of discussion rooms, each of which was dedicated to a certain discussion topic and had the related conference papers available in the room. The rooms each had a bulletin board associated with it but were otherwise empty. In addition the serious area also provided one general purpose discussion room.
Sometimes we scheduled meetings with a small group to discuss a specific question. When the time of the meeting approached members of that group would log in and head for the social area as a "meeting point". A free-wheeling discussion began already when only a few of the group members were on-line but although this was a serious discussion already, it took place in the social area.
What surprised us was that each time one of the participants eventually would suggest that "we are in the middle of our discussion already -- we should go over to the meeting room". The group then moved to the other room and continued the discussion there.
In the Juggler system there was no functional difference between that social area and the meeting room -- the only difference was in the social connotations they carried. One of the places was an open, comfortable place with a bookshelf, toys and other objects people could interact with. It was a place that people could feel comfortable in because they could somewhat personalize it.
The meeting room instead was an empty, sterile place without decoration. It was perceived as a more formal place, where "serious business" could take place. Frequently one of the group members would volunteer to log the meeting in the meeting room. Such a proposal was never made in the social area. We also had the impression that even the sentence structures and the language used in the meeting room were more formal and elaborate than in the social area.
We think that the need of the conference participants to move their discussion proves the importance of social connotations for collaboration and communication tools, especially for direct social navigation.
In a place that is perceived as informal a user will not hesitate to address another user and ask for help, whereas in a place where a formal discussion is taking place users will do everything to avoid disturbing the current speaker and use private communication channels, wait for their turn to speak up, or rather use other navigation tools available.
In shared information systems it is also a very common situation that a user has to ask two similar questions if the first answer was not completely satisfactory. The behavior in this situation is again determined by how users perceive the place this discussion takes place in. Both in a real place as well as in a virtual place it is not always appropriate to immediately address the next person with the same question.
So far we always talked about places that also were spaces. This is a simplification because although spaces can be places, places don't necessarily need to be spatial. An example is an email discussion list. Although such a community quite obviously is a "place" as in the definition of "an environment about which users have a common understanding" it is not a space, because there is no real spatial metaphor in use. Still this place does have social connotations that govern what behavior is appropriate and what is "out of place" -- see also (Harrison and Dourish 1996). In this paper we focus on social connotations of spatial metaphors and spaces and do not consider social connotations of non-spatial places.
Although we do have an intuitive understanding of social connotations for physical places, it is not necessarily true that the same social connotations exist in virtual space.
Presently we are conducting a pilot study to determine how social connotations might carry over from a physical space to a space modeled in a textual virtual environment. The pilot study consists of two questionnaires. Each questionnaire presents a set of 10 situations described by text.
We restricted the pilot study to textual descriptions because we are interested in social connotations of textual virtual environments. According to Barbara Tversky (personal communication) it has been proven that textual route descriptions through an environment can cause people to form the same mental models of the described space than direct experience of the environment. This and knowing how strong a mental image a good textual descriptions can create in our minds (for example in literature), let as assume that it is valid to study social connotations of spaces from textual descriptions.
The 10 situations in our survey each describe the need to obtain some information, which can be satisfied best by interacting with people in this place. Participants see a list of possible strategies on how to approach this task and have to rate them as appropriate or inappropriate. We also show a set of possible social connotations and ask participants to decide if they perceive these connotations from our description or not.
The first Web questionnaire was posted to the Web in early February and was completely filled out by 74 people (some situations were filled out by more participants).
The second questionnaire will use essentially the same situations, but rewritten to feel more like MOO room descriptions. This Web questionnaire will be presented only to experienced MOO users. By comparing the results of both questionnaires we hope to get a better idea which social connotations exist unchanged in both physical and virtual spaces.
Although the pilot study is only partially complete we found a few interesting results already in the first part of the questionnaires. In addition we received many comments from participants that pointed out important issues in our study.
The 10 situations we looked at were:
In most of these situations it is possible to get the desired information from navigation tools in the environment (signs, maps). We defined the situations as being under time-pressure hoping this would force participants to rather seek help from other people than look for signs and maps. As could be expected a few participants commented in each situation that they would rather look for maps.
One, maybe not surprising result was that the perception of what was considered appropriate and inappropriate behavior seems to be quite consistent for most situations, especially where the answer boiled down to a matter of politeness.
The questionnaire was anonymous but we posted it to a few mailing lists that are focused on HCI and user interface design. We assume that our participants are mainly professionals in the HCI and design field and this general agreement on appropriate behavior might be slightly biased towards "professional attitude".
For example about 60-80% of the participants considered it not appropriate to just walk up to somebody and ask a question without apologizing for the intrusion first (still 12-15% found this behavior appropriate in practically all situations). About 70-80% considered it appropriate to first apologize for the intrusion and then to ask the question (86% in the apartment, over 90% at the bus stop and in the cubicle maze). The only exception was -- not completely unexpected -- the rock concert / sports event, where 42% thought that a question without introduction was appropriate (43% inappropriate) and where a question with introduction was considered appropriate by only 74%.
For the physical world these results are not really surprising. It will be interesting to compare this result with the data from the second questionnaire. Many virtual communities have a very specific code of conduct and a strict social hierarchy that determines if, when, and how "newbies" (newcomers) can ask questions of old-timers.
Another interesting point was if it is appropriate to immediately ask another person when the first person asked doesn't give a satisfactory answer. In most situations this behavior was considered appropriate (by around 70% of the participants and even 89% at the bus stop) but only 46% of the participants considered this appropriate in a discussion and only 56% thought this was appropriate in a cafe.
At first we thought that some spaces are perceived as more polite or formal than others. While this is indeed true we later realized that the reason for this particular result might be a very different one. Both in the cafe and in the discussion the same people are likely to stay nearby over some time. In all other situations people will move around (reception, mall, rock concert) or be very anonymous (bus stop). A semi-stable group of people might be perceived as a temporary community and people probably try to avoid looking foolish even in such a group.
We would like to point out that this theory is not entirely bullet proof but has to be tested in an additional study. The result itself is interesting in any case because asking a slightly modified question or query is a typical reoccurring task in social navigation. The design of a place where questions are frequently asked has to keep this difference in mind. For example, this suggests it would be inappropriate to design a meeting space for technical support for a software product as cafe or discussion room.
In the questionnaire we also asked participants about the following social connotations:
Participants could also suggest additional connotations. The results for this part are (percentage of participants):
| Hotel | Mall | Recp. | Cafe | Libr. | Disc. | Tour. | Conc. | Offic | Aptm | |
|---|---|---|---|---|---|---|---|---|---|---|
| No connotation | 0.01 | 0.02 | 0 | 0.02 | 0.02 | 0.01 | 0.04 | 0.05 | 0.03 | 0.05 |
| Anonymous | 0.59 | 0.92 | 0.05 | 0.66 | 0.77 | 0.39 | 0.86 | 0.85 | 0.36 | 0.55 |
| Public Space | 0.82 | 0.96 | 0.19 | 0.79 | 0.48 | 0.13 | 0.92 | 0.8 | 0.03 | 0.08 |
| Cameras | 0.39 | 0.67 | 0.21 | 0.16 | 0.53 | 0.14 | 0.13 | 0.56 | 0.65 | 0.66 |
| Switch Off Cameras | 0.02 | 0.01 | 0.04 | 0.06 | 0.01 | 0.04 | 0.01 | 0.01 | 0.05 | 0.04 |
| Informal Space | 0.4 | 0.84 | 0.39 | 0.73 | 0.21 | 0.09 | 0.67 | 0.64 | 0.07 | 0.37 |
| Move Freely | 0.72 | 0.89 | 0.85 | 0.5 | 0.7 | 0.09 | 0.84 | 0.57 | 0.07 | 0.3 |
| Restricted Space | 0.09 | 0.02 | 0.75 | 0.16 | 0.43 | 0.73 | 0.01 | 0.23 | 0.93 | 0.78 |
| Know Most People | 0.03 | 0.01 | 0.62 | 0.04 | 0.01 | 0.24 | 0 | 0 | 0.01 | 0 |
| Know Some People | 0.39 | 0.32 | 0.76 | 0.42 | 0.43 | 0.89 | 0.07 | 0.37 | 0.57 | 0.4 |
| Rules | 0.74 | 0.53 | 0.81 | 0.65 | 0.83 | 0.86 | 0.33 | 0.4 | 0.84 | 0.44 |
| Safe | 0.5 | 0.45 | 0.73 | 0.55 | 0.54 | 0.56 | 0.16 | 0.2 | 0.49 | 0.32 |
| Want Tour | 0.02 | 0.01 | 0 | 0.01 | 0.04 | 0 | 0 | 0 | 0.09 | 0 |
One row in this table that really surprised us was the connotation "want to switch off cameras". We intended the camera to be a metaphor for a logging device in the physical space. With these results we are not sure any more if we can really compare logging with surveillance cameras because we believe people would feel stronger about logging conversations than that.
It seems that most participants have very little concern about camera surveillance in public or semi-public spaces. We were surprised by the lack of desire for privacy. The author personally would switch off almost every surveillance camera he came across if he could.
In the comments participants hinted at several other social connotations for our situations:
While out study of social connotations is not completed we began working on a simple classification scheme for places. In our initial position paper we tried to distinguish between places according to their scale. We still think that a distinction according to scale would be important, but we leave this distinction out of our present classification scheme. We think that most social connotations automatically determine a feeling of scale as well.
Our current list of social connotations is grouped into several categories. In each case social connotations lie somewhere between two extreme values. This is an incomplete list under construction. It is open to discussion and likely to change in the future.
| Space | |
| Large, open space | Small closed space |
| Light space | Dark space |
| Loud space | Silent space |
| Background music | No planned background noise |
| Safe (guarded?) place | Unsave place |
| Recording / surveillance devices | Complete Privacy |
| Food space | Non-food space |
| Resting space | Work-only space |
| Access | |
| Public access | Restricted access |
| It is obvious if I belong here | Once I am in, nobody can tell if I belong here |
| Community | |
| Unfocused | Focused on one person / direction / activity |
| Common goal | No common goals |
| Code of conduct | No code of conduct |
| People know each other | Strangers |
| People know me | I am a stranger |
| Forgiving community | Unforgiving Community |
| Small community | Large community |
| Stable community | Community constantly changing |
| Newcomer friendly | Newcomer unfriendly |
| Behavior | |
| Formal behavior | Informal behavior |
| Accepted Behavior is obvious | I don't know how to behave |
| Dress code | No dress code |
| People know this place | People don't know this place |
| Other | |
| There is an agenda | No agenda |
| Signs and tools help newcomers | No help for newcomers |
We describe the distinction between the concept of space and place, which is a space with meaning. Spatial metaphors are directly or indirectly used to describe systems that support people communicating and collaborating on the Net. Like all social activities, also social navigation is strongly influenced by the place it is happening it. We distinguish between indirect and direct social navigation. Especially in the case of direct social navigation the social connotations of a place have a strong influence on what behavior is deemed acceptable and what is not. It is therefore important that virtual places for social interactions are designed for a specific goal -- otherwise the social connotations might interfere with the activities taking place.
We describe a pilot study that tries to find in how far social connotations of real places are carried over to (textual) virtual places used for social navigation. In a first part of this study we conducted a Web survey that determined how people perceive social connotations in real spaces. We currently work on a second questionnaire which will study the same questions in a textual virtual environment.
We also present a draft classification of social connotations. The classification is meant to help designers of virtual communities to describe spaces in exactly the right way to allow the creation of a certain place and thus support or discourage certain types of social navigation behavior.
We would like to thank Andrew Frank who brought up the idea of looking at social connotations of space at a conversation COSIT'97. We further would like to thank Amy Bruckman and Tom Erickson. Their ideas have a profound influence on this work.
Bruckman, A. and M. Resnick (1995). ÒVirtual Professional Community: Results from the MediaMOO Project.Ó Convergence 1(1).
Curtis, P. and D. A. Nichols (1993). MUDs Grow Up: Social Virtual Reality in the Real World, Xerox Parc. accessible at ftp://parcftp.xerox.com/pub/MOO/papers/MUDsGrowUp.ps.
Dieberger, A. (1996). Browsing the WWW by interacting with a textual virtual environment - A framework for experimenting with navigational metaphors. Proc. Hypertext'96. Washington DC: 170-179. http://www.cc.gatech.edu/gvu/reports/techreports97.html.
Dieberger, A. (1997). ÒSupporting Social Navigation on the World Wide Web.Ó International Journal of Human-Computer Studies 46: 805-825. accessible at http://www.cc.gatech.edu/gvu/reports/techreports97.html.
Dourish, P. and M. Chalmers (1994). Running out of Space: Models of Information Navigation (short paper). HCI'94 (British Computer Society). accessible at ftp://parcftp.xerox.com/ pub/europarc/jpd/hci94-navigation.ps.
Erickson, T. (1993). From Interface to Interplace: The Spatial Environment as a Medium for Interaction. COSIT'93. Elba: 391-405.
Harrison, S. and P. Dourish (1996). Re-Place-ing Space: The Roles of Place and Space in Collaborative Systems. Proc. of CSCW'96. Boston, MA, ACM Press: 67-76.
Hill, W. C. and J. D. Hollan (1992). Edit Wear and Read Wear. CHI'92. Monterey, ACM Press: 3-9.
Lakoff, G. and M. Johnson (1980). Metaphors we live by, Univ. of Chicago Press.
Resnick, P. and H. R. Varian (1997). ÒRecommender Systems.Ó Communications of the ACM 40(3): 56-58.
Terveen, L., W. Hill, et al. (1997). ÒPHOAKS: A System for Sharing Recommendations.Ó Communications of the ACM 40(3): 59-62.