to the top publications

PhD Thesis © Andreas Dieberger 1994, 2000
Navigation in Textual Virtual Environments using a City Metaphor

[Contents]--[Abstract]--[1]--[2]--[3]--[4]--[5]--[6]--[7]--[8]--[References]


8. Conclusions and further work

Virtual worlds need to be designed. The world and the experience that one can have in it must be consciously shaped. So far, those using virtual worlds have never been caught up and taken away by the experience. The event usually involves a bad copy of the real world, verging on the tedious, tiresome and downright boring. We need to create worlds that are worth entering - that are inviting - and worth spending time in - that are CAPTIVATING. They should be powerful, interesting, pleasurable, a medium of expression and experience as well as a way for people to interact with each other. Its about time we stopped tinkering with the equipment and started playing with the worlds.
[BEST94, p.2]

This chapter tries to summarize the main conclusion from the navigational study in the textual virtual environment. The focus is on results that are of importance for the realization of an information browser in a textual virtual environment based on the Information City metaphor. General results are contained in the first section of this chapter. Results of relevance for the Information City are focused on in the second section. The last section points out future work.

8.1. Navigation in MUDs

Spatial metaphors are a promising extension to the set of user interface metaphors. The spatial metaphors used today do not really exploit the spatial concepts of everyday life. Those concepts are not restricted to the spatial organizing principles. Instead this concept engulfs aspects of rich spaces, of social spaces, communication, culture and the interaction between people. These aspects together can turn a space into a place - a space that has meaning to its inhabitants and users.

The space concept

MUDs essentially are topological spaces but show Euclidean properties locally. They can be used to design a spatial environment to implement a spatial hypertext system in a MUD.

Spatial discrepancies in the environment are noticed by most users. They are not perceived as disturbing as long as they are more or less local and as long as the overall benefit of the spatial disturbance is greater than the problems caused thereof. Larger spatial discrepancies are not perceived as discrepancies but as a transition to an entirely different "mode" or level.

It is important that such transitions are well enacted to tell the user immediately that she is entering a different mode or level in the system. In user interface design "modes" are seen as a disadvantage. In the spatial user interface they are sometimes justified since a mode can be seen as entering a place dedicated to a certain activity. This spatial separation of activities is one of the benefits of the spatial organization.

Rooms and places in the textual environment are perceived by users as distinct places. Users seldom think of places as big or small unless they need this information for navigation. Sizes and distances are perceived mainly from the wording of the description of the space. Since the triangulation theorem in virtual spaces is not valid this size information does not help users in navigation - except in very well designed areas.

Many users do not perceive the textual environment as a three-dimensional but a mostly two-dimensional space - that is a plane or a set of planes. Up or down exits are commonly used - still the environment is seen as being flat. Using these exits as a spatial metaphor for going "deeper" or to "more" information therefore will work for only a restricted set of levels of detail.

The space concept of the city metaphor seems to be a useful metaphor to navigate large structured information spaces. The spaces should show a partly hierarchical structure however. Totally unstructured information spaces, like entirely unstructured hypertext webs, can be represented using the city metaphor but the resulting environments are unpleasant and hard to navigate. They function similar to magic mazes and will be avoided by the users.

8.1.1. Infrastructure

A virtual environment needs infrastructure for the activities of navigation, communication and providing information. The placement of such infrastructure can and should be done similar as in real life - according to knowledge about designing spaces that exists in the fields of architecture, city-planning and related fields.

Exploiting this knowledge in an interdisciplinary design act can help make a virtual environment not only a space but a "place". Such a place could even turn to a mirror-world of the real world where people go to do business, to get information and to communicate. Creating large environments is never the act of a few individuals but a collaborative dynamic process. Design knowledge can help guiding this process but it cannot replace it. It is the people living in a community, who shape the community and therefore the space the community uses.

Information providers and transports have been studied as examples of infrastructure. Depending on the richness of the description of space there are several ways of providing information in a textual environment. The more complicated but more realistic ways to provide information are preferred by players. The separation between structural information and actual information therefore should be strong.

Navigational infrastructure makes it possible to navigate rapidly through the environment. Two kinds of transports - the point-to-point transport and the transport system have been identified. The transport system is suited for general connections in an environment whereas the point-to-point transport lends itself to specialized, hard-wired connections in the environment.

8.1.2. Magic features and the need for enactment

Magic features are the main advantage of virtual environments. They help in navigation when designed well. However magic features can confuse the user more than they help when they are not designed well - such features will be avoided by users.

Well-designed magic features in the virtual environment enhance the efficiency of navigation enormously as they provide novel and very effective means of interacting with an environment. Users have to learn and get used to these new ways of interacting as people had to learn to cope with the invention of the railway in the last century.

Learning how to use magic features requires, that the functionality and the working of the feature is clearly communicated to the player. A well-designed enactment of the magic feature helps enormously in understanding and learning its use.

Navigation in textual virtual environments is a collaborative process. Users are used to be able to ask somebody else if they are lost or too busy to search an area for a specific object or place. Collaboration is done either by describing the path to the desired location or by guiding the lost user. The main advantage of using spatial metaphors is the support of communication of spatial properties of objects and places to other users. People are used to work with inaccurate path descriptions in real life. Very superficial route descriptions are sufficient in an environment like the Information City to point out the location of information. In a single-user system this activity of guiding has to be realized by a virtual user or an agent.

Using magic features exclusively for a long time can cause forgetting of knowledge about areas. Still the knowledge of the existence and general location of landmarks remains even if the user is not able to find those places without searching the area.

8.1.3. Guidelines for the use of spatial metaphors

At the ECHT'94 a poster was presented outlining results of this thesis that are valid for general spatial environments. On this poster five guidelines for the design of spatial user interface metaphors for spatial systems were given. These guidelines are: -- [DIEB94]

8.2. Consequences for the Information City metaphor

8.2.1. Is it realizable?

The Information City metaphor is realizable in a textual virtual environment. However a few of the features of the metaphor are hard to design well in a text-only system. The Information City is ideally realized in a semi-graphical MUD client or in a graphical virtual environment.

On the other hand the expressiveness of written language will make the enactment of magic features much easier in a textual virtual environment. Magic features are difficult to design well in a graphical system because there aesthetics in the interface plays a more important role than in a text-only environment. This is a situation similar to the design of special effects in film-making.

As in real cities navigation in virtual spaces can be facilitated by making single locations special. This can be done by pointing out landmarks, paths and edges, by enriching spaces, by enforcing activity and so forth. In purely textual environments not all of those methods are directly usable since the level of abstraction may to be too high and the separation of structural information and the information content itself is difficult. A fitting use of text-layout in rooms or pseudo-graphics can be used to partly overcome those limitations of the text-only system.

8.2.2. Providing other structures in the city

Maybe the most important design issue in the Information City is the clear distinction between the general parts of the city which follow a mainly Euclidean space concept and closed-off areas where a different space concept is used to provide for instance general hypertext linking. An example are fully connected web-like structures. These could be realized using a viewing-screen metaphor inside the virtual world metaphor so that navigation in a spatial environment is not necessary.

While the main advantage of virtual environments can be seen in the fact that the user moves through data instead of data being moved in front of the user. Those web-like structures do not lend themselves to the realization in a spatial metaphor and should therefore be represented using the conventional method. The disadvantage of that method is that the user will not be able to control his viewpoint as easily as in a real virtual environment. Realization of such a web structure using the spatial metaphor would yield a space that is similar to a magic maze and could be avoided by the users.

Areas of this kind should be kept small and closed off from the rest of the system. They can be seen as special services in the city like for instance a World-Wide-Web (WWW) service in a special room and should be designed as such. The Information City is a place to organize and navigate structured information. Entirely unstructured information cannot be provided in the city without referring to magical features.

For these closed off subspaces in the city the hierarchical structure of the city metaphor is ideally suited. An example is a building providing general hypertext linking in its inside. It is important though to keep these exceptional space concepts closed off inside buildings or even rooms, to make clear when users enter such an area and when they leave it. Such a transition has to be well enacted to be understandable. Only then it will be perceived as entering a different "mode" in the environment.

Similar consequences are valid for any hypertext system that tries to facilitate navigation using a spatial metaphor. Spatial metaphors are useful mainly for structured hypertext. Spatial user interface metaphors must not be considered a universal cure for interface problems.

The opposite approach - forcing a spatial structure on a web structure can be realized when parts of the linking structure of the web are assigned the semantics of navigational links and the remaining links are represented as magic features. If the selection of the navigational or structural links is done well that web structure can be used in a spatial environment - the general outcome will be that of a magic maze however. It is possible to perform such a structuring for a single user. This process would change the appearance of the spatial structure for every user however and collaborative navigation in such a system will be impossible. This will eliminate one of the main advantages of the spatial metaphor.

8.2.3. Hierarchical structure

As the city metaphor is well suited for a mainly hierarchical structure of a large information space by separating the space into districts which are further separated into sub-districts, streets, buildings, floors, rooms, and so forth it is advisable to structure navigational support in the environment similar to a several-level scale concept of space like that described by Montello (see section 2.2.4). As the city is be structured according to a distance measure on the information contained, for instance using document similarity [SALT89], [SaAB94], information belonging together will be located in clusters - that is the structuring of the city will automatically lead to the creation of hierarchical structures in the city.

Navigational infrastructure for rapid transportation according to this hierarchical structure should be provided. Four different classes of navigation seem advisable:

Links covering large distances should be realized using transport metaphors like subways or taxis. The main difference between the kinds of links is if the user is interested in the area between link source and link destination. In the first two types of links - where the user is likely to see only a few rooms in between, that are closely related to the topic - she even could be interested to see those rooms. In the other cases seeing all the rooms in between is more confusing than helpful. It will be necessary to design a transport that transports the user to a destination or a landmark near the destination. The remaining distance then can be covered by using a guide. If the destination is of prime interest however it should be made a landmark and the user should be able to navigate there directly. Those transports are easy to realize using the subway metaphor. Subway stops are typical landmarks in a real life city and the rider of a subway is not aware of the area she travels through.

Enactment of links shall provide the user with a feeling of distances and direction - if this feeling is not communicated to the user she will be unable to create a coherent picture of the environment as a spatial structure but instead sees it as a set of information islands, that are locally Euclidean but are drifting in the void.

8.2.4. Restructuring the city and extending the metaphor

There is a fifth kind of link, that leads to an entirely different information system like another information city. Including this concept would extend the Information City to the metaphor of the information islands or to an information world consisting of several information cities. As each city can be structured only once according to the distance measure a major change in interest of the user could require restructuring the city. In a larger metaphor like the information islands a viable approach to this problem could be to create several cities containing the same information in different structures.

The information islands are a very useful metaphor for less structured kinds of information. The information islands metaphor is used by the Singapore National Information Infrastructure project [WATE94]. Islands are a useful extension of the city metaphor and can be used as a metaphor to structure an information environment that encompasses several Information Cities. This is a promising solution when an Information City grows too fast. Since cities are more a plane than a cube distances between documents can become unbearingly long. Breaking the information space into several islands drifting "in the void" is therefore a logical extension of the information city metaphor. As the space between the islands can be seen as indefinite the islands can expand dynamically in the void without creating spatial inconsistencies. This process is similar to the stretching a rubber sheet.

8.3. Future work

This section points out two issues that could not researched further during this study. One of those issues is the possibility to "measure" the navigability of an area using a heuristic measure which takes several aspects that can influence navigation into consideration. Another issue is researching into collaboration in navigating virtual environments. Additional further research on navigation, on the spatialization of hypertexts and on the perception of time and change in information environments will be useful in textual and graphical environments.

Assessing navigability of areas

As the case study showed there are several aspects that can heighten the importance of a room in the cognitive representation. In a visual environment it is rather easy to decide for an object if it could be a landmark or a major orientation point. In MUDs the information in the room cannot be perceived in a glimpse. This fact makes assessing landmark status of rooms in textual virtual environments much more difficult.

The following list of aspects of room descriptions that can be assumed to influence the importance of a room:

These and similar aspects could be collected into a heuristic measure which could provide a way to assess the ease of navigation in a area. This heuristics would give a numerical value to weight presence and absence of such features in rooms. In MUDs designed as information providers it should be the aim of every author of an area to build a space that has rooms with high values of that metric spread evenly - but not too evenly - throughout the area. An area where all room have a high value according to the metric can be as hard to navigate as an area that has low values on every room. The fine-tuning of the metrics strongly depends on the system that shall be built and on the users who use it.

A useful application of such a metric can be in an iterative improvement of the navigability of an environment. Such a heuristics could be useful in every information structure providing navigation. An example is the World-Wide-Web. People using the net collect information and often include it in their "homepage", when they think it is useful. This again makes that information more useful since it gets more known and more available to other users. The WWW is a typical example of a community in a virtual space that collaboratively expands that space and adapts it to make it a more useful space.

An environment could be assessed as simple to navigate if there is an even distribution of nodes with high values of that metric - an assumption that certainly has to be verified. However designing a useful metrics still requires much future research on what makes rooms really special.

Collaboration in virtual spaces

Another important area for future work in information environments is support for collaboration which was excluded in this thesis. The spatial conception of the Information City makes it an ideal area to navigate collaboratively as people are used to communicate about space and to exchange directions how to find objects and locations in space.

For this work it is needed to further research how people describe environments and routes and how easily it is possible to navigate spatial environments using incomplete information. It is assumed that navigation using incomplete information is the easier the richer the environment is, the more coherent the spatial concept is designed and the more navigational and information infrastructure is provided.

When presenting these ideas to various people often the argument was brought forth that in a real information city people would be much too busy to give directions to other people. I do not believe that, simply because exactly this collaborative navigational behavior can be observed on the Internet in the World-Wide Web.

The Web as a general linking structure with no spatial concept makes it hard to find information on the web. Unless somebody announces an URL (a web address) nobody will find the information contained there. Interestingly it has become very common practice for people to provide "home-pages", that function like a business card on the web. Most people spend a lot of their time designing their home pages and making them publicly available. Those home pages often do not only provide information about the person whose home page it is but give access to literature databases, publications of that person and last but not least a large collection of links to other locations in the net which this person thinks are useful.

This is typical collaborative navigational behavior. People collect navigational information, bring them into a usable form and make them publicly available. However homepages can be only a small start. They require exact link information to be useful. In a spatialized information environment like the Information City similar information can be useful even it is incomplete. As the city is a multi-user environment people can not only provide fixed lists of pointers but communicate and interact, they can hand information objects to each other and get in social contact.

Future work in the field of collaborative navigation has to look at how to support this process. This collaborative navigation eventually might lead to the realization of the concept of cyberspace - a virtual social space and place people inhabit as a parallel world to the real world we daily live in.


[Contents]--[Abstract]--[1]--[2]--[3]--[4]--[5]--[6]--[7]--[8]--[References]

last modified on 10/31/96
Andreas Dieberger
andreas.dieberger@acm.org