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

5. Navigation in computer environments

While the last chapter looked at user interfaces in general this chapter focuses on the navigational task in user interfaces. The concept of hypertext is described and it is shown how hypertext navigation can profit from a more spatial conception of the "hyperspace". In hypertext based information systems navigation of information is the primary task of the user. The user interface has been identified as an environment that commonly makes use of a spatial concept to support navigation. This chapter builds on this assumption and describes navigation in computer systems from a spatial viewpoint. A spatial user interface metaphor which supports navigation in collection of hypertextual data based on the metaphor of a city is described.

5.1. Hypertext and Hypermedia

The section first describes basic principles of hypertextual systems. The navigation task in hypertext is specified and commonly used tools to help in navigation are described. The advantages of spatial hypertext are described.

5.1.1. Principles of hypertext and hypermedia

Another basic principle of hypertext is "non-linearity". Where most types of information have a typical order in which they should be apprehended this is not true for hypertexts. Hypertext reading can be seen as "associative reading", as the reader determines the order in which information is consumed. Where - for instance - books have their own way of telling a story and steer the user deterministically and linearly through a "space" described by the author, interactive books and hypertexts allow the user to navigate through the information space freely.

Hypertext typically is a computer-based medium. In a computer-based hypertext it is possible for instance to select a word "annotation" and a small information-window opens to describe the contents of the "annotation". The annotation again can contain a number of links and by following them the user of the hypertext can navigate the network structure at will. Hypertexts are often compared to dictionaries and dictionaries indeed are a form of paper-based hypertext.

This section described the concept of hypertext. Hypertext and hypermedia are often used synonymous. Hypertexts consist of nodes and links which are navigated by the user. The chief principle in hypertexts is non-linearity in information access.

5.1.2. The navigation problem in hypertexts

The main difference between file systems for computers and hypertextual information is that file systems are hierarchical constructs whereas hypertexts are networks and therefore mostly non-hierarchical. That means that there often are several paths to a certain node. Except for that the navigational task is similar (see section 4.1.)

Hypertexts can be degenerated to linear lists or strict hierarchies. Although these structures sometimes are not recognized as hypertext many hypertexts contain linear or hierarchical substructures. Figure 15 shows an overview map of a hypertext which shows a network structure. Still clusters and sequences of information can easily be identified in this network.

Fig. 15 - A Storyspace map showing general linking and substructures.

This section described the navigational task in hypertext systems which is commonly associated with the problem of "getting lost in hyperspace". Recent research brought forth that disorientation problems in hypertexts are not a problem of the hypertext concept but of the user interface. Disorientation occurs when the structure of the information space is not well communicated to the user.

5.1.3. Navigational aids in hypertext systems

• Backup-Links. Most hypertext systems allow to move back node by node to where the user came from. By backing up the user eventually will encounter a node she recognizes. Backing up finally leads back to the starting node. Some systems use intelligent backup links that do not jump back one node but to the last "concept". For instance when reading a book such a link would not jump to the previous page, but first to the beginning to the chapter and then to the previous chapter read.
• History lists are list representations of nodes visited. They allow users to jump back to a certain node instead of backing up.
• Lists of recently visited nodes present an unsorted list of the last nodes visited or even small pictures showing these nodes, like in the HyperCard system.
• "Bread crumbs" are markers that show users where they have been already. They help mainly for getting an overview of how much of the hypertext has been seen already. It was often seen as a indication of disorientation when nodes are visited several times and bread crumbs should help users to avoid revisiting nodes. Therefore bread crumbs are sparingly used in modern hypertext systems. However revisiting nodes is an important aspect of navigation in an environment - it fosters the familiarity with certain nodes that can then act as "orientation nodes" or even landmarks.
• Interaction histories or "survey objects" try to capture parts of the system that have been seen already or that have been found in a search operation. Users can manipulate these objects and generate "subspaces" of the hypertext. An example of the use of such objects is to generate a subspace of "all the nodes I did not look at yesterday and that contain the keyword 'space'" [PuGr93].
• Typed links restrict the number of possible links when looking for certain information. An example for link types are links leading always to pictures, or links always leading to references, to counter-arguments or to more detailed information.
• Maybe the most famous way to aid in navigation is to make structure explicit using overview maps. Such a map represents the structure of the hyperspace as a directed graph - an example being figure 15. Overview maps suffer from the fact that screen space is limited and that they often can show only parts of the hypertext. Solutions to this problems are the use of nested maps or fisheye-views which expand the map around a certain focus. Overview maps can be hand-crafted by the author or be generated automatically. They are most helpful when based on a space concept. An example are automatically clustered overview maps based on "textual similarity" of the node contents (see for instance [MuFH94] or [ZiBl94]).

5.1.4. A hypertext using several navigational tools

This hypertext has a mainly hierarchical structure and contains two overview maps, that show the structure in two levels of detail. The overview maps are hand-crafted. It is possible to navigate to another node simply by clicking a box in the maps. The check-marks in a several of the boxes show the use of bread-crumbs in the system. The bent arrow in the lower right corner of the system is the backup-link. In the text-page shown there are no links however many other pages of this hypertext contain links to other nodes. The structure of this system is mainly a hierarchy, that contains links across its structure.

Fig. 16 - Communicating the structure of a small hypertext as maps.

Other navigational tools present in this system are the history list, links to the starting node and to the next and previous node in a series. The small bar below the text-window shows the position of the reader in a linear series of nodes, that makes up the section "interactive fiction". This section consists of 4 nodes and we are looking at the third node of the series. This bar is an active indicator again: by sliding the black box in the bar it is possible to navigate to the other nodes in the series.

This section described how navigational tools can be used in combination to support navigation in a hypertext. It should be pointed out that these tools are useful mainly for avoiding disorientation - they do not really help to find information.

5.1.5. Spatial hypertext

As in the case of user interfaces a possible solution to such problems is to make the space concept explicit in the hypertext and to spatialize hypertexts on the first hand. Hypertexts and user interfaces are deeply intertwined and spatial hypertext therefore make use of spatial user interfaces.

For the process of spatialization a promising meta-concept has been proposed at the European Conference on Hypertext 1992 by Jay Bolter. He suggested two approaches to combine virtual environments and hypertext:

• Hypertextualizing the space. This can be realized by giving graphic elements in virtual space a symbolic significance. This concept is related to "writing on the world" as it uses a sort of pictorial language to present information on objects in the world. Examples of such writing on the world have been identified already in section 3.3.1.1.: stone-inscriptions, gable stones or guild-signs.
• Spatializing the (hyper)text. This principle essentially advocates to make the space concept explicit and to use more spatial representations of hypertexts to make hyperspace a real "space".

Conclusion

This section described the concept of hypertext or hypermedia, two terms which often are used synonymously. The essence of hypertext systems is non-linearity. They consist of nodes containing information and links to connect those nodes. Navigation in hypertexts can cause disorientation problems if the space concept of the hypertext is not clearly communicated to the user. Hypertexts rely on a topological space concept but many tools meant to help in navigation are based on metric space concepts. Spatial hypertexts try to communicate their space concept clearly to the user. It is assumed that clearly communicating the structure helps in navigation.

5.2. Navigational problems in computer environments in general

Tunneling through space

One example of such a problem is that two hypertext nodes which may be distant in an overview map may be considered close at the same time when connected by a link. When the overall structure of the hyperspace is such that closeness in the overview map expresses similarity of content then such a link is perceived as a tunneling through space.

The problem becomes apparent mainly when the user interface metaphor stresses the concept of the metric, flat space. In this case tunneling is perceived as a disruption or the metaphor and as a magic feature. If the tunneling is a useful feature is determined mainly by how the working of the magic feature is communicated to the user by making use of "enactment" (see below).

Tunneling occurs not only in hypertext linking. The Macintosh file system provides so-called aliases. Aliases are pointers to files located elsewhere, maybe even on another machine. Aliases can be used as if they were the actual file. They provide the user with functionality which is very similar to hypertext linking. Extensive use of aliases may restructure the file system from a hierarchy to a network.

Directionality in hypertexts

Although most hypertext system do not stress their spatial concept, users learn a mental representation of the hypertext structure. Based on this representation people may associate distances and directions to links. Closeness in hypertext is often interpreted as the length of the access path. This concept is similar to the reachability matrix described in section 2.1.2. Two nodes that are two links away are then seen as being closer than two nodes that are five links apart.

However this distance implicitly assumes that links always lead straight away from a node and that they have the same length. However it is well possible to follow two links and reach the first node again.

One possible solution is to associate directionality to link typing. A link-type that provides more detailed information could lead into a conceptual "direction". In case following several such links and arriving at the node one left in the first place is a clear disruption of the spatial structure and therefore should be avoided or at least be well enacted.

Searching computer space

Another problem of computer environments is closely related to the tunneling problem: the search problem. A typical task in a computer environment is to look for information one knows is existing in the system - only the location is not known. The standard solution for this situation is to make use of "search"-functions or a "find" command. The find command is the silent acknowledge of navigational problems in the desktop metaphor. If the desktop and the so-called "finder" of the Macintosh user interface really would allow users to "find" their documents easy, the find command would not be necessary in its present form.

Search functions navigate the computer environment using an identifier strategy (see section 3.1.). The search therefore degenerates to exhaustive search of the information environment. As modern computers are fast machines we normally do not regard this strategy as a problem. Still the information available to our machines grows much faster than the performance of our systems and in distributed systems this strategy is not very reasonable. This strategy is used in distributed system like the World Wide Web anyway where "Internet Worms" or "Knowbots" freely navigate the net to collect information.

When searching for information in the Macintosh desktop, the desktop space is searched linearly without making use of implicit information in the spatial arrangement on the desktop. When an item finally is found the user is moved to the location the item is contained in. This happens by opening a window showing the contents of the right folder. Little is done to communicate the location of this folder to the user. In a small, well-known computer environment the user may know from the window title where this location is. When searching in network-mounted folders or on very large hard disks such knowledge can not be relied on. Macintosh users frequently can be observed to use a function showing the path to an opened folder after a "find" - this function is similar to a history list up to the root folder. This indicates that users want to know "where they are" after a search.

Enacting the tunneling process

It was mentioned, that magic features are easier to understand when they are enacted well. The Macintosh user interface provides enactment for the tunneling problem however this hint is often ignored as it gives only a rough indication of the location: When a folder or document opens, a gray frame zooms open from the opened object till that frame reaches the size of the window to be opened. In the case of the find function this zooming often occurs from the symbol of the hard disk the opened folder is located in. The enactment of the find then shows only superficial information and is not as helpful as it should be.

As such an enactment has proven to be valuable it becomes a common practice to use more animation and enactment of user interface functionality. This gives users better feedback on what is happening and makes user interfaces more enjoyable to use. Well designed enactment can enforce the perception of the user interface as a space instead of only a flat picture on a screen [ChUn93], [BERN93a], [LAUR93].

Keeping an overview in large information spaces

In most computer environments it is impossible to get a quick overview of all folders available except on the highest hierarchical level. A lot of navigation is required only to locate folders that contain for instance project names - only then the quest for the document looked after really starts. There can be hundreds of documents, that all show the same icon and have only a short filename.

The problem of this type of navigation is that the representation of the information space is mainly flat. In a more spatial environment it is easy to move older documents more to the back, that is away from the user. This is the approach advocated in the information landscape metaphor (see section 4.3.4.). This approach keeps all the documents visible - only they get smaller and visually less important as they move back.

A 3D space concept allows much more information to be shown at the same time. This is shown impressively in systems like the Information Visualizer which is able to represent the complete file system of a workstation in a window [RoCM93].

Enrichment of information spaces

Enriching and spatializing a virtual world is possible in various ways. Several examples are discussed in [STAP93], for example the use of a virtual sun and virtual shadows to show the passing of time or using different sizes for information objects according to their size. A spatialized virtual space allows moving objects closer of farther from the user. Objects, that are important for the present work should be closer than others. This spatialization leads to the extension of the desktop space to room, house or city spaces.

The use of colors is a possibility to encode additional information in user interfaces. It is not always clear which color should be used to indicate, for instance, the age of a document [SmBo93], [LBSS85], [SOLO90]. Instead objects should be represented much richer thereby providing the user with important navigational cues [HoSa93].

Conclusion

Although computer systems are not seen as hypertexts the navigational task is strikingly similar and similar problems occur in both domains. Most navigational problems occur when the system uses a different spatial concept than is communicated to the user. As example the find command in the desktop space was compared to a tunneling through space. Such magic features are no disadvantage per se - indeed they can make an environment more useful. However their working must be clearly described by a well-designed "enactment" -- otherwise the user cannot make use of these features.

5.3. The Information City metaphor

5.3.1. Overview of the Information City metaphor

The city metaphor was chosen as a basis because people seldom get really lost in a real city even when they are in this city for the first time. Cities provide people with lots of informational and navigational infrastructure. Besides in all cities there are always other people around to help in navigation. It seems reasonable to make use of the city navigation skills most people have to navigate a complex computer generated information city.

The Information City shall not be designed by simply putting a few buildings into an empty space. It shall be tried to design a city environment by making use of knowledge in the fields of architecture and city planning. The Information City shall not only be a tool but an enjoyable place. The city will be clearly structured and the design of the city will strive to provide all those elements of the city image people use to navigate the environment. This description of the Information city follows the same structure as the ontology of the real city described in section 3.4.

[I]f the environment is visibly organized and sharply identified, then the citizen can inform it with his own meanings and connections. Then it will become a true place, remarkable and unmistakable [LYNC60, p.92].

Structure of the Information City

In the Information City buildings act as containers for documents. These documents can be for instance entire hypertexts. While navigation in the city structure is navigation in the collection of documents, entering a building allows navigation in the (spatialized) hypertext.

Buildings not only are containers for hypertexts but they show a lot of information about their contents. By making use of the writing on the world concept and additional concepts like read wear the facade of the buildings shows what type of information is contained in the building. The state of doors and windows gives information about the accessibility and the state of the facade gives hints about the age of the information contained in the building. The Information City therefore will be a rich spatial environment.

Navigation in the city is done either by walking on the streets, which is a metaphor for browsing between documents, or by using navigational infrastructure like taxis or subways. Navigation in the documents is done by walking the structure of the buildings.

Information in the buildings essentially is contained in information walls located in rooms or hallways. By navigating through the building structure the user walks along information walls and by doing so she navigates the actual information structure. Information walls and all other objects in the Information City exhibit read wear and show signs of use. This makes it easier for people to assess the age and relevance of information contained in the buildings.

The city is structured according to an abstract distance concept which is based on document similarity and user needs. Buildings thus are grouped into districts of interest and close buildings therefore should contain related information. Navigation in the city by walking will therefore allow accessing related documents easily. Navigation using navigational infrastructure will connect larger scale structures in the city - for instance districts each of which contains documents of a related topic. Navigation in the hypertexts or the documents is done by using the navigational structure of the building - that is hallways, stairs, elevators, rooms and so forth.

The space concept of the Information City

The Information City will be mainly based on an Euclidean space but is complemented by a topological space in the subway system. Additional magic features provide hypertext linking in the buildings and across buildings.

Buildings essentially are mainly hierarchical constructs which will be a problem in case a building contains an entirely web-like hypertext. This case can be handled in two ways. The first possibility is to use an embedded metaphor in parts of the building - for instance a special room using a view screen metaphor.

The other possibility is to make use of the building block concept [DIEB93b]. This concept has the disadvantage that it can be applied only during the authoring process of the hypertext. The building block concept essentially assigns the role of a skeleton to parts of the overall linking structure. The remaining link structure is then "wrapped around that skeleton". The skeleton structure can be visualized as the structure of a building - that is as hallways, rooms, doors and so forth whereas the remaining link structure is visualized using magic features - like magic mirrors or windows that look into other parts of the building. Magic features like these will enable users of the Information City to rapidly locate and access information.

Navigation and magic features

For these magic features link enactment is of special importance (see section 4.2.4.3). The main advantage of the city metaphor is that people cope well with navigation in cities. Using magic features for navigate basically disturbs the user's orientation. Enacting the magic features well can help to avoid disorientation problems. The enactment must give information about the direction and the distance traveled.

Of special importance in this respect is the subway metaphor in the city. It provides fast access to a landmark in a certain district and users of the subway do not expect to see the environment during the ride. As the user emerges from the subway system at a landmark reorientation is no problem for her.

At first sight realization of linking using the metaphor of flying would seem a good solution in a graphical virtual environment. On second thought flying is too distracting and visually intense to be useful for general hypertext linking. The user probably has forgotten why she traversed a link when she finally reaches the link destination. In the city a subway metaphor is therefore preferred. Subways are visually less distracting and always deliver the user at known landmarks from where it is possible to navigate further to the destination of a link.

Navigation starting at a landmark can be supported using light cues, signs or guides that lead the user to the destination. As in a real-life city navigational infrastructure shall be available at landmarks. Another possibility for linking would be to zoom out to a map view of the city and then to zoom in again to the destination. It is used in the Pad++ system [BeSt94]. For quick scanning of the neighborhood of information houses a surface tramway service can be used [TOGN93].

These examples of magic features needed in the city show that the city metaphor should not be carried so far that it limits its purpose. A single metaphor cannot cope with everything and it is necessary to make use of magical properties of the virtual world where it suits best. Such deviations should, however, be very well-chosen [WATE92].

Embedded metaphors

The Information City can be used as a front-end for various information services, like for instance the network based services Gopher, World-Wide Web and so forth. It will contain electronic mail, bulletin boards, meeting rooms where people from physically separate locations can interact virtually.

Specialized information rooms represent links to other cities, that represent data-bases at remote Internet sites. Users can automatically download data by requesting a copy of information they found in the city. There will be a place where people can request to have their own information linked to the city. The city provides information in information walls and other objects, which act similar but can be carried around and given to other users. Examples are newspapers or business cards. These objects can contain hypertext links but are not fixed to a location and are not visualized as houses [TROM93b] (see also [CuNi93] and [WATE92] for similar visions).

Not all these services can be realized well using the basic information building and information wall metaphors. Instead those services use embedded metaphors like the view screen, the post office and so forth. They will help to expand the Information City to be a useful place.

5.3.2. Metaphorical mapping from a real to a virtual city

Structural objects and functionality

Unlike in the real city flying is a commonly used type of navigation in the Information City, although most navigation is done by staying on the ground. Flying allows taking a look at the city structure in the top-down fashion described in section 3.4. Like in the real city major navigational paths come into view first. These paths should separate large districts in the city. As districts signify areas of related information the separation into districts could be highlighted using colors. At this level of detail major navigational paths, crossings of paths, squares and districts can be recognized.

Between major areas in the city there are undefined areas These areas contain the "void". City areas are drifting in the void like swimming islands in the ocean. The void cannot be closed and therefore allows unlimited expansion of an area. Expansion of an area over a certain size may result in separation of the area into smaller areas or in creation of an envelope (see below). The areas in the city are connected using long-distance transports like subways, flying or magic features.

Looking at single districts lower order paths, edges and nodes come into view. At this level of detail buildings and small-scale areas can be recognized. The next level of detail shows single buildings. Again this level shows landmarks and allows viewing architectonic properties of single buildings. Specialized buildings like subway stops and transport functionality can be seen.

The elements of the environment at this level are:

• Paths connect two locations in the city. Therefore they have a starting point and an ending point - both of which should coincide with landmarks. Along the path there is an arbitrary number of distinct locations like crossings with other paths, squares or buildings. Like in the real city a path is a continuous element of the Information City.
• Lines are linear elements in the city that follow a different space concept. Where paths are structure of Euclidean space, lines represent topological space concepts. Public transport, for instance the subway, travels on lines. Lines are connected to paths and squares at distinct locations which often are located at important locations like landmarks and squares. Temporary stops to enter the transport system can be created everywhere.
• Intersections of paths are squares. Large squares are major elements in the city. They contain stops of public transport systems and they can contain landmarks.
• A building is a container for information or infrastructure in the Information City. Buildings have a unique address along a path or on a square. Buildings show their accessibility using doors. Additional information is contained in signs, and in the look of the facade. A more elaborate access scheme can be represented using a fence metaphor.
• The facade of a building is an information carrier for the building it is placed on. It can contains signs and provides information about the age and state of the building and its contents through various parameters like open or closed doors, the color or state of the facade and so forth.
• Landmarks are a special type of building. Landmarks either are no access buildings or public access buildings in the Information City. The first type occurs when the landmark has only landmark function. Such a landmark can be used to provide a clear vista at an end of a path. The second, more common, type of landmark contains public accessible infrastructure. This type of landmark is used wherever public functionality is coexisting with a landmark. In a graphical Information City landmarks can be seen from far and paired landmarks can provide directional information. Major landmarks should be higher than most buildings to provide orienteering aids for the those users using flying in the city.
• Some buildings provide functionality to everybody - examples of these buildings are railway stations, churches and so forth. These buildings shall be realized as public access landmarks in the Information City.

• Hallways and stairs are the major navigational structure inside buildings. Hallways can make use of a red carpet metaphor to point out information of special importance or to show a preferred sequence in which to access parts of the building. Besides this metaphor hallways and stairs make use of light and color cues to help in orientation and to point out related items. These cues are a visual way of communication between the city as an agent and the user. Hallways and stairs have walls which may simply separations or information walls (see below).
• Similar to lines in the city there are elevators. Elevators in houses provide these structures with a topological space concept. Elevators - like lines in the city - have predefined stops at lobbies or important rooms. Temporal elevator stops can be created to enter the elevator at every location in the building. Elevators should provide direct and fast access to city-wide transportation - for instance subway lines or the flying metaphor.
• Lobbies have a similar function as squares in the city. They commonly connect hallways, stairs, and elevators. They sometimes are landmarks. In lobbies signs and information kiosks provide informational infrastructure.
• Rooms are locations inside buildings. Typically they consist of walls or information walls (see below). The walls may contain doors or windows to access other rooms or the outside of the house. Rooms show their accessibility by the opening or closing of the door. Rooms are seen as containers which can look different in case a different spatial concept of access to another information system is provided. These rooms are special and are called "envelopes" (see below). For example it is possible that rooms contain only one information wall which acts as a view screen metaphor. Rooms not necessarily have to be really contained in the building. Instead the door leading into the room can function as a magic feature to teleport the user to a special mode in the city - see envelope.
• Envelopes are special rooms. They do not fit into the Euclidean space concept of the city and the building and can be regarded as being located in another plane of reality. Like rooms they are accessed through doors. Envelopes can contain very different structures or even another Information City. A city in an envelope could represent archived data for instance. A transition to such a city has to be clearly enacted as a magic feature. The link to it is performed as an "envelope link". A "quit envelope" command transfers the user back into the house in front of the door leading to the envelope - like an intelligent "back" command in hypertexts.
• Doors in the facade connect buildings to the city and doors in other walls connect elements of the building - like hallways, lobbies or other rooms. Doors can be opened or closed. Closed doors may be locked or unlocked. The state of the door shows information about the accessibility of the room reachable through the door. Some doors can be opened only from one side. Open doors can be peeped into. This provides previewing information about the contents of the room. This peeping has to show clearly if the door leads to a room or to an envelope.
• Windows are similar to doors. Most windows are located in facades. They can be used to reach the path in front of the house quickly and can serve a departure points for flying. A special type of window is the magic window (see below).
• Magic windows are a special kind of information wall that provides either only a topological connection to other parts of the building or another location in the city or a view-only connection to another location. View only connections can be used to refer to information in other locations as "annotation". For instance such a window allows to see resolve the references in a scientific paper but it does not allow to link further from these papers. Magic windows have to show clearly that they are magic features. Magic windows should be used only sparingly.
• Normal walls separate rooms, hallways and other elements of the house. Walls contain doors, windows and magic windows.
• Information walls are the main information providers in the city. They present information and show read wear. They can provide linking to other information walls either by moving the user to the next wall in sequence. As an example consider several walls constituting a sequence. Activating the "next" link moves the user into the adjacent information room or simply to the next wall in the room. Information walls can provide linking using the metaphors of the elevator or the subway to link to other locations. The links on the information wall provide read wear. Information walls can contain magic windows in which users can provide visual links to annotations about the contents of the information wall. Information walls cannot be removed.
• It is possible to provide information dispensers in a room. These take the form of a pile of "removable information object" like business cards, papers and so forth (see below).

• Facades provide information about the contents of a building as was described above. Facades should show read wear to give information about the read wear status of the building. This either is the read wear of the main lobby or an average read wear of structurally important rooms.
• Signs are a type of information walls which provide no linking. Signs cannot be removed.
• Traffic sign are signs providing mainly navigational information.
• Removable information objects are similar to information walls but can be taken away. They provide information which is not in direct location context to a room. These objects may contain magic window links to the room they were contained in or to additional information. Examples of such objects are business cards. They provide information about the owner of the card, provide links to the room they were taken from and links to additional information. This information can be viewed in a magic window or can be "ordered". Ordering downloads the information to the site of the user holding the card. Copyright and copy-fee issues is not specified in the metaphor but could be supported easily using for instance a mail-order metaphor. Other such objects are address lists which remember locations in the city and provide preview and fast access to these locations and maps.
• Information can be bundled in a kiosk. Kiosks are public access rooms that often containing information walls with general information about the surroundings of the kiosk. Kiosks contain piles with removable information objects giving navigational aids, like maps, address lists with important locations nearby, preprogrammed guided tours, and so forth.

• Walking is a slow form of transportation mainly for short distances. Walking can use paths, squares and all open access house structures. Walking is either interactive, when the user controls direction and speed directly or automatic when an address on an address list is selected for instance.
• Cars are a metaphor for fast walking. In the Information City can be summoned at every place and can be disposed of at any place.
• Taxis are almost identical to cars. They can be summoned instantly everywhere and are able to navigate using incomplete information. They cannot be steered interactively but they can provide a guided tour when given a preprogrammed guided tour object.
• Buses and tramways give predefined slow tours in small areas. They can be used to scan buildings along a path. They can be summoned on lines, have predefined stops but can be left everywhere during the ride.
• Subways provide long-distance transport in the city. They provide a set of predefined stops which always coincide with landmarks. Leaving the subway at those landmarks either places the user in the main lobby of the landmark or in the front of the landmark. The travel time gives a rough indication on the distance traveled. Subways shall show the direction of travel but they do not show the environment traveled. Subways can be left only at predefined stops but it is possible to summon a temporary subway stop everywhere outside buildings. From inside buildings they can be reached using an elevator. The use of the elevator is forced upon the user to clearly show that the user leaves the building she was navigating. Subways use a topological space concept in the city. They are able to cross the "void".
• Flying is long distance navigation which visual connection to the environment traveled. During flying users can orient themselves using major landmarks which are designed as higher buildings. Flying can be quit by "landing" at any place in a path or a square or on top of a building which provides access to an elevator. Flying can cross the "void".

Communication between users is provided in the information city using post offices and telephones. Communication with the user's agent requires no communication device. Ideally the city should provide helper agents that can be used to watch information walls for changes. These helper agents can communicate with the user via the user's agent. Communication is not treated in detail in this description.

Limits of this environment

The Information City is a very large structure which partly will be changing dynamically. It is to be expected that radical changes in the structure cannot be supported well. Multi-user access in the city is possible. However in this case it is not possible to restructure the city to optimally fit each user's needs.

The Information City is not ideally suited to represent entirely web-like structures. These structures depend mainly on magic features. Magic features are very useful in the city but should not be the main way of navigation.

In order to show information about buildings on the facade a sort of pictorial language has to be developed which can be used consistently in the city. For instance it has to be clearly defined what a certain change in color means.

This section outlines a spatial user interface metaphor based on the city model. The Information City relies mainly on information richness and psychological aspects of space to structure and navigate the space in the city. The metaphor is aimed at organizing and navigating large collections of hypertextual information.

Summary

This section described a novel spatial user interface metaphor called the Information City. This metaphor is based on the Euclidean space concept of a real city and enhances this space with magic features to support easy navigation and hypertext linking. The city metaphor is designed to make use of knowledge in the domains of architecture and city-planning to create an area that is easy to navigate. This includes the strive for designing a rich environment which provides informational and navigational infrastructure. Based on the ontology of the real city, which was described in section 3.4., this section gives a detailed ontology of the virtual city environment.

5.4. Open questions in the Information City

• Navigation and orientation in the city metaphor: Do people find their way around in textual environments easily? What kinds of landmarks are useful for navigation? What are basic properties of landmarks? Navigation in real life cities is not always easy. How to avoid these problems in a virtual city? Which magic features are to be used and what are their consequences for the spatial perception of the space?
• Modes of transportation: In the description of the Information City metaphor the topic of transportation is outlined only briefly. The concept of building blocks could provide "landmarks" in every building block. Those landmarks are the corner stones of every non-walking navigation in the city. The question is which metaphor is most useful under which circumstances. A transport can be a guided or an unguided way of navigation.
• Perception of room size, position, and geometry: Closely related to issue of transportation in the city is the perception of size and distance. Distance in the subway could be conveyed by sound or travel time. When walking from a place to another in a textual environment this transition is enacted as "you walk <direction>" or a similar message. This enactment does not give any information on the distance traveled. As MUD rooms often are assumed to be placed on a regular grid missing rooms require some paths to be longer than others. Are people aware of that? Do they assess a sort of path length? Do the descriptions of the rooms modify the perceived size of the rooms? This again would modify the placement of rooms on the grid and therefore modify the distance between rooms.
• Navigating very large information spaces: Are large distances navigated in a very different way than shorter distances? Does this imply the need for different transport metaphors? What are the limits of useful navigation in a spatial environment like the information city? In real life when going for instance from Vienna to Boston people apply hierarchical navigation. They do not know how to get to Boston directly - instead they have the knowledge that Boston is to the west, that it is reached best by plane and how to navigate to the airport. They start worrying about local navigation when they arrive at Boston, starting from the next available navigational infrastructure, for example an information kiosk. Similar navigational concepts are needed for navigating very large scale information spaces.

This section outlines several problems in the concept of the Information City metaphor which have to be researched into before being able to implement a system based on the Information City metaphor in a textual virtual environment. These questions are the basis for the study described in the next two chapters.

5.5. Conclusion

The Information City metaphor is a spatial user interface metaphor based on the metaphor of a city. The city metaphor is designed to be a rich and well structured environment and shall make use of principles from architecture and city-planning. It is based on an Euclidean space concept and hypertext linking therefore relies on the well designed use of magic features.