Exploring the design space of networked technologies

Michael Twidale
Graduate School of Library and Information Science
University of Illinois at Urbana-Champaign

Keith Cheverst
Computing Department
Lancaster University


Given the explosion of new gadgets and networking technologies, how do we decide what to try and build?

In the past, computer science was a matter of coping with shortage (limited processing power, speed, memory; excessive cost, bulk and weight, constraints of location, time and mobility). Neat new software aimed to push at the limits of those shortages. Now the rate of improvement in hardware and infrastructure means that the options for innovative software development explode in all directions. As a result we are spoiled for choice in what we can choose to try and build. Worse, we have to cope with such a rapid rate of change that software development projects can be overtaken by the rate of change of hardware. Imagine a three year project involving the use of mobile phones in new computer-supported ways, and how the project would have to cope with the change in the capabilities and public use of phones in that time period.

Thus we move from the problems of design under scarcity to the problems of design under plenty (but very rapid change). So the problem, the research question, changes from;

"What can we possibly build under current constraints that will just about work well enough to prove the point?"


"Of the vast number of things that we could build now, which ones offer the most productive insights for the future?"

Just trying to build the first neat idea that pops into your head may well not be the most productive approach.

Our approach involves an attempt to explore the design space for networked personal technologies. We are using two related contexts for that exploration: tourism of an historic city and visiting a museum. However we claim that the need for a principled exploration of the design space is necessary for any given context or combination of work, home life or leisure.

We use various mechanisms to generate use scenarios, and to critique and modify those scenarios in order to consider design options and research questions that arise. The aim is to produce a more principled approach to the exploration of the design space and the selection of certain technologies to develop in order to understand the associated problems in more depth. The use of scenarios enables an exploration of issues in a way that is less technology-centric. In particular we have in our case study chosen to focus on issues of collaboration. This leads to an exploration not of the potential of a single technology (such as a PDA) but rather on how combinations of technologies might be fluidly combined in different ways to address particular needs.

The Cyberdocent project

The availability of advanced computing technologies is leading to radical changes in business, education, and society in general. The aim of this project is to explore the potential of new technologies to enhance the experience of visiting a museum. The focus is not remote access to the virtual museum; rather, we wish to examine a relatively under-explored area, namely how to enhance the experience of an actual museum visit in real life, through the deployment of different information technologies. Our approach is to draw on the experience and expertise currently used in museums to address this issue, and to match it up with conventional existing, cutting edge, and soon to be developed technologies that could be applied in innovative ways. The work considers a range of types of museum- not just science museums, but those oriented to art, culture, human history, the natural world, etc.

From Docent to Cyberdocent

One very common way in which museums attempt to enhance the experience of visitors is by providing docents. These are usually volunteers with particular interests and specialist knowledge who are trained to provide guided tours and to answer visitors questions. We have explored the nature of 'the docent function' in an attempt to clarify the different kinds of things that can be done (Rayward & Twidale 1999). In this work, we came to the conclusion that there were a wide range of things that museums did to 'add value' to their collection, all of which could be considered part of the docent function, even if they were not actually done by docents. For example, while a visitor is looking at an artifact (such as an amphora) a docent could explain more about where and how it was found, how it was used and its significance in helping our understanding of the people who made it and who used it. A written display could also provide a similar explanation, although it would probably not be as entertaining. Both the talk and the display serve the docent function, but use different media and technologies with their own particular attributes. Indeed it was our contention that it was precisely the docent functions that made a collection of artifacts more than just a collection and turned it into a museum.

Although we coined the term 'cyberdocent' as a shorthand for the exploration of technologies that might be used to provide certain docent functions, there already exist examples of ways of providing that information other than just by using people. Clearly a wall chart of text can convey information that a docent may explain to the user, but naturally the two media have different functionalities. The wallchart is always present, but the wording is fixed. There is only so much text that can (or indeed should) be presented and revising it is time consuming. Too much text is intimidating, but too little information is frustrating. If you are reading the text, you aren't looking at the artifact. By contrast, a docent can modulate her presentation based on the backgrounds, interests and even the real-time reactions of her audience. However you must be there at the right time to join the tour to hear the explanation, and for a group of mixed ages and experiences, the docent must choose an appropriate single level of detail to pitch at. Docents may not always be available at the time you want them, and you may not want to follow at the exact speed and the exact itinerary of the group.

A classic example of an early cyberdocent is the cassette audio tour. This is a kind of duplicated disembodied docent voice. Personal audio machines are available for hire and are often provided on the occasion of special exhibitions. In effect, they offer the individual viewer a personal but necessarily invariant pre-recorded commentary about selected items located throughout an exhibition. On the other hand, despite these constraints, the audio-guide provides a certain kind of flexibility not possible with a live docent and a tour group. They can be switched off and back on as the visitor takes a detour or backtracks and then resumes the tour, and they can repeat their commentary as often as needed.

New technologies offer new opportunities. CD-ROMs are now being used for audio tours. They have better sound quality than audio tapes. But more importantly they can hold a large store of information which can be accessed randomly, whereas audio tapes hold only a limited amount of information which has to be accessed sequentially. Thus, with CD-ROMs it is possible to design mechanisms that enable a user to move at his or her own pace and in any order through the collections, retrieving the commentary related to particular items in an exhibition as needed by inputting a code provided near the item. Moreover it is possible, given the capacity of CD-ROMs, to envision the provision of multiple audio extracts in order to offer commentary in various languages or at different levels of detail or focusing on different aspects of the artifacts. One extract might be geared to children, another might offer general overview information and another provide more detailed information. However, the amount of audio data that can be held on CD-ROMs is itself limited, and the increased multimedia capacity of DVD technology provides opportunities for greater flexibility in the amount and kinds of information that can be presented. With wireless technologies there is in theory no need to store the information on the device held by the user (although issues of bandwidth and reliability will impose some constraints). A simple example would be a variant on the audio tour where a mobile phone served as the interaction device. Taking advantage of the keyboard and limited LCD display creates opportunities for selection of tour options, pausing, replaying, asking text questions or even talking to an expert at a help desk. We are not saying that mobile phones should be the device to be used, only that it is a possibility in the design space and one where the consequences of recent mass production and mass use have now radically tipped the cost surface of that design space.

One of the ways that docents can help visitors to increase their enjoyment of a visit is by providing alternative insights into viewing and interacting with artifacts. This might be accomplished by sharing skills of how to look at a work of art, or reassuring for the visitor that their reactions and impressions are indeed appropriate and a good way to study a painting. It might include providing background information, sharing the results of scholarship in an accessible form, or showing practical or contemporary linkages with an artifact that may seem remote because of distance, time or complexity.

The tour group also emphasizes the social aspect of museum visiting. Many visitors arrive in groups (couples, families, friends, etc.) and discuss together what they see and experience. This social interaction is part of the pleasure of the museum-going experience, as well as having significant educational importance. A docent can enhance this, both by creating a larger social environment in the tour (adding to the social aspect for solitary visitors), and also by making comments, providing information, or asking open-ended questions that serve as topics for post tour conversations and discussion after the tour members resume their visit as separate groups. It is precisely an issue such as this that a purely technological approach may overlook. The social aspects of current use may be regarded as merely a side effect of the constraints of the physical world. That is, one might regard the requirement of going round the museum with other people in a guided tour or milling round with other independent viewers at best as a necessarily evil. With such a view there is the temptation to abolish all the other people in a pristine personalized digital museum accessible from one's desktop. There are many occasions where such a functionality would be desirable and useful, but it would be a mistake to consider that it is in almost all ways better to get rid of those other people in the physical world with whom one is 'forced' to share scarce resources. This applies to museums just as much as to libraries (Twidale 1995)

The development of multiple layers of meaning for a given artifact

A museum artifact can be understood and explained in many different ways. Ideally, you would have a personalized tour by a friendly expert sharing their enthusiasm and helping you to engage with the meaning of the artifact based on what you know and what you are interested in. There are many levels of detail about the item to suit the needs of visitors with different levels of interest, amount of time to spare, previous education in the area, experiential background, etc. We can think of the different kinds of explanations as multiple layers of meaning for the artifact. A printed exhibition catalog can thus be regarded as attempting to address this layering by offering additional information for people who want to know more about a particular artifact. However, the constraints of cost and space limit the amount and variety of layers that can be offered this way. New technologies can productively regarded as tipping the cost surface of the design space. In many cases they provide features of services that have always existed for the very wealthy, but that are suddenly now affordable for a much wider market. In order to help designers to explore these new opportunities it can be productive to look at the behaviours of the wealthy and ask: "Which features of the services available to the wealthy can now be made available at much lower cost to many more people?". Thus in a museum case, a wealthy patron (and likely donor) will be given a personilised tour by an expert. That may not be possible for everyone, but a less personalised audio tour by that same expert (or a well paid actor with an excellent speaking voice reading the script) does provide some but not all of the features at a drastically reduced cost. With computers incorporated we can add greater interactivity and personalization to the visitor's experience.

As well as layers of increasing sophistication of analysis, there are also layers based on different themes. For example, an Egyptian mummy could generate many different types of discussion and analysis: Egyptian history, religious practices, the physics, chemistry, and biology of mummification, the techniques of analysis, radiology, carbon dating, pollen analysis, pathology, how archaeologists work, current academic debates in Egyptology, whether mummies should be put on display at all (repatriation and reburial), comparisons with other cultures (including our own), mummification, embalming and other funeral rites.

Just as a single artifact can be used to tell many different stories, a combination of artifacts can create many different tours of real or virtual exhibitions. For example, a steam locomotive next to a canal boat helps explain the evolution of transport infrastructures, while the same locomotive next to an early steam powered water pump from a tin mine helps explain the evolution of steam technology.

One aim of museums is to share with a visitor different ways of 'seeing' the artifact. Different tours for example can feature the same artifacts for different purposes. This may be part of the aim of enhancing a deeper understanding of the natural world, helping people appreciate an important but difficult scientific concept or helping the visitor to appreciate different insights from viewing a work of art. All these fit within a broader educational aim and new technologies afford novel ways of supporting this process.

Technologies under consideration

The main technological focus of this project is the use of wireless PDAs and their potential interaction with walk-up computer workstations, large screen displays, conventional PCs in a media center and (via the museum's web server) the visitor's home computer. For particular artifacts, various kinds of augmented realities will be explored. These can vary from, say, a remotely controlled laser pointer to highlight parts of an artifact being discussed or explained in other media to projections of images onto or over the artifacts, as well as onto nearby display screens or even onto white space. We are also considering the potential for integration with other consumer technologies such as mobile phones and various kinds of wearable computers.

Related Work

Thomas and Mintz (1998) survey a number of recent approaches that employ varying multimedia technologies in different kinds of museums. They examine the potential for productive interactions between virtual representations and actual physical artifacts. Mase et al. (1996) propose the use of techniques related to intelligent agents and virtual reality to augment artifact representations. Milosavljevic et al. (1998) explore the use of dynamic artifact labels to tailor descriptions to the individual visitor (see also Oberlander et al. 1997). There are even attempts at "automating" the human docent by building robots to provide guided tours (Burgard et al. 1999). The use of PDAs as educational devices for children is being investigated by many researchers (Inkpen 1999; Soloway et al. 1999), as is their use as input devices for large shared displays (Myers et al. 1998; Greenberg et al. 1999).

Potential scenarios of use

A key part of the work of the project is to critically design, analyze, test and revise detailed scenarios to uncover the systems requirements and progress towards the development of prototypes and demonstrators. As an example a scenario is sketched out below.

Caveat: The scenario below is currently science fiction. We are not proposing to implement it! Rather, with careful analysis we can explore the implications of this rough sketch and propose a research path to develop self-contained components of such an idealized tour. Each component would be intended to add to the museum visit experience in its own right and also to contribute to the evolving vision. However, for the reasons outlined later, this requires careful exploration of the space of possibilities before diving in to the development of exciting technologies.

Example Scenario: The Enhanced Tour

The visitor to the Science Museum rents a PDA with attached headphones. Using a walk-up graphics terminal she chooses a particular tour that highlights the achievements of women in science and engineering. There is not a specific gallery devoted to this. Rather the galleries are themed by broad concepts such as "The Environment", "Our Bodies", "Work", "The Home". The tour will involve walking around the galleries, looking at particular exhibits from the perspective of the themed tour. The visitor indicates she wants the one-hour version of the tour (she can change this if she decides she wants to spend more or less time on the tour), and because she is a regular visitor to the museum, the system uses details about her profile to construct a tour using reasonable assumptions about levels of interest, background knowledge and references to things seen on other tours. (All these features can be over-ridden by the user, but allow slightly more tailoring than a standard one-size-fits-all vanilla tour). Last time she took a tour, she had planned her visit from home, using features at the museum website, and so could just pick up the PDA and get going. But this one is a spontaneous visit.

Putting on the headphones and looking at the PDA, she starts the tour. The PDA shows a map and an indication of where to go. As she is a regular visitor, this is a little superfluous, but useful for first time visitors who can get rather lost in finding the way to an out of the way collection. Best of all, because of the wireless connection, the map can update based on where in the museum the user is currently standing. Although map reading is perhaps a useful skill to be practiced, it can get a little confusing, particularly when your kids need the bathroom in a hurry.

Arriving at the Difference Engine exhibit, she listens to a more elaborate description of the life and work of Ada Lovelace than is available on the displays. Being a software engineer, she is interested in Lovelace, and chooses the option for even more information. The PDA has details about books and websites about Lovelace. She decides she doesn't want to use the nearby monitor to look at the websites recommended (PDAs are great for bits of information but its nicer to use a bigger monitor for lots of text), but notes that she wants the system to make this information available for use later (maybe at home) and to remind her when she gets to the gift store to take a look at the Byron family biography they have in stock. The PDA communicates with the museum web server, adding the chosen information to our visitor's personalized web pages there. It will use the spatial awareness from the wireless network to flash a reminder on the screen when it detects that she is near the bookstore.

Some kids are trying out the virtual Difference Engine simulator shown on the fancy plasma display nearby. As there is a whole bunch of them, their teacher has them using their PDAs to issue votes on what to do next. "So much better than handling the fights over who gets to hold the joystick - that's fine for small families but not for my girlscout pack", she thinks. Which reminds her, she must use the 'make your own tour' feature to put together something for them. She's promised to take them all to the museum next month and it would be good to assemble a load of hands-on activities, that also relate to the theme of her own work, so she can talk about that as well. But first, she wants to see what information they have about Grace Hopper, so she decides to carry on with the tour up to there and then she'll play around with the 'fun and educational activities to do and talk about with your kids' feature. She remembers a colleague talking about Hopper and decides that if the exhibit is good, she'll send him an e-postcard about it, choosing a few pictures and chunks of text and adding a quick note of her own. It's fun to spread the word about the museum's latest innovations this way.

The scenario illustrates some of the key issues that need to be explored:

Improving interactivity

As noted, for many people a museum visit is a social event. Technologies should encourage rather than inhibit this. Thus part of the work is to examine ways in which this interactivity can be enhanced. It might be by the system asking questions for people to think about and discuss. It might be by playing games, either using a simulation, trying to achieve a goal that requires a certain understanding of a concept, or by something like a treasure hunt around the museum, collecting clues that dynamically change the next thing to hunt for. As the scenario notes, PDAs can be used to control more sophisticated monitors or physical or virtual simulations. They can also be used for collective interactions such as voting, or sharing information as with MIT's work on nametags (Borovoy et al. 1998). Many things are possible and the question is how to choose activities that can achieve the different goals, including learning goals, for a museum visit. Research in constructivism emphasizes the importance of activity in learning. This is not just pressing buttons in a science museum. It can also include easy ways for the visitor to tell stories, take notes, construct diaries of a visit, or send messages to friends, either in the museum or elsewhere. Given that a wireless PDA knows where you are, it is possible to link brief notes to particular exhibits and to pre-existing text or graphics.

The GUIDE Project

The GUIDE system (Cheverst et al. 2000a&b) integrates the use of personal computing technologies, wireless communications, context-awareness and adaptive hypermedia in order to support the information and navigation needs of visitors to the city of Lancaster.

An initial requirements capture exercise involving experts in the field of tourism was conducted in order to identify some of the limitations associated with the traditional information and navigation tools available to city visitors. This process highlighted the need to provide visitors with a highly personalized and context-aware experience. In particular, information presented to visitors should be tailored to both personal context (e.g. the visitor's interests and current location) and environmental context (e.g. the current state and position of attractions within the city). In addition, the system should support a range of functionality in order to enable visitors to explore and learn about the city of Lancaster in their own way. The GUIDE system has thus been designed in order to enable visitors to use the system in a passive way (i.e. by following a structured context-aware tour) or in a more active way (i.e. by using the system like a guidebook with the goal of information retrieval) depending on their own preferences.

The system assists visitors with the task of retrieving information by utilising context (most notably the visitor's current location) in order to both constrain and prioritize the information presented. For example, the visitor may ask the system to display information on 'nearby' attractions. In this case the system presents a list of those attractions within a few minutes walk of the visitor's current location and the list is ordered in such a way that those attractions that are open, and have not already been visited, are placed higher up the list

A field-trial based evaluation of GUIDE has been conducted (Cheverst et al. 2000a) which focused on ascertaining the quality of the visitor's experience when using the system. The evaluation revealed the public's general acceptance of using a system such as GUIDE and that many visitors, especially those belonging to a younger age group, found the technology employed by the system exciting. However, the evaluation also revealed the need to avoid overly constraining the functionality provided by the system based on the current context and the need to provide less function rich systems for those visitors wishing to interact with something more resembling an information appliance.

Future plans for GUIDE include extending the range of context-aware services to meet the needs of city residents in addition to city visitors. We also intend to extend the communications infrastructure in order to provide coverage within buildings such as the city museum.

Issues in GUIDE

Many of the potential uses of a system for visitors to a historic city are just the same as those relating to visiting a single museum, as noted above. The technological constraints and opportunities are slightly different however, subtly shifting the design space. A different kind of wireless technology may be used, which may or may not change the nature of the functionalities that can be developed (or can most easily or economically be developed). In general, it will be more complex to provide wireless access around a city as opposed to around a museum. This means that issues of quality of service become more prominent, particularly those relating to varying quality of service. Thus if we aren't sure that the dynamic updating of information (from either push or pull technologies) is guaranteed to occur in a timely manner, we need to design in features to warn the user that at the moment they have only a weak or non-existent connection, and that things will probably get better as they move around. In the museum it is much easier to guarantee a minimum degree of access, obviating the need to even make the user aware of the issue.

In addition, the needs of the intended users change somewhat. Visitors to both museums and cities wish to be educated, entertained and informed, but in the case of city visitors, extra practicalities arise. While there is little difference in the content aspects of listening to or reading information about a statue in a museum room or a statue in the market square, the potential for augmented reality is more limited (but not eliminated) in the latter. Also it might start to rain (not a rare event in Lancaster…)! That is, there are more unexpected issues requiring re-planning that can arise in citywide use than in museumwide use. This adds to both the complexity of the scenarios and the opportunities for supporting visitors. Examples include: helping people to navigate round the city, integrating cultural activities with shopping, rest breaks, visits to cafes and planning for subsequent activities later that day or during that visit.

Adding collaborative features to GUIDE

As in the case of the cyberdocent, we are investigating the support of additional forms of collaboration in GUIDE. These include the use of social navigation of possible tourist activities, incorporating collaborative recommendations, the sharing and re-use of other people's trails round the city and the use of expert intermediaries (such as Tourist Information staff). We are using a similar technique of generating a set of different scenarios to explore and illustrate how different technologies can help in regular tourist activities of pre-planning, visiting places, interacting with the resources and people, purchasing, dynamic re-planning, contacting people back home and taking advantage of serendipitous discoveries.

A similar approach to scenario-based design can be used to explore interesting new functionalities. For example, while walking round a city en route to a recommended attraction, you see an interesting craft shop. You decide not to go in right now, but note it on your PDA as a place you'd like to go back to. The PDA knows where you are because of the wireless connection, so selecting the relevant shop is easy. Useful information can be called up (like this shop's rather erratic opening hours). Later, when time permits, it is easy for the system to help you find the shop again. Contrast this with the common occurrence of walking past an interesting place in a strange city and being unable to find it again. Indeed, even if the user had not made an explicit note of the shop, the system can still give a time-stamped trail showing the streets that they have walked along, and consequently the buildings and shops that they have (and have not) passed.

We are currently investigating how the techniques of context aware computing and the notion of social awareness can be combined in order to support and encourage cooperation between GUIDE users (Cheverst et al. 2000c). In more detail, we have modified the system in order to encourage social interaction by providing visitors engaged in exploring an attraction virtually with an awareness of those visitors that are physically located at the corresponding attraction. Further work will explore ways in which context can be used to help visitors determine the appropriateness of interrupting another visitor. For example, given the context 'I am sitting in a café and I am alone', then other visitors could be made aware that this is a reasonable time for me to be interrupted. However, given the context 'I am visiting the museum and it is about to close', other visitors could be made aware that this is not an appropriate time for me to be interrupted.

Scenarios relating to social awareness and collaboration include the following:

You are thinking of going to the Maritime Museum, but in checking out the current city visitor activity screen you see that it is packed out with people. The system notes that this is relatively unusual (actually it is caused by two coach loads of tourists arriving at once). You decide to wait until it quietens down a little and so request the system to keep you posted and head off for a coffee. Later, after the museum, you are at a loose end for an hour and want some suggestions. In addition to recommendations based on you current patterns of activity and what others who have done similar things have also done (collaborative recommendations), you also check out the overall visitor activity screen to see what is currently all the rage. Note that this is the exact opposite of the Museum example - here the popularity is being used as a positive rather than a negative recommendation.

Another example of social awareness involves temporal and spatial coordination. John and Mary are on a coach trip. They want to do some things with the group, some things together and spend some just doing their own thing (John likes ships and wants to visit the Maritime Museum, Mary can't think of anything more boring). Drawing on analogues of how people use mobile phones to coordinate their lives we are exploring functionalities to help people keep track of specific individuals and groups, send messages, re-plan as circumstances change and still get back to the coach before it leaves. Examples include John finding two other people who are also model ship builders, Mary getting recommendations of the best commercial galleries in the city, John and Mary arranging to meet up in a café neither has actually been to before, and both of them finding out where the rest of their group is.

Both GUIDE and the Cyberdocent are systems intended for the context of leisure use. This requires careful design to ensure that we provide services and features that people want to use, and that the various mechanisms employed (including controls for addressing privacy concerns) are sufficiently easy to learn and use that people will be willing to bother to deal with the technology at all. Leisure use such as this is a good extreme test, since users are only likely to be using the technologies for a few days at most and typical use will be for just a few hours. Benefits of use must be clearly apparent and costs of learning and use extremely low.

Broader Research Problem: Towards Informed Systems Design

It is a persistent problem in Computer Science that there is inevitably a gap between the developers of a system and the intended end users. In Software Engineering this is known as the Requirements Capture problem. Life is so much easier when Computer Scientists design systems to be used by themselves or by other Computer Scientists. That is because they know thoroughly the nature of the problem, what users currently do, the difficulties and inefficiencies that they encounter, what they would like to do, and what is currently technologically and economically possible to design. This applies both to regular applications and to innovative research and development using cutting edge technologies. From a consideration of this idealized state, we can see the problems that will arise when addressing more complex situations involving various kinds of users or publics. What do the end users really want to do? What do they actually find difficult with the existing arrangements? What is currently possible? Effective design needs a considered interaction between user needs and the current economic-technological space of possibilities. All too often that does not occur. Where users are insufficiently involved, there is the danger of technology-led applications, solutions in search of a problem, applications that solve problems that people don't have, and that fail to solve, or even worsen the problems that users actually have.

There is a substantial literature, particularly in Computer Supported Cooperative Work research, of case studies of deployments that failed precisely because of an inadequate understanding of what real people really do in their work and the consequent development of systems that although masterpieces of ingenuity, solved the wrong problems (Grudin 1989). Equally, just involving the local experts is insufficient. Such a person may be very clear what they want ("I want to tell my computer to get me all the correct information, like on Star Trek") but unaware of the realistic potential of current technologies. This may be an over-estimate of current abilities based on hype and science fiction, or equally unfortunate, an underestimate, or lack of systems development insight to see how a combination of technologies could be employed, maybe not to 'solve' their problem, but to provide a distinct improvement on existing practice. Thus if you don't know that a solution to a problem is now possible, you may not even think of it as a problem, you may not even be aware that you have the need for it. This is illustrated by the frequent difficulty of marketing radically new technologies. You have to explain to people why they would want them at all, if they've never had that option before. Hence, you couldn't have invented it in the first place solely by asking them what they wanted.

What is needed is more work on bridging the gap between the explicit and implicit needs of the user and the current state of the art (and what could be reasonably done to advance the state of the art). In this paper we have considered the case of city tourism and museum visits, but clearly the approach has far wider implications. To be effective, we believe that the approach has to resist the Computer Scientist's temptation to 'rush to code'. Normally that is taken to mean the desire to start coding and skimp on systems design. We are taking it to mean an even broader approach - that as soon as an issue arises, there is a temptation to work on an intriguing design solution. Instead we believe it is in the long term more productive to first explore the larger design space of many possible design challenges, to review the technical feasibility and likely payoffs of the different approaches and only then begin considering the normal design issues. Such an approach is based on user studies, detailed analysis and the application of scenario based design. At its best it can lead to the choice of a particular project area to focus on that shows the greatest chance of eventual end user take-up and success. It can indicate critical factors, especially about usability, that must be addressed and are often ignored in the pursuit of ever more impressive functionality. It can even produce a whole range of development options and uncover critical technical research issues that only arise when a certain juxtaposition of technologies and needs occur. For example, we expect that this work will raise significant issues for Quality of Service, data compression, data quality, CSCW, Computer Supported Collaborative Learning and HCI research.

Scenario based design (Carroll 1995) does not just consist of thinking up illustrative stories. It involves the use of these stories of plausible activities to explore the design space. By examining a scenario, alternative functionalities can be considered. These functionalities can then be ranked both by expected impact on different (perhaps sometimes competing) design goals (such as maximizing learning, maximizing enjoyment, maximizing interaction) as well as by expected development costs (including for innovative designs the tractability of the research issues that would need to be addressed). Choices can then be made about future research and development based on this multi-factorial analysis. Furthermore, to explore a design space in greater depth, a scenario can be adjusted by changing some of its assumptions to other plausible values and exploring likely breakdowns and how these should be addressed. These what-if sub-scenarios form a wider space around the originating scenario, and some of them can be promoted to first class scenarios themselves.

Selected References

Borovoy, R., Martin, F., Vemuri, S., Resnick, M., Silverman, B., and Hancock, C. (1998). Meme Tags and Community Mirrors: Moving from Conferences to Collaboration. Proceedings of the 1998 ACM Conference on Computer Supported Cooperative Work 159-168.

Burgard, W., Cremers, A.B., Fox, D., Haehnel, D., Lakemeyer, G., Schulz, D., Steiner, W., and Thrun, S. (1999) Experiences with an interactive museum tour-guide robot. Artificial Intelligence (in press).

Carroll, J. M., Ed. (1995). Scenario-based design: envisioning work and technology in system development. New York, John Wiley & Sons, Inc.

Cheverst, K., N. Davies, K. Mitchell and A. Friday. (2000a). Developing a Context-aware Electronic Tourist Guide: Some Issues and Experiences. In Proc. of CHI'2000, Hague, Netherlands, 17-24, ACM Press. (April, 2000).

Cheverst, K., N. Davies, K. Mitchell and A. Friday. (2000b). Experiences of Developing and Deploying a Context-Aware Tourist Guide: The GUIDE Project. In Proc. of MOBICOM'2000, Boston, ACM Press, pp 20-31. (August 2000).

Cheverst, K., Mitchell, K. and Davies, N. (2000c). Providing Tourists with Social Awareness Through Context-Awareness. Submitted to the Workshop on Awareness and the World Wide Web: ACM CSCW'00 Conference on Computer Supported Cooperative Work, (Philadelphia, December 2000)

Greenberg, S., Boyle, M. and LaBerge, J. (1999). PDAs and Shared Public Displays: Making Personal Information Public, and Public Information Personal. Personal Technologies, Vol.3, No.1, March. Elsevier.

Grudin, J. (1989). Why groupware applications fail: problems in design and evaluation. Office, Technology and People 4(3): 245-264.

Inkpen, Kori M. (1999). Designing Handheld Technologies for Kids. Personal Technologies Journal, 3(1&2), pp. 81-89.

Mase, K., Kadobayashi, R., and Nakatsu, R., (1996) Meta-museum : A Supportive Augmented Reality Environment for Knowledge Sharing. Proceedings of the International Conference on Virtual Systems and Multimedia '96 (Gifu, Japan: IEEE Computer Society, 1996): 107-110.

Milosavljevic, M., Dale, R., Green, S.J., Paris, C., and Williams, S. (1998) Intelligent Interactive Virtual Museum on the Information Superhighway: Prospects and Potholes. Proceedings of CIDOC '98, the Annual Conference of the International Committee for Documentation of the International Council of Museums. http://www.dynamicmultimedia.com.au/papers/cidoc98/

Myers, Brad A., Stiel, Herb, and Gargiulo, Robert (1998). Collaboration using multiple PDAs connected to a PC. CSCW '98. Proceedings of the Conference on Computer-Supported Cooperative Work, Seattle, WA, . ACM Press, pp. 285-294.

Oberlander, J., Mellish C., and O'Donnell, M. (1997) Exploring a gallery with intelligent labels. In Proceedings of the Fourth International Conference on Hypermedia and Interactivity in Museums (ICHIM97) (Paris, September 1997).

Rayward, W.B. & Twidale, M.B. (1999). From Docent to Cyberdocent: Education and Guidance in the Virtual Museum. Archives and Museum Informatics 13(1) 23-53. http://www.lis.uiuc.edu/~twidale/pubs/docents.html

Soloway, E., Grant, W., Tinker, R., Roschelle, J., Mills, M., Resnick, M., Berg, R., & Eisenberg, M. (1999). Science in the palm of their hands. Communications of the ACM, 42(8), 21-26.

Thomas, S. and Mintz, A., (eds), (1998) The virtual and the Real: Media in the Museum. Washington, DC: American Association of Museums).

Twidale, M.B. (1995). How to Study and Design for Collaborative Browsing in the Digital Library. Invited talk given at How We Do User-Centered Design and Evaluation of Digital Libraries: Methodological Forum, Thirty-Seventh Allerton Institute, Allerton Park, University of Illinois, Monticello, IL, October 1995. http://edfu.lis.uiuc.edu/allerton/95/twidale.html