[SICS | KFB | NUTEK | Workshopens hemsida ]

A Summary of the Discussion

Below follows first a summary of the discussion and comments done by Kristian Simsarian and then a transcript of the discussion as interpreted by Kristina Höök. Participants in the workshop are most welcome to add comments - send those to kia@sics.se and she shall link them to this page.

Summary by Kristian

Transcript of discussion by Kristina

Summary and comment

by Kristian Simsarian

Through the course of the meeting a number of subjects were brought up and discussed. Kia has categorized the comments into the following: Problems for the disabled, How do we recruite users, Elderly Users, Tools change people, where do new designs come from?, What are agents good for?, see below. Because of the time limitation of the workshop, many of the issues did not receive the in depth discussion they possibly deserved. Also in the end we realized that the discussions did not necessarily center upon the themes invoked in the workshop title (i.e. VR and Agents). Themes centered more around the area of "computer-interface technology and users with disabilies." Within that area there is an identifiable set of broader themes that came up and are perhaps closer to the application process. These are worth bringing out again in this summary. These themes are the concepts of the Initiation, the Construction, and the Acceptance in respect to novel technological systems.


There is probably a fundamental difference between the systems users ask for and the systems technologists propose. In the discussion it was mentioned that user originating ideas may tend to bring small improvements while ideas that originate in technology could provide leaps. As was said at the beginning of the meeting "technology researchers are not always aware of particular problems but may have the solutions." In an ideal situation, a user will ask for fantastic tools they think are not possible and the technologist will realize that is simple application of understood (or near-understood) technique. Workshop particpants seemed to agree that the optimal case is somewhere inbetween, e.g. a cooperation between users and technologists, where this link comes at the beginning. This would strongly suggest the need for communication (both formal and informal) between the people with different expertise.


We discussed a few of the ways systems can be contructed in conjunction with users (see discussion below). In the optimal case we want end-user involvement throughout the process, but there are costs associated with this, both in time and resources. How this should be done is an actively debated aspect of the HCI field.


Acceptance of a system can be a long-term area in terms of feedback. It is an unfortunate, but true, factor that issues such as resource investment and time pressure can cloud this issue. Previous biases will also be factors here when trying to introduce technologies that have simularities to earlier flawed projects. Knowledge of such flaws are thus important in the construction phase. There are also some sociological notions involved here where it is easy for technologists to be seen as forcing technology on users, possibly for the sake of technology. These sort of misunderstandings can be due to a lack of good communication (due to distance and/or language) between involved parties.

In summary there is a set of issues that emerge in this process and should be addressed. These centered around:

Technologists need to be alert for the introduction of extraneous technology, and need to remain conscious of the end goal (e.g. "what activity is this tool meant to achieve, are we producing side-activities?"). Also technologists need to be willing to involve the users in their development process (this is certainly not the norm). This involvement is especially true when the users have special needs that are different from the greater population. People need to be found in the care industry who are open to new technology. The particpants in the medical and social care side need to be willing to be active in helping to bring the technology in.

A way of networking should be available that allows informal contact in this field to take place. Both Jan-Ingvar and Hans mentioned that they have established contacts with users that have disabilies and are interested in exploring technical tools. As a demonstration of this of this we should make sure that all particpants have a address list of the workshop particpants including a brief description of the institution they represent. In the future we should work toward adding to this list contacts that sponsor disabled groups that might be willing to work with technology.

As final words, there should be something said about the two technologies that were demonstrated. Through the course of the morning and early afternoon we were given an outline of some of the applications of the technologies of Virtual Reality and of Agents. These technologies are separate fields but we did see a few examples where they were used in conjunction. For example a user can enter a world and be approached by an "agent" who can take the user on a tour, or provide requested information.

Generally, I see the broad area of VR as helping users to experience situations where they might otherwise encounter physical restrictions due to physical aspects of the world. Thus VR can help to create another reality where events can be slowed down, controlled, and adapated to users needs. It can also create a new tool for communication were such communication or collaboration would not have been possible before.

While VR might reduce the physical demands on a user, the field of Agents can contribute to the reduction of the cognitive load on a user. A user may be relieved of having to explore the details of a number of situations, or consider every option, if a "personalized" agent is able to present the options that an individual user is interested in at that moment.

Transcript of the discussion

by Kristina Höök

Below follows a transcript of the discussion.

Problems for disabled

Lars-Äke initiated the discussion by pointing out that it is of crucial importance to start out with getting a better understanding of disabled users' needs. It is furthermore important to see the differences that comes with different disabilities.

Kia pointed out that design of systems for disabled is as system development in general: a process where it is both necessary to understand user needs, but also see the technical possibilities and limitations. Design is an art. This means that a complete understanding of users' needs will not automatically render a good design. Creativity is needed in the process. Kia also pointed at the need for methods, and discussed one such method (that has some severe limitations) namely user-participatory design.

Britt wondered whether user-participatory design meant that a "user" is constructed, but Kai explained that it means that actual "real" users participate in the design meetings and help to come up with ideas. Unfortunately, it is a long process since both designers and users have to learn eachothers' languages and understand the terms that define the space of possibilities.

Jane pointed out that a reference group of "experts" that understand the needs of a particular group of disabled might be a better alternative.

Paul pointed out that most existing methods for HCI design have been developed to fit with "normal" users and that we might have to adapt or invent new methods.

How do we recrute users' to these projects?

Jan-Ingvar then brought up the problem of recruiting disabled users to these projects - both to participate in the early design phases but also for evaluating and using the systems produced. Anders B. pointed out that e have very strong and active organisations for disabled in Sweden. They are also highly motivated to participate and really want technical aids. In particular, he talked about his work with mentally disabled children. They see a need for technical aids since they really want to improve their problem solving skills. Once they get these kind of tools they cn express themselves and show that they are "humans".

Gerd brought up another problem with recruiting users, and that is the risk of only getting in contact with users who are active and already knowledgeable in how to use computers. Some systems also needs to be tested on disabled users who are novices in computer use. Jane pointed out that "day centers" are a good place to recruite users.

Tools change people - how good are the methods on evaluation?

Jan-Ingvar expressed a fear that the methods used to test whether tools are usable or not seem to neglect the fact that users change through using the tools. The methods used only involve a small set of novice users and it is not clear what happens when the tool is used in the long run?

Britt added that this points at a focus problem for researchers: should we study the reality that disabled have or should we invent new "realitites". The fact is that as soon as we enter new tools into their reality, we shall change their reality no matter what we think. Maybe it is better to start from this perspective?

Jan-Ingvar replied that there is a difference between doing it on behalf of users of just force it upon them.

Kai pointed out that novices stay novices for a short time. Users change. That has to be taken into account.

Kia pointed out that evaluations of systems should control for differences in experience of both computer use and degrees of disability. The problem is to do both in the same study - again new methods might be needed.

Paul pointed out that there are two quite different approaches to research: one is problem driven (users' needs) and the other is vision (or technique) driven. The former may provide small steps forward, the latter might give big steps but also involves greater risks as one cannot be sure that the technique developed will meet the users' needs. Both are needed!

Gerd proposed that case-studies of real situations where you take whatever subject you can get is most benefitial in the early stages of design. Later on in the design cycle you can do more formal evaluations with many real users. But Kai replied that it can be very hard to do case-studies when we are trying to invent completely new technology? Sheri proposed single-user studies as an alternative method. By studying one subject and how s/he changes over time, much can be learnt. This is a method that fits well with disabled users as these often differ a lot and cannot be compared with each other. Also it is not so costly.

Hans came with a down-to-earth realistic suggestion and that was to get in contact with the users that he and his collegues had connected to the Internet. Both disabled users and hospital personel were attached. They have tried email and chat etc. Chat did not work at all. He also proposed that dyslexiker might be one group that would benefit a lot from utilising agents.

Elderly users

Britt has investigated what might be particular to the elderly user group (which is not caught by for example comparing them to other disabled groups such as blinds, hearing impaired, etc.). She had found that three problems are prominent:

She pointed out the importance of differentiating between old-age diceases and changes to peoples cognitive abilities. Britt also pointed out that the group of elderly people we have today are used to change and new techniques being brought into society.

Kai added that in general it takes something like 2000 hours to become an expert on a complex system. Most computer systems that we have have not been around so long so that users could have had a chance to become experts. In fact, they will not be around so long since they are recurrently released in new versions.

Coming up with new designs?

At this point Lennart raised his voice and declared that he found the views expressed boring. People do not change. We must come up with new things derived from people's real needs. What we have now are small useless machines designed by and for nerds. We must be the ones to design the new systems for the future. They must be a natural to people as a pen, hammer or telephone. The most sophisticated use of technique that Lennart can think of is two people sitting silent together on the phone. Through out the history and think about how we can construct completely new applications. Right now we are surrounded by systems built by technicians that forces the users to adapt (rather than the other way around?).

Britt agreed and said that often people phone her up and ask for check-lists that they can use to check whether their design is good enough for disabledor elderly users. That is of course the wrong approach. But she also felt that the answer is between user-needs driven and technique-driven research.

Lennart then pointed out that one must not be afraid of technology. Good designs can come from anybody - one does not have to be good at maths to influence the design of a system.

Anders pointed out that this is a circle: first some examples of technique (prototypes) then we can take user needs into account and build better systes. Also sometimes it is necessary that many users use a particular system before its usefulness can be determined.

At this point Kia wanted to become a little bit more academic and started to discuss methods that claim to be able to foresee or at least be aware of that fact that new tools will change what people do and why. In particular, she was referring to activity theory (see Bonnie Nardi's book Context and Consciousness). Even of activity theory may not succeed in actually being able to predict what sort of changes (in both individuals and organisations) that will occur when a new tool is introduced in an organisation, it is still interesting to read about their critique of the Human-Computer Interaction field. Activity theorists say that no really innovative and good design ideas have come from HCI researchers performing studies of users and their needs. Instead the most creative and useful designs have been initiated by technicians. Of course, activity theory researchers claims that activity theory will do the trick - but this is not at all clear.

Anders B. pointed out that the activity theory is a good description of the fact that people (also mentally retarded) will be active, conscious beeings that actively take in information. The view should not be that the machine communicates information to the user -- it is the human that actively takes the information in.

Britt added themethod named SCOT (Social Construction of Technology) as an alternative. (Pinch and Baker?).

Jan-Ingvar was more cautious and said that we must not forget that we work with people that have a limited ability to take in information. They are more controlled by their environment. This is an ethical problem to us researchers, since we may force solutions upon them. Jane added that it is important to realise that these users may not be able to nor allowed to decide whether they can participate in an experiment or not. Instead their parent (or ward) will be the one who decides.

What are those agents good for?

Lennart raised his voice again and was annoyed by the fact that all technicians were accused of seeing human beings as machines -- he certainly does not! In fact, he believes that little is known about human intelligence and that is the lesson learnt from the four waves of Artificial Intelligence. So instead of trying to imitate human intelligence we would do good in studying interactivity. Interactivity allows people to take control, to change the environment created by the system. In a Gibsonian-sense he would like to see "affordances" as a means to communicate to people what they can do in the environment. Self-organisation, etc. Otherwise nobody will like to be in those environments.

Kai pointed out some limitations with agents, such as they can only operate on what they know. They can sometimes not explain how they work to the human user since their internal representation may be a neural net.

Kia pointed out that complete control can also be a hindrance. We may not always be interested in complete control over the systems we work with. In fact, we sometimes willingly hand over control to for example car mechanics to fix our cars without having complete control over what they are going to do. As long as their service is good enough, we won't be bothered. The same goes for agents: if an agent can provide me with a really good service, for example, finding high-quality information that interests me, then I may not be bothered by the fact that I cannot control its inner workings. Kia was also worried by the connection between "old" AI and agents -- that connection is not a good one.

Kai at this point wanted to bring up the philosophical discussion of what intelligence is and whether we can know what it is.

Anders H. then posed the crucial question (that of course should be asked): we shall judge systems from they can by us in terms of mileage, what can agents provide us with that is not possible already with ordinary systems?

Related to the question introduce by Anders H., was a question by Sören. He could see what agents would be able to do for him, but he was wondering what agents would do for his 80 year old father? He is clear over what he wants out of life: socialising with his grandchildren, and handle the shores in his daily life. Sören expressed a fear that unless we can help elderly people to use IT, our society will be divided in half.

Kia said that agents might not be useful at all to Sörens father but that Internet might well be as a means to communicate with children and grandchildren. Her view on what agents should look like, whether they should be controllable or not, what they should be used for, etc. really depends upon the task we want to solve. Sometimes agents should be visible in the interface as little faces or dogs or squirrels, and sometimes not. Kia said that three problems were such that direct-manipulation did not suffice:

Possibly we could also include the need to enhance communication between people (as in marketplaces, MUDs or whatever).

What was this discussion good for?

Kristian, who had been sitting silent listening up until now, now poses the question: what is it that we want out of this disucssion? He had the feeling of being an indian in a teepee-tent: everybody got to say what they think, but that does not bring us forward. Kristian wants to have some more brainstorming around the actual ideas for systems rather then being told that he should take care to think about users' needs. He wants some networking to come out of the workshop - to get in contact with people.

Kristina Höök