Contents - Previous - Next

This is the old United Nations University website. Visit the new site at http://unu.edu

4. What can the technology do now?

The technology has moved rapidly in the past few years. The enabling technologies required to support a multimedia workstation go far beyond provision of the media themselves. A typical list [15] would include:

- Bit-mapped displays
- Audio and video support
- Device independent graphic rendering
- Control and synchronization
- Data compression
- Networking support
- High CPU/disk performance
- Authoring tools

To illustrate how far current workstations have come in being able to deliver such a collection of enabling technologies, Malleo-Roach cites the SUN SPARCstation as an example and provides the figures given in the table.

To give an idea of the costs associated with multimedia support, he suggests three possible configurations: "minimal," "enhanced," and "operational," with costs at about US$10,000 and US$35,000 for the first two (the third not yet being available). He expects an "operational" version to be available in 1995 for about US$10,000.

Figure 2

Multimedia capabilities have also already arrived on most personal computers. There are multimedia extensions to Windows, and Apple now offers "Quick Time" on the Macintosh. Quick Time offers three new components: the Movie Toolbox, the Image Compression Manager, and the Component Manager (figure 2).

The Movie Toolbox provides facilities for creating, editing, and playing back moving images. The Component Manager enables new devices (such as VCRs, camcorders, etc.) to be attached as well as third vendor software modules. The Image Compression Manager uses one of three compression techniques for animation, video, and photos. The JPEG (Joint Photographic Experts Group) algorithm is used to compress photographs. It removes data that is redundant to the human eye. Compression ratios of between 500:1 and 5:1 can be achieved with a full screen colour image taking about 10 seconds to compress. The video compressor can achieve ratios of between 25:1 and 5:1 giving about 15 seconds of compressed video per megabyte. Boards are available for capturing video directly from cameras.

Compression techniques are essential since a standard television frame contains nearly 750,000 bytes of information. In uncompressed form, even a compact disc can only store 30-40 seconds of video. Furthermore, since a compact disc delivers data at an error free rate of 150 Kbytes per second, and 10 per cent of this needs to be reserved for audio, each picture needs to be reduced to about 5,000 bytes to maintain motion. This corresponds to a compression rate of 160. This can be done by DVI (Digital Video Interactive) algorithms.

5. User centred or design centred?

Articles on multimedia tend to stress the technological side rather than utility aspects. There are many who appear to view multimedia techniques as a technology looking for an application. This technological-push view of multimedia is summed up in the introduction to an extended set of articles on multimedia in a recent edition of OpenLook [19]:

For while multimedia has the potential to transform computing, many of the components that create it are already available - it's just that users haven't brought them together yet.... Admittedly, the one thing that multimedia's been waiting for is applications. But they . . . are now coming on stream.

In other words, we have the technology, why don't you users want it? Because of the availability of the CD-I ROM and the richness of the media, educational technology has been a major target area for multimedia development. Many people extol the virtues of multimedia education, however one does tend to get the impression that the designers start with a multimedia tool kit and then try to find an application. As a result, a number of people are rather sceptical about current developments and the progress really made.

Tools for the construction and manipulation of collections of multimedia material are becoming widely available on successively cheaper hardware platforms. In terms of the educational use of these systems, however, it is unclear that any progress has been made. The effort in developing the technology has not been matched by a similar concern with the pedagogy.... At present it is an article of good faith that multimedia is a good thing for education and training. There is no evidence that multimedia enhances learning, or makes it more cost effective. [12]

Thus, the importance of multimedia applications lies not in what it can do but in what user goals it can solve more effectively. In Process Control, for example, the goals of the operators could be interpretation accuracy; problem comprehension; task performance; decision quality; speed of comprehension; decision speed; recognition and recall; or viewer preference [11]. Some of these goals will be more important than others - e.g. recall is usually less important than task performance. It is vital that interface designers always keep these user goals (rather than technological goals) uppermost in their minds. If this approach is not adopted, we could find that things get worse rather than better. The challenge of multimedia technology is to understand how to minimize information overload by the appropriate use of different media.

6. The PROMISE multimedia interface project

The overall objectives of the PROMISE project (PRocess Operators Multimedia Intelligent Support Environment) are twofold:

(1) the design' construction, and evaluation of a multimedia tool set for improving user interfaces to advisory and diagnostic expert systems used by single operators in process control environments;
(2) the development of a methodology to support designers using the tool set.

6.1 The Media Supported

The PROMISE system is multimedia, as it supports many ways of communicating information, and multimodal, as many styles of interaction are supported. The system supports a number of multimedia options [2], including:

- text, graphics, and sound output
- full colour
- live and still video output
- text and Mouse input
- two-dimensional animation
- unmediated video
- natural language output (speech)

The media "unmediated video" [1] perhaps require some explanation. The term subsumes many others, including algorithm animation, data fusion, visual realization, program visualization, and abstract representation. It is best illustrated by an unmediated audio example. In the 1960s, ICL engineers linked the data bus of a 1900 series computer to a loudspeaker, no doubt for hardware debugging purposes. The loudspeaker was left in the operator console of the computer and could be heard by increasing the volume control. It is almost certain that this information was never intended for operator use, but it soon became an extremely useful operator aid. Although the connection between the sounds emitted and the machine activity was never explained, operators were rapidly able to use the information to tell if the machine was idle, which job was running, and if a job was in a loop. At night, this use of the auditory channel also enabled such monitoring to take place in the operator tearoom adjacent to the computer room, probably the first recorded example of a multimedia interface advantage. There are other, more visual examples of this approach that have been used in process control.

6.2 The Environments

The multimedia approach in PROMISE is being applied in two distinct types of environment that will help us in evaluation - a real-time application and a simulated one. The real environment puts serious constraints on the evaluation environment, since the clock cannot be stopped and unanticipated events can occur. Such an environment also poses other problems from an evaluation standpoint. Errors cannot be deliberately introduced or reproduced, and controlled experiments are difficult. Therefore a simulated environment where we can control the nature of the upsets and monitor operator performance will also be used. The two application areas chosen are control of a chemical plant (DOW Benelux) and the use of power station simulators (Scottish Power).

6.3 The Unique Features of Multimedia Usage in Process Control

Most multimedia ideas grew out of work in computer education. The educational multimedia approach involves the use of videocomputers, interactive television, and electronic books to provide a richer educational environment. It is envisaged that a user will be able "to browse through vast libraries of text, audio and visual information" [4]. These libraries will be highly interlinked using a technique known as "Hypermedia," a concept developed from Hypertext that was first defined in the 1960s and led to the Xanadu system [18]. Most of the current literature about multimedia approaches derives from the educational sector. These educational applications tend to be activated by the learner alone where the user of the multimedia system is in complete control.

In contrast, the process control environment places extremely demanding requirements on the architecture of a multimedia presentation system.

- The operator may be engaged in many distinct tasks simultaneously, for example monitoring, tracking alarms, etc.
- New tasks and hence interactions may be instigated at any time by either system or operator.
- The process state may be extremely dynamic, so information may need to be presented rapidly and in a form that is readily understood by the operator.

The designer of a user interface cannot know in advance what combinations of multimedia resources might be required. The problem may therefore be thought of as a resource management problem, managing the competing requirements of different media for the limited resources of the interface. The existence of additional media provides problems and opportunities. The problems are those of "media-clash" - for example, several alarms wish to use the audio channel at the same time. The opportunities come from the ability to switch media to overcome the problem.

6.4 The Architecture

The overall architecture is shown in figure 3.

The actual realization of an object is derived from:

- designer options: the designers specify alternative renderings of an object and provide measures of their preferences for such rendering;
- resource limits: the availability of the physical rendering resources;
- multimedia options: given a choice, what would be the preferred rendering?
- operator preferences: are there special operator requirements (i.e. colour blindness)?

At the far right are the dynamic system to be controlled and the "supervisory and control system" that controls it. The PROMISE system is an advisory system so that the operator can completely ignore it if required and can control the process separately. Indeed, no direct control of the system is possible through the PROMISE terminal. An "advisory/diagnostic knowledge based system" is shown in the figure, though such a system is not actually part of PROMISE. At the heart of the tool set is the "interactive data model." This module models the relevant portions of the plant and all objects rendered to the operator. It is therefore a key player in consistency maintenance. The "presentation server" actually renders the images, sounds, etc., under control of the "resource manager," which decides what media will be employed taking into account the designers' preferences, the current resource usage, the user preferences, and general rules about rendering (the M41 Knowledge in the diagram).

Figure 3

Currently, this tool set is being installed in the chemical plant and in a nuclear power station simulator so that experiments can be carried out on the effectiveness of multimedia interfaces. In parallel, a series of experiments are being carried out in the behaviour laboratory at Loughborough University.

7. How does one design a multimedia interface?

Although the project is ongoing, we can already provide some guidelines as to how to design a multimedia interface. Our approach is one of experimentation combined with pragmatism (figure 4).

Figure 4

A task analysis [20] will be carried out on the operator task to establish the information needs. These needs then must be characterized in terms of a defined information processing need using a knowledge characterization scheme. The options that the various media provide are then matched to the needs to obtain a set of multimedia possibilities. Finally, an envelope of defined interfaces can be derived by checking the possible interfaces against our knowledge of the human cognition system.

The knowledge characterization scheme identifies a number of important information communication features that describe the knowledge to be communicated to the operator, for example. More details are given in Alty and Bergan [3].

- The number of dimensions needed:
- Ordering or sequence:
- Relationships:
- Dynamic:
- Static:
- Persistence:
- Instance:
- Information content:
- Urgency:
- Context change:

There are, of course, many more. We then map these information requirements onto appropriate media using another list of media properties, for example:

- ICON is 0-D, has no ordering, limited relationships, static, is persistent, and has limited information content.
- PICTURE is 2-D, strict ordering, probably strongly related to real world, static, persistent, and has high information content.
- VOICE OUTPUT is 1-D, strict ordering, strongly related to real world, dynamic, non-persistent, and has medium information content.

Finally, the options that result are examined for psychological appropriateness, i.e. those that best exploit the human cognition system. At this point, all objects currently being rendered (if known) are considered. An example of the sort of knowledge that is used in the psychological underpinning is the set of rules developed by Kosslyn et al. [13] - five principles of articulate graphics. Kosslyn points out that graphical displays must be articulate to succeed. They must transmit clear, compelling, and memorable messages. His five principles are:

(1) Gestalt laws of perceptual organization
(2) The influence of knowledge on perceptual organization
(3) Incremental transformation of images
(4) Different visual dimensions are processed by different channels
(5) Colour is not perceived as a continuum

For example, incremental transformation of images is important for creating and maintaining mental images. In the real world, objects do not disappear and reappear elsewhere. So pictures on VDU displays should change incrementally. Other important psychological principles are the 7 + 2 rule of Miller [16] and issues concerning the coherence or interference between visual and auditory channels. Application of the latter two rules, for example, make audio explanation with video highlighting a superior mechanism for explaining a complex diagram than the use of the visual channel only.

8. Some initial guidelines

Multimedia interface design has only one purpose - that of achieving some goal in a "better" manner. It is therefore important to clearly establish the goals to be achieved and what is meant by "better." In the PROMISE application, we decided early on that the key problem in process control was information overload with too little time for cognitive processing. Thus our major goal is to use multimedia in such a way as to buy time for the operator by rendering the information in such a manner as to enable it to be assimilated more quickly. Other goals include error reduction and better job satisfaction. Multimedia design is therefore critically bound up with the tasks being performed and the skills of the operators at the interface. Even so, we can make some general observations:

8.1 Real World Sensing

Multimedia interfaces allow us to reconnect the operator/user with the real sense world. The history of process control has been the gradual separation of the operators from the system they are controlling. In former times, operators were able to see, hear, and feel the plant. We have created a media-poor world that the multimedia approach can recreate. When operators close a valve they should see and hear it close. In PROMISE, we expect to have a microphone icon that can be moved round the process diagram. The operator can then hear information from selected parts of the plant.

8.2 Maintaining Information Source Integrity

Human beings often do not know how they use information. From a set of sensory inputs they "know" what decision to take. It is therefore important to preserve the integrity of information sources, particularly if we are not absolutely sure what is important. In the PROMISE project, for instance, our studies revealed that converting a particular test to a number lost a great deal of important visual information. Now we preserve a video still of the test so that the operators can still make a full judgement.

8.3 Exploiting Human Cognition

There are certain aspects of human cognition that lend themselves to a multimedia approach. Overloading of one sense can often be avoided by using a second sense in parallel. In particular the audio +visual approach can be particularly powerful.

8.4 Changing Context

This aspect does seem to be best dealt with using the audio channel. Auditory warnings are most effective and allow the operator to concentrate on other aspects of the task.

8.5 Simulation

The visual approach is, of course, particularly well suited to simulation. A useful combination is that of real video and overlaid graphics.

8.6 Use of Non-verbal Sounds

Very short sounds are useful for alarms or status identifiers. Continuous sounds are more difficult to use effectively. They are best used when identifiable changes really matter. Their use for absolute measurement should, of course, be avoided. The critical issue is whether the important changes can be recognized by the human observer. Continuous sounds can also be very irritating.

9. Conclusions

I hope that this paper has shown that the key issue facing multimedia technology implementers is not primarily a technical one (although technical performance is important) but a design one. We need to know where and when particular media combinations will enable us to achieve our goals more effectively. Because the advantages (or disadvantages) of multimedia usage depend upon the tasks being performed, it is difficult to set up precise laboratory studies and apply the results directly to real life situations. The answer here is probably more longitudinal studies in real situations. However, we are now reasonably confident that our list of guidelines will eventually extend into a methodology.

There are issues that have not been addressed in this paper. For example, dynamic allocation of multimedia resources raises major issues about interface consistency. We believe that, provided the representations change only occasionally and only in critical resource situations, this need not be a serious problem. However, this is a matter for debate.

10. Acknowledgements

The work reported here is part of the PROMISE project funded by the European Commission as project 2397 in the ESPRIT II programme. Their funding support is gratefully acknowledged. Thanks are also due to the partners in the project, namely Scottish Power (Scotland - Prime Contractor), Tecsiel (Italy), Dow Chemicals (Netherlands), University of Leuven (Belgium), IDS (Spain), Work Research Centre (Eire), University College Dublin (Eire), and Algotech (Italy).

References

1. Alty, J.L. (1989). The concept of "Unmediated Video and Audio" was first discussed in the original PROMISE Technical Annex (1989) (ESPRIT II Project 2597).

2. Alty, J.L., M. De Winter, P. Del'omo, and M. Zallocco (1989). "D2- Interface Definition for first M41 Toolset (PRO/6)." Deliverable no. 2, Esprit Project 2397.

3. Alty, J.L., and M. Bergan (1992). "The Value of Multimedia interfaces in Process Control." To appear in the Proc. of MMS '92, 5th IFAC, IFIP, IFORS, IEA Symposium on Analysis, Design and Evaluation of Man-Machine Systems, The Hague, Netherlands, June 1992.

4. Ambron, S. (1988). "What is Multi-media?" In: S. Ambron and K. Hooper, eds. Interactive Multi-media. Washington, D.C.: Microsoft Press, p. 5.

5. Beagles-Roos, J. (1985). "Specific Impact of Radio and Television on Adult Story Comprehension." Presented at a meeting of the American Psychological Association, Los Angeles, September.

6. Bolt, R.A. (1977). Spatial Data Management. Interim Report, Architecture Machine Group, MIT. DARPA Contract MDA903-77-C-0037.

7. Bush, V. (1945). "How We May Think." Atlantic Monthly 176 (1): 101-108.

8. Bush, V. (1969). "Memex Revisited." In: Science Is Not Enough. New York: Apollo Editions.

9. Carswell, C.M., and C.D. Wickens (1987). "Comparative Graphics: The Historical Development and Experimental Comparison of Alternative Formats." Aviation Research Lab., Savoy, Illinois. Unpublished manuscript.

10. Cawkell, A.E. (1991). "Multimedia and Hypertext." In: World Information Technology Manual 2. New York: Elsevier, p. 674.

11. DeSanctis, G. (1984). "Computer Graphics as Decision Aids." Decision Sciences 15: 463487.

12. Elsom-Cook, M. (1991). "Multimedia: The Emperors New Clothes. " DLT News 6: 1-2.

13. Kosslyn, S.M., C.F. Chabris, and S.E. Hamilton (1990). "Designing for the Mind: Five Psychological Principles of Articulate Graphics." Multimedia Review 1 (3): 28-29.

14. Maekawa, M., and K. Sakamura (1983). "Multimedia Machine." In: R.E. Mason, ed. Proc IFIP 9th World Congress, Paris, Sep. (Information Processing '83). Amsterdam: North Holland.

15. Malleo-Roach, J.A. (1992). "Bringing Multimedia Technology to the SPARCstation." SUNEXPERT3 (2): 60-65.

16. Miller, G.A. (1956). "The Magic Number Seven, Plus or Minus Two: Some Limits of Our Capacity for Processing Information." Psychological Review 63: 81-97.

17. Moray, N., P. Lootstein, and J. Pajak (1986). "Acquisition of Process Control Skills." IEEE Trans on Systems, Man and Cybernetics SMC-16 (4): 497-504.

18. Nelson, T.N. (1981). Literary Machines. Swarthmore, Penn.: Edition 87.1. Available from the author.

19. OpenLook (1992). "Multimedia: The Key to a Thousand Doors." OpenLook Editorial 3 (2): 39.

20. Payne, S., and T. Green (1989). "Task-Action Grammars: The Model and Its Developments." In: D.A. Norman and S.W. Draper, eds. Task Analysis for Human Computer Interaction. Hillsdale, N.J.: Lawrence Erlbaum.

21. Pezdek, K. (1987). "Television Comprehension as an Example of Applied Research in Cognitive Psychology." In: D.E. Berger, K. Pezdek, and W.P. Banks, eds. Applications of Cognitive Psychology: Problem Solving, Education, and Computing. Hillsdale, N.J.: Lawrence Erlbaum.

22. Rewey, K.L., D.F. Dansereau, L.P. Skaggs, R.H. Hall, and U. Pitre (1989). "Effects of Scripted Cooperation and Knowledge Maps on the Processing of Technical Material." J. Educ. Psychol. 81: 604- 609.

23. Vernon, M.D. (1952). "The Use and Value of Graphical Material in Presenting Quantitative Data." Occupational Psychology 26: 22-34.

24. Walker, J.T., and K.J. Scott (1981). "Auditory Visual Conflicts in the Perceived Duration of Lights, Tones and Gaps." J. Exp. Psychology: Human Perception and Performance (7): 1327-1339.

25. White, M.A. (1990). "Imagery in Multimedia." Multimedia Review 1 (3): 5-8.

26. Washburne, J.N. (1927). "An Experimental Study of Various Graphic, Tabular, and Textual Methods of Presenting Quantitative Material." J. Educ. Psychology 18 (6): 361-376.

Discussion

The discussion on session 3 began with an intervention by F. Thompson about the paper on "The Potential Offered by 'Extended Retrieval'." He said that the result of retrieval was expected to be a "file" that is to be directly used by the requestor. In the future, however, the computer itself will need to retrieve an item of data to be used in conjunction with many others to "compute" the response to the user. He asked: "How are 'addressing' into files to be developed and used?" M. Buckland explained that data within files may be found through indexes in the files themselves, and that inferences from the data might be made or aided by computer processing.

N. Streitz wondered about research conducted in library science, computer science, and computer linguistics in an attempt to solve the problems users face in looking for information. M. Buckland said that research was in progress in all three fields and that "successful research is likely to be a combination of these and other fields."

W. Rouse then took the floor to ask why the same technology could not be used to improve source documents themselves. In other words, could computer technology not be used to assure that documents are understandable? - otherwise they would not be published! M. Buckland agreed that new files derived from existing files could well be useful. He said that greater standardization of documents could well be helpful, but difficult to achieve.

Commenting on W. Rouse's position that authors should have responsibility for writing their articles so that better retrieval would be possible, J. Alty informed the audience that the Turing Institute had used the SEML Makeup language to do just this. SEML was used to mark up documents to show structure. The marking scheme was defined by the author. The Institute extended this idea to "semantic" marking - that is, words or fields in the document were marked up and could subsequently be retrieved. "This approach," Alty continued, "provides significant advances in information processing." He explained that they were able, for example, to code up the essential "if-then" rules in operator documents and later extract them and check them for consistency and completeness.

The next comment on the same paper came from C. Cooper. He expressed that M. buckland's paper left the impression that a system that supplies more information on each item is self-evidently a better system. "This surely cannot be correct: what we need is just that critical amount of information that is required to make sensible choices." In his reply, the author recalled that historically, information retrieval systems had been designed to retrieve everything with some specified attribute - a fact that can be useful in limited circumstances, such as patent searching. However, it would be better to design systems so that they would normally select the references best matching the user's preference - with an option for requesting more, if need be.

In concluding this part of the discussion J. Kendrew found the idea of restricting the language in documents difficult to accept. "Documents - all documents - are literature; they may be bad or good, but they are literature. Is it not an arrogance to suggest rewriting them? Shakespeare? James Joyce?" It is the job of the information scientists, according to J. Kendrew, to cope with the infinite variety of language as it is, rather than to make life easier for themselves by restricting and so destroying language.

Moving to the paper on "Information Retrieval: Theory, Experiment, and Operational Systems," W. Rouse referred to Thomas Allen's work mentioned therein and commented that traditional information retrieval systems cannot meet users' desires if all they do is provide large numbers of pointers to possible answers. While the information retrieval system need not provide the answer, "more value added is needed - a few good pointers. What do you think?" In his reply, S. Robertson agreed, but underlined that pointers, as opposed to answers, are a valuable function for information retrieval systems, though perhaps Allen's work might suggest pointers to people rather than documents.

N. Streitz then took the floor. He said that while S. Robertson reported on new and innovative ways of providing better access to information retrieval systems, commercial, large-scale systems do not essentially change, staying with their "not very user-friendly interfaces, requiring knowledge about Boolean queries." He asked whether there was any reasonable explanation why this was still the case. Why do they not make use of graphical user interfaces or new retrieval techniques? S. Robertson replied that there were several explanations, though not necessarily good ones. One was that the major host systems use very old software, modified but not fundamentally changed over 20 years. A second reason was that operational systems designers were frightened of the possible computational requirements of associative methods - although some associative methods can in fact be implemented in computationally undemanding ways. Nevertheless, said S. Robertson, "there are now signs of change."

Making a general comment, I. Wesley-Tanaskovic referred to the need for a general concept of information technologies enabling improved or changed conduct of science and engineering, such as post-processing in "front-end" systems by enhancing the delivery of messages or communicating knowledge. S. Robertson agreed and remarked that we have a vast amount of stored human knowledge, in various forms, a very small proportion of which is in machine-readable form. "We have to work with our history," he remarked; "we are not in a position to start from scratch."

In concluding the discussion of his paper, S. Robertson made a general comment on databases. He said that full-text databases, such as press cuttings, have many of the characteristics of bibliographic databases, except that they contain the "information" rather than a pointer to it. Reference works also exist as databases. One challenge is to combine text-retrieval facilities with those associated with relational databases.

D. Lide then addressed the third paper of the session on "Computerized Front-ends in Retrieval Systems." He thought some projects that had developed front-ends to lead the user to the most appropriate database for answering his query have encountered resistance from database vendors, who feel that this might put them at a competitive disadvantage. He wanted to know if this was still a problem. L. Smith replied that there was an example in the front-end EasyNet system of a solution that may be acceptable to the database producer but not to the user. In this particular front-end, database selection involves having the user go through a series of menus, specifying the subject areas and type of material of interest. However, very often more than one database may satisfy this specification. In these cases, the front-end was programmed to divide usage among the different databases, choosing each base with equal frequency but without informing the user that there was more than one possibility in the particular search under consideration. Thus, a user might be led to a better or poorer database, depending on the day the system was used - a situation that satisfies the database producers but cannot be adequate from a user's perspective.

M. Takahashi then asked two questions. First, he wanted to know if there were front-ends that include post-processing capabilities to process the documents drawn from databases. L. Smith explained that some front-ends provide capabilities to build local databases with downloaded records. She added that there was experimentation, as described in papers by N. Dusoulier and M. Buckland, to analyse output using certain attributes, such as year or language of publication. She concluded by saying that "we need to learn more about what users would like to do with text or bibliographic records to further develop this capability of front-ends."

The second question of M. Takahashi concerned non-professional people. He wanted to know what kind of front-end systems were good for them. L. Smith said that it was easiest to design front-ends in support of access to particular databases for particular tasks. "This will be easier for a user of such a front-end since they are working in a particular context."

The discussion on front-ends was concluded with a comment from W. Rouse, who believed that the front-end can be more friendly than the back-end if we separate traditional searching and downloading from specialized and tailored searching of local, smaller databases. "While downloaded databases cannot be any better than the source database, their dramatically reduced size can allow much more sophisticated and friendly processing."

Regarding multimedia technology, W. Rouse asked if it had been found that colour, graphics, animation, video, etc., may be more useful for selling a system and training in initial use, but less so once someone became a "power" user. J. Alty in his reply admitted that "fancy tricks" do help to sell an initial system. It is also true that when new techniques or facilities are provided, designers usually tend to "over use" them, as was the case with the initial use of colour. He said there was an urgent need for a multimedia methodology, because the number of combinations of media is huge. Since media can often assist in transmitting "difficult" knowledge, they are particularly useful for new users or those in need of training. It has been found in experiments that more complex media do not always help. Texts may be found useful in many situations whereas in others graphics, colour, and sound may be crucial. The methodology ought to help us identify the situations where multimedia are useful.

N. Streitz next took the floor to make two comments. The first on the figure in J. Alty's paper plotting "usability" against "knowledge and experience of the problem area." According to N. Streitz, another dimension was needed: the task. J. Alty agreed, but said, however, that the diagram was not originally his; it was from Marmollin and could not be changed. The other question was about the "comparison of texts." N. Streitz felt that there were better ways to do this, such as DIFF programs. J. Alty replied that the example he gave was meant to make a point for multimedia - how, by using properties of the human cognition system, we can significantly simplify a task. If the task becomes more complex, such as paragraphs getting out of order, then DIFF may not work well. He said that they were hoping to set up a system where one document is read over whilst the other is visually followed. There will also be visual aids to help with paragraphs out of sequence. At that point, "we will campare it with DIFF," concluded J. Alty.

Contents - Previous - Next