This report is also available as an Acrobat file.
The Design of Virtual Environments with particular reference to VRML
Design faults in virtual environments
In addition to those new problems associated with virtuality, we may expect to transfer some of the problems we have with 3D environments in the physical world to the computer domain. Some problems arising from poorly designed environments and objects include:
It is assumed that virtual environments will be interactive in the full sense, so that the users' interaction is not confined to altering viewpoint. VRML is changing to accommodate such needs.
- inadequate identification and signage
- lack of affordances
- poor use of mappings
- inadequate feedback
Inadequate identification and signage
As in large buildings and cities, poorly designed Virtual Environments will have inadequate identifying features and insufficient navigational cues (Passini 1992).
Lack of affordances
Affordances are those traits of a designed artefact that tell us how to use it (based on Gibson's ideas about visual perception (1979) and used by Norman (1988) and others). A well designed handle will look like an object that can be turned, while on the other hand the multiplicity of identical buttons on a consumer electronics device normally give no clue - other than their labelling to their purpose or mode of operation. It would be sad if we built a virtual VCR as appalling as most of the actual VCRs on the market (Thimbleby 1991). There is some reason to believe that the visual uniformity and resultant incomprehensibility of electronic devices arises from the freedom from physical constraints in their construction. If building a steam-engine out of brass and steel, there are many limitations of physics on what can be designed; but if building a music amplifier, the physical components (being small and not easily identified by the lay person) cannot either 'explain themselves' nor dictate the form of the assembled article - another black rectangular box is produced. The relevance of this to virtual objects of course is, that an even greater disconnection is possible for a virtual artefact between its form and its behaviour or mode of use. The designer of virtual objects has the freedom to make even worse objects than those commonly available in the real world. However by applying a proper understanding of human perception and behaviour, and transferring to virtuality the best of design principles from the familiar world, the designer also has the freedom to exceed the usability of real objects. This is design considered as a gift to the user, rather than as an opportunity for the designer's self-indulgence.
Poor use of mappings
Other aspects of physicality have influenced the form of real-world artefacts. The form of objects and constructed environments is not only determined by the nature of their materials but by the fact that they will interface with human beings. A door-knob is the right size and shape to fit the hand. If the mapping of virtual objects to the user were well designed, then the forms designed for use immersively with a dataglove would be dramatically different from those designed for desktop use with a mouse-pointer.
The objects with which we interact must be capable of registering our actions. Feedback should be clear, instant and appropriate. GUIs have made considerable strides in providing kinds of feedback, including:
- continuous representation of the object of interest a defining characteristic of Direct Manipulation such as the file being moved across the desktop
- input-acknowledge indicators, such as buttons which highlight when the pointer intersects them
- progress indicators, such as a rotating cursor to indicate a delay while processing takes place
- error messages of many kinds
In current VRML browsers, it is easy to be unsure whether a delay is, for example, caused by a failure of user-input such as a mouse-click or by processing delays. Just as the language of feedback in current GUIs has evolved over a period of more than ten years, it can be assumed that it will take a while for an appropriate range of feedback styles to mature in Virtual Environments: another difficult but stimulating task facing the designers of Virtual Realities.
What you see is ALL you get
The history of desktop computing is sometimes taken to demonstrate that GUIs are superior in every respect to CLIs. A single example of poor design in many GUIs will demonstrate that this is not so, and point to problems which we must avoid in Virtual Environments. Using a CLI, we can interact with objects regardless of whether they are currently represented on screen (for example
copy a:myfile.txt b: may be used even when neither file's name was previously on the display). By extension, we can also easily set up batch processes to work on a set of objects over a period, to which we need not attend. But in some GUIs, we can interact only with those objects currently in view, and we must manually undertake each interaction. The significance of this is that it is symptomatic of the excessive concretising which we have warned against elsewhere in this report: it is an unfortunate corollary of elevating the concrete at the expense of the symbolic. The fact that in a Virtual Environment we can interact with Virtual Objects by direct manipulation, should not mean that we must do so.
More prerequisites for successful virtuality
What are the preconditions for successful virtuality? What prompts the sense of presence, of directness, of engagement? Part of the point of VR systems is their very sense of 'realness' (which we would argue is not dependent on mimicry of real world referents), and with sufficient computational power and the design of effective feedback already mentioned, we can also hope for engagement to arise from the way in which objects in the scene will react to the user's actions. With current technology, a responsive 2D system may be more satisfying than an unresponsive 3D one.
In addition to these obvious requirements, there may be more subtle psychological characteristics which would assist. The ideas of Csikszentmihalyi (1988, Bulmer 1995) offer an additional insight into what we might mean by successful virtuality. He has tried over an extended period to gather evidence of what it means for a person to 'lose themselves' in an activity, to achieve a state of experience which he has named 'Flow', whose characteristics include:
One of his useful observations is that this kind of super-engagement arises intrinsically from the activity. While an activity may be begun for extrinsic reasons, it later becomes self-motivating. This seems to apply equally to work activities and recreation.
- a sense of complete engagement
- alteration of self-perception (both less and more perception of self)
- increased perception of external systems and devices as extensions of self
Csikszentmihalyi is less useful when it comes to helping us state what might be the preconditions for Flow. He suggests that the difficulty of the task in hand should be maintained between the boringly easy and the frustratingly difficult, hardly providing new insights unknown to the Higher Education community! His populist writing is aimed less at the designers of systems than with the attitude of the participant. Nevertheless, his work may be worth study for anyone trying to devise inherently absorbing systems. We must always remind ourselves that the motivation which users may experience when first they encounter Virtual Environments will be transient, assuming that VR will become a commonplace mode of interaction.
Academics, and especially the creators of courseware, sometimes look enviously on the engagement which computer games bring to users, and games are certainly among the first widespread uses of 3D interaction. There seems little doubt that the provision of intriguing, puzzling, often confrontational environments assists in engaging the user. However, we must remember that an important contribution to the appeal of games is their very lack of content and that they are elective - people are not obliged to play a computer game in the way that they are obliged to take a particular course module. Some of the attraction of computer games is therefore nothing to do with their design, and Education can never by its nature achieve the same style of devotion.
Virtual Environments Visualisation