Question 1

a) Infrequent experts will have some superficial similarities with novice user of a computer system. They will require initial guidance and additional help facilities in order to learn how to access basic functions. However, they are likely to be able to transfer expertise gained with other systems to support their use of this novel application. This will increase the robustness and accuracy of their response to any errors. However, skill transfer will only be successful if the interface is externally consistent with other applications that the expert has met before (i.e. external consistency supports skill transfer for expert users).

Frequent novices will superficially be similar to expert users. They will develop considerable skills within specific areas of an application. However, those skills will not be robust in the face of failure. Relatively simple error messages that they have not met before can have a devastating effect. Also, because the users have had little experience with other applications, it is important that the interface is internally consistent t ease their initial learning time with this system.

b) There are a number of dialogue styles that might potentially be recruited. These range from text-based command lines to full graphical user interfaces. The full costs and benefits of these different approaches should be analysed using a QOC diagram and it is likely that some combination of styles will provide the greatest support. for example, test based command lines offer extremely high rates of interaction for expert users. However, in the context of a bank, there can be a significant level of staff turnover and so it is important that there be some means of reducing the initial learning times for the interface. This might suggest the use of menus for novice users in conjunction with keyboard accelerators. However, the nature of the data to be supplied also suggests that forms might be introduced to provide information about the various fields to be filled. This, in turn, implies that the user must be able to rapidly navigate between those fields without their hands leaving the home position on the keyboard (Tab for moving between fields, escape to cancel etc).

c) Any system would have to undergo profound changes if it were to be extended from the bank staff to support more general groups of users. Some of these changes can be summarised as follows:

- terminology - bank employees are likely to be much more familiar with the terms that are used in a banking system than members of the public for instance, they may not even know the name of the particular account that they have (5 day saver etc)

- dialogue structure - members of the general public will have to be guided through each stage of their interaction with the system. Great care must be taken to ensure that they can recognise closure on a task and the consequences of partial task completion.

- data entry - members of the general public may lack both mouse and keyboard skills. A trade off may need to be made - by simplifying the available input devices you may also restrict the functionality f the interface.

- errors - the above factors may have a profound affect on the sorts of errors that the different groups of users will experience. Appropriate help messages will be different between the two user groups.

- help facilities- see previous comments both about errors and terminology.

Question 2

a) Java and AWT provide a number of benefits for the prototyping of a general purpose kiosk. One advantage is that the interface can be ported easily between a number of different platforms using the Java virtual machine. This provides flexibility in the choice of implementation platform to be used to power the system. A further benefit is that the information system can be ported between the kiosk platform and the web with little or no maintenance costs. It might even be possible to network the kiosk and drive t directly from a restricted set f web pages.

b) Java may not be a good choice of prototyping language because it can be a "heavy weight" solution in comparison to Director, Hypercard or even simple HTML. One consequence of this is that developers may be very reluctant to throw away an initial prototype as development progresses. A second issue is that the Java virtual machine imposes speed overheads as an interpreted language. This may not be a serious problem for a general purpose information server which has relatively modest computing requirements. A third problem is that Java virtual machines may NT exist for many of the non-standard devices that are employed by kiosks - this would have to be checked during initial requirements elicitation.

c) Many possible sketches might be provided to answer this question. They must, however, provide context information explain what the information is about and who is providing it. There must also be control information about how to issue a request and navigate within the system. There should be help functions.

The evaluation section must focus on appropriate formative techniques because the question makes it explicit that it is only a prototype. However, observational evaluations might be possible if the system were relatively complete. I would probably opt for co-operative techniques in conjunction with scenarios in the initial stages and then move to amore ethnographic approach as the development cycle continued. This situated technique is very appropriate given that it is difficult to predict all of the problems that might arise when they system goes live in a location which might not provide help if the users have problems.

Question 3

a) A mode arises when the same input sequence can have potentially different effects in two or more different context. For instance, a mouse drag may cause a circle to be drawn if the editor is in drawing mode. The same movement may have no effect if the editor is in selection mode. Designers have to be very careful when they introduce modes into a user interface because they are a potential source of considerable irritation the system may appear to behave in an unpredictable manner if they cannot understand the reason why their input is not having the effect that they predict.

b) It is very difficult to determine whether or not modes will cause particular problems because their impact is very different for different sorts of users. experts are likely to quickly identify modes because they will have seen similar effects in other interfaces - they can look at information in a tool tip or changes to a cursor to illustrate a mode transition. Novices may lack the necessary experience to find an appropriate explanation for the errors that they observe. Further problems arise because if you tell users that you are looking for problems to do with modes then they may modify their behaviour and be more careful in seeking explanations of the effects that their input produces. It can also be difficult to device tasks that force users into the areas of an interface that rely upon moded interaction.

c) Changes in mode must be indicated by more than one change in an interface if users are to notice. In drawing tools, the shape of the cursor usually changes from drawing, to selection, t rotation mode. This is because the users' attention is usually directed towards the drawing canvas and, in particular, to the mouse cursor that is used to perform the drawing activities. ToolTips may also be used to provide information about the effects of selecting a particular mode (e.g., "Draw a circle..."). This is useful because it provides a prompt for text that might be entered into search within a help facility. Other variations on this include a status box at the bottom f the screen to provide information about the current mode of the system. Audio feedback may be used to provide information when input has no effect (e.g. typing when a moded dialogue box requires that you press the confirm button).

Question 4

As this is an essay question there are many possible solutions. The key thing is to define what you mean by a principled approach to design. One interpretation might refer to the use of a standard set of interface guidelines, such as the Windows Guidelines or the Smith and Mosier work mentioned in the notes. Another interpretation might focus on the principles of psychology that focus on the interplay between cognition; physiology and perception.

Once an appropriate definition has been established the next stage is to focus on the importance of short term and long term effects. Visual appeal is clearly significant for consumer acceptance in the early stages of use - if it looks awful then people may be less inclined to download other related pages. However, the impact of this may fade over time. Users may return more on the basis of information content and layout than on simple visual appeal but this depends on the nature of the site (see below) Speed of retrieval may have more of an effect the longer that a user interacts with a set of web pages.

Later sections of an answer might go on to discuss task dependence of these observations. For example, visual appeal may be less of an issue than appropriate font selection if the user is accessing a mass of on-line technical information. This creates mixed approaches where initial pages for public access may be designed for visual impact and that subsequent information pages may be tailored more for legibility/information layout etc.

Finally, individual user differences and preferences may be considered.