Being Explicit about Implications - Subliminal cues
All user interfaces make implications about the way they should be used. The position of a switch can indicate how severe the consequences of using it are. Placing it in an awkward position implies it should not be used unless absolutely required. Placing it in easy reach suggests it can be used any time.
It is well known that the expression on someone’s face when they tell you something makes a vast difference to your reaction. While they may explicitly be saying one thing, the facial expression may say something entirely different. In cases where you don’t know the person, and what they are saying explicitly and what their face says contradicts there is an uncomfortable feeling of unease, because you are getting conflicting signals and don’t know which to trust. So it is with user interfaces.
So when designing a user interface it is important to try to understand the implications you are making, and to make sure these are consistent with the actual message you wish to put across. A very crude example is the button bar in a word processor like Word. Because these are highly accessible, and therefore more prone to accidental use there should be undo capability on all of these functions. The very accessibility implies to the user that these things are safe to use. Dangerous tools are kept in a more secure location in a toolshed, the same applies to virtual tools.
Another example is the close window button. In most applications closing a document window throws away the document. You may be warned that this will happen, and you get a chance to save, but ultimately the document will be discarded. MacOS control panels though, when closed, do record the changes made. They interpret closing the panel as “all done”, and by default keep any changes done. This has two undesirable effects. One, it makes it much harder to cancel a change, and two, it leaves the user in an uncomfortable position since they are not quite certain that their changes were committed. It would be better to have an “OK”, or “Done” button instead, which also closed the dialog, since the close button could then be used in the normal way, losing the changes. A cancel button would also be good, since that makes things even clearer, and on some systems modal dialogs have both these and no close window button, which makes the implications very clear.
Being explicit about implications essentially means that when a UI is designed care must be taken to consider what is being implied, and to make this tally with the explicit message. The ability to see implications, even if only at some emotional level, is one of the skills that separates a good designer from a poor one. A good designer may get a feeling that a design is good or bad based on those subliminal cues. Good designers are also probably more consciously aware of cues than others, but this does not mean that others are not aware at all.
Attempting to make implications consistent with the explicit interface is clearly a good goal. The question is how can it be achieved. One problem with this is that implications are often predicated on a users previous experience, and not all users have the same experience. In user interfaces which are not metaphorical the implications are derived solely from past experience with that and similar UIs. This suggests that making non metaphorical UIs should, in some ways, be simpler since there are fewer unplanned implications derived from the real world. Where user interfaces are metaphorical implications from the real world are important, and this can make design much more complex since it means that the cues required can change from culture to culture, and person to person. This either means that care has to be taken to remove such confusing cues, or to have multiple interfaces for the various user communities. This highlights the importance of understanding the users.
Having argued the case against using implications, I should redress the balance. Subtle cues, be they visual, audible or whatever, are useful. The mouse pointer is a classic example. Once the user is used to the pointer changing shape it is a very powerful hint that there is an active region. Cues like this should be used where appropriate, but used with delicacy. Overloading a cue becomes confusing and the implication looses it’s value. Other implications are colour, green and red are typically meant to suggest safe and dangerous operations respectively (although colour may be culture sensitive).
In conclusion, since human beings observe their environment at many levels, from conscious awareness through to subliminal hints, it is useful to present interfaces which trigger each level of awareness. However, this is only useful if the the messages all concur. Taking time to ensure this is a worthwhile part of the design process.