How faithfully should the electronic office simulate real one?

The naïve answer to this question --- "as closely as possible" --- is wrong for several reasons. First, it fails to recognize that user dexterity in manipulating simulated objects on a computer screen is not as high as it is in the physical world. Second, it wrongly subordinates users' goals to the actions that they perform in the physical office to achieve those goals. Third, it often results in unnecessary development effort and poor system performance. Fourth, it limits the functionality of the system to that provided by its physical counterpart. Finally, it presents a user interface that, though familiar, may not be optimal. There is no simple rule for determining the correct level of detail that a Desktop Metaphor system should have. Such a determination requires either a careful analysis of the desired functionality (taking into account pragmatic considerations such as the difficulty of implementation) or considerable experience in designing such systems. Many developers lack such experience and are unwilling to do the necessary analysis, and thus wind up applying the naïve approach. Some heuristics are presented to help guide designers away from the naïve approach.