Direct Manipulation

Examples of Direct Manipulation Usage

Characteristics of Direct Manipulation

Benefits of Direct Manipulation Interfaces

Potential Problems with Direct Manipulation Interfaces

Theoretical Approach to Direct Manipulation

The Future of Direct Manipulation Interfaces

Introduction to Direct Manipulation

The guiding principle behind direct manipulation (DM) is as follows: the user should be able to engage in an interface dialogue which mimics those actions which would be required to perform a task if it required the manipulation of physical entities. Actions are carried out on interface objects themselves, as opposed to an intermediate command language. Typically, DM systems have icons representing objects, which can be moved about the screen environment and manipulated by controlling a cursor with a mouse.The possibilities of direct manipulation can be seen by observing the successful programs for which it has provided the underpinnings.

Examples of Direct Manipulation Usage

The Apple Macintosh was the first highly successful commercial computer system which included direct manipulation features. This system used the metaphor of the desktop, in which icons represented common desktop objects (e.g. files, trashcans). Following Apple, many CAD systems for engineering began to use the direct manipulation principles. As of today, some of the most successful software and computer products have been based on direct manipulation ideas. For example, text editors (Word, Wordstar), spreadsheets (VisiCalc, Lotus 1-2-3, Excel) and computer operating systems (Windows) have become the standard in world-wide computer use. And, most notably, video games now provide examples of well-engineered and commercially successful DM systems. As this concept has evolved, specific common characteristics have successfully carried over from one interface design to another.

Characteristics of Direct Manipulation

The evolution of successful direct manipulation interfaces has been a trial-and-error process. It became clear that not all interfaces would be best suited for a direct manipulation interface. Likewise, not all possible direct manipulation features are equally efficient acoss different types of interfaces. As of today, there are six characteristics which exemplify well-designed direct manipulation interfaces.

• Explicit and visible action: the user can point and manipulate objects, making it explicit in terms of what is being operated on.
• Immediate feedback and modeless interaction: the effects of object mainpulation should become immediately apparent to the user. For example, in text editors with separate "insert" and "edit" modes, errors commonly occur due to failures to change the mode when the activity changes.
• Incremental effects: Actions should parallel the movements of the users, so that the user can break a complex task into smaller components which have a summative effect.
• Intuitive interaction: the interface must match the user's model of the task. In general, direct manipulation interfaces require the simulation of real world tasks. The aim of this is to enable operation of an interface without prior learning.
• Learning by onion peeling: this describes the idea of gradual revelation of system complexity. The user should explore the more complex parts of an interface only when their knowledge is sufficient for them to do so.
• Pre-validation: only actions that are valid for the current task and state of the system should be possible; for example, menus and dialogue items should be greyed out to signify their appropriateness.

Benefits of Direct Manipulation Interfaces

Not only have such DM interfaces shown to be commercially successful, they tend to engender enthusiasm and great praise from users. This energetic response can be explained by the following list of benefits.

• Novices can learn basic functionality quickly, usually through a demonstration by a more experienced user.
• Experts can work rapidly to carry out a wide range of tasks, even defining new functions and features.
• Knowledgeable intermittent users can retain operational concepts.
• Error messages are rarely needed.
• Users can immediately see if their actions are furthering their goals, and, if not, they can simply change the direction of their activity.
• Users experience less anxiety because the system is comprehensible and because actions are so easily reversible.
• Users gain confidence and mastery because they are the initiators of action; they feel in control, and the system responses are predictable.

Potential Problems with Direct Manipulation Interfaces

One obvious problem with direct manipulation is that some tasks cannot be described with concrete objects and not all possible actions can be performed in a direct manner. With the Apple Macintosh, designers had to deal with conceptualizing a buffer (a place where information could be temporarily stored). The solution involved utilizing a hidden "clipboard" where the information would be temporarily displayed. Initially, this concept showed to be problematic for novice users. This is just one of many possible types of problems associated with DM interfaces. Here are some other common problems:

• Repetitive tasks: while incremental effects break complex tasks into manageable components, users may have to repeat the same actions again and again when the task is repetitive. A well-designed system needs a macro or scripting mechanism to handle repeated manipulations.
• The concept of variables: distinguishing individual objects from a class of objects: direct manipulation requires that objects are tangible entities, yet this does not fit well with the concept of a variable as a slot that can represent other objects.
• Pointer limitations: the pointer system is not necessarily accurate enough or sufficiently controllable for some tasks which require accuracy. The advantage of issuing commands is that the user can give explicit coordinates.
• Mapping knowledge and skills: direct manipulation requires a close mapping between existing knowledge and interface skill. This might restrict interface design to tasks that users already do, and prevent the design of new tasks. A corollary of this is that users may restrict their use of interfaces to highly familiar routines. Learning by onion peeling may restrict users to simple or obvious features of the interface.
• Transferring task difficulty to the user: since direct manipulation requres actions to be carried out explicitly, there is a danger that task difficulty can be transferred directly to the user. For, example, visual programming may enable programmers to link objects together by making pointer selections, but it does not support the task of planning programs. Tasks such as programming require some degree of abstract reasoning, whereas direct manipulation is about concrete action. There is a danger that direct manipulation may discourage the planning necessary for complex tasks.

Thus, clearly DM technology is not the end-all solution for all interface designs. A definite trade-off exists between issues such as expert/novice, speed/accuracy, efficiency/thoroughness, etc. In order to better understand all the trade-offs involved, attention has switched to a more conceptual level. In the 1980s, an attempt was made to put DM within a theoretical framework.

Theoretical Approach to Direct Manipulation

A noteable attempt has been made to evaluate the idea of "directness". In this thoery, directness is characterized in terms of two opposing gulfs which indicate the existing gap between the goals of the user and the system image. The extent of directness in DM is measured by the extent of the discrepancy between these two gulfs. These two gulfs are termed the gulf of execution and the gulf of evaluation. The distance between these gulfs represents the mismatch which occurs between the way in which an individual thinks about a task and the way in which it is represented in the system.

The gulf of execution is defined as the distance between the goals of the user and the means of achieving those goals through the system.

The gulf of evaluation is defined as the distance between the behavior of the system and the goals of the user.

These gulfs demonstrate the bidirectional relationship between the user and the system. In order to impove an interface, the discrepancy between the goals of the user and the physical state of the system must be reduced. If this occurs, the mismatch between the users' way of thinking and the representation of the system is reduced. As a result, it is easier for the user to complete her task. Two possible ways to do this include changing the users' goals and interpretation of the system versus changing the input/output components of the system image. The proponents of this theory suggest that the latter alternative is usually preferred.

For more information on this theoretical approach, see the chapter titled "Direct Manipulation Interfaces" in the book User-Centered Designs (1986, Hillsdale).

The Future of Direct Manipulation Interfaces: Hypertext, Multimedia and Virtual Reality

Given the great success of DM interfaces, it is probable that its implementation will continue throughout the evolution of computers. However, there are clearly some applications which might be extraordinarily enhanced if not exclusively propelled by DM technology. Three such applications are Hypertext, Multimedia and Virtual Reality

Hypertext

Hypertext can be defined as bodies of text-based information which are linked in a non-linear fashion (in opposition to paper-based text), and which allow the user to browse through such text in any desired order. In traditional "bookish" information sources, a user must search forwards and backwards through a series of chapters, files or pages in order to access the desired information. Using hypertext, the user is able to create any possible number of links between similar items of information, allowing her to explore concepts through direct access to similar information. This type of information organization is especially useful in content domains which tend to be organized via hierarchies. Examples of current hypertext applications include Hypercard stacks, library search systems and most of the world wide web.

With such a system, the user can decide what information to view at any particular stage of the search by simply pointing a cursor and clicking on the text or image which is relevant. For example, a hypertext system which represents information about universities would allow a user to gain information in increasing detail with a single click (e.g. GA Tech, College of Computing, Students, John Doe).

However, hypertext is not without problems. For example, when a user already knows exactly the "what" and "where" details of a piece of information, hypertext may prove to be cumbersome. Also, users of such systems commonly have trouble locating exactly where they are within a hypercard systee; feedback systems are necessary.

Multimedia

Multimedia is the application of hypertext to a mixture of computer-based text with pictures and sounds from video and audio-based sources. By giving the user simple access to large amounts of information, these systems might allow the user to engage in creative behavior which was not possible with linear, single-medium interfaces. Furthermore, some have argued that this format may be particularly applicable in learning environments as the user determines the specific route taken through the gathering of information. Examples of current multimedia interfaces include CD-ROM encyclopedias and the latest world wide web sites.

Virtual Reality

Virtual Reality takes multimedia one step further. This interface uses computer-generated graphics to simulate a virtual world for the user to enter, explore and interact with. These systems typically use very sophisticated input devices such as data gloves and head-mounted displays. Virtual reality is still being developed as a technology and its strengths and weaknesses have yet to be fully evaluated. Also, the high costs of input devices and hardware dictate that, for now, uses are very limited.

In conclusion, the future of computing may certainly be fueled by the ideas of direct manipulation. Advances in the speed of computer chips, parallel processing and hardware, are making more advanced and complex systems possible, increasing the possibilities with which DM features can interact.

General Sources for this Report

Eberts, R. E. (1994). User Interface Design. Englewood Cliff, NJ: Prentice Hall, .

Lansdale, M. W., & Ormerod, T. C. (1994). Understanding Interfaces: A Handbook of Human-Computer Dialogue. Academic Press: San Diego, CA.

Preece, J. (1994). Human-Computer Interaction. Addison-Wesleyas: Wokinghav, England.