Table of Contents→

Joseph Licklider’s vision was the symbiosis between man and machine. Early systems – like Sketchpad and NLS – have reached a level of intimacy with the user that is still remarkable. But the effort to learn the correct handling of those systems was a stumbling block for users. The five-finger chording keyset is a typical example. Once you have learned how to handle the device the usage can become habitual. It can be operated with ease and feels natural. But the chording keyset is not approachable for a novice user. It is most likely to make many errors in the beginning, because the keys are not labeled like the keys of a standard keyboard. Furthermore, the user has permanently to keep in mind in which mode of the five cases the keyset currently operates. This is an extra load for the memory.

The mouse is not an intuitive input device either. But the user becomes familiar with it very soon.** In Intuitive equals Familiar [Raskin 94b] Jef Raskin argues that ‘intuitive’ is often used inaccurately by non-HCI specialists. Very few systems are intuitive in the sense to be immediately usable without explanation and training. A better denotation of an intuitive interface would be the quality to use «readily transferred, existing skills» [Ibid.]. Alan Kay has shown up the relation between the mouse and Jerome Bruner’s notion of the enactive mentality. The mouse evolves to a seamless extension of the hand to interact with objects on the screen. The term direct manipulation has been coined in 1983 by Ben Shneiderman for this style of interacting with the computer (e.g. in Designing the User Interface [Shneiderman 98, p. 185]).

The targeted user plays a key role in interface design. Back when computer systems were used only by a few computer experts, not much attention was paid to well designed user interfaces. The user was required to adopt to the machine. Things changed with the development of the Xerox Star 8010 Information Systems. Now the intended user is an average office worker. The desktop metaphor was invented to cover the technical details of the computer in order to create an friendly and familiar working environment. The user’s mental model is based on everyday objects like paper, documents, and folders. This world of office metaphors helps the user to gain an practical understanding of the system.

Despite the promising approaches of NLS, Xerox Star, and Apple Lisa we are facing today computer systems that are complex and awkward to use – for office workers, as well as for scientists. All current PCs fall short compared to the original vision of personal computing.

The next three sections will discuss the relation between documents, application programs, and the filing system. The difference between a file system and a filing system can be best clarified with the following analogy: What a file is to the file system, is a document to the filing system. Both, document and filing system, belong to the user model. File and file system are technical implementations to physically store the content of documents. The critical comments on the filing system will be followed by a recapitulation of Xerox Star’s and Apple Lisa’s document-centered design approach.
This chapter will close with some reflections on metaphors and the chances of the medium computer.

3.4.1 Filing

The desktop metaphor allows to put documents on the desktop, and to file them in folders. Folders can also reside on the desktop, or they can be filed away in volumes. As the real screen estate is limited the desktop metaphor is extended to a hierarchical structure, i.e. folders can be stored inside of other folders, and so on. This approach worked quite well as long as the number of items was in the range of hundreds. But it does not scale to thousands or ten thousands of files. The hierarchy of folders was not invented for graphical user interfaces. It was adopted without reconsideration from the model of hierarchical directories of previous generations of computer systems. In a talk – given at the symposium Engelbart’s Unfinished Revolution in December 1998 – Ted Nelson argues with sarcasm [Nelson 99b, p. 4],

we’ve got hierarchical directories which we accidentally invented like in 1947 – ’where are we going to put all this stuff?’ ’well, lets make a file’ – they named it files, right. So, the hierarchical assumption has passed on to us which assumes that there is no overlap between things we do. You work on one thing, then you finish that, put it away neatly and then you work on something else. {{laughter}}. right. There is no overlap, there is no interpenetration, projects are never redefined, we don’t have to change our terminology once we’ve started…

The consequences of this decision penetrate the way how the desktop model works today. Users have problems to keeping track of their documents. They do not remember, for example, where they have saved the letter to a specific customer a couple of months ago. The only handle to a document is its file name and perhaps its position in the hierarchy of folders. Both criteria don’t scale very well. Furthermore, they are also weak with respect to the user’s real office experience.
File names are artificial and do not work the way one would expect. Jef Raskin describes the situation when file names are created in the following way [Raskin 2000, p. 118]:

File names are bothersome when you are about to save work, because you have to stop in the middle of your activity, which is trying to store your work away, and invent a file name. Creating names is an onerous task: You are required to invent, on the spot and in a few moments, a name that is unique, memorable, and within the naming conventions of the system you are using.

Names for folders and their position in the hierarchy are generated in similar fashion. What sounded reasonable in the moment of naming, is unintelligible when the item needs to be retrieved.

Means to identify files other than by name are underdeveloped in the present desktop model – although some promising approaches have been taken in the history of personal computing. Those include the employment of visual qualities. SDMS’ Dataland is founded on the idea of spatial arrangement of items. The remnants of this approach lead to the persistent arrangement of icons in WIMP interfaces like the Lisa Desktop Manager and the Macintosh Finder (cf. top left window in Fig. 3.13). Visual clues for documents are also rarely used. SDMS displayed facsimile – scaled down images of the original content. Opposed to that, Xerox’ desktop model introduced icons – abstract graphical images with a well defined semantics. The role of thumbnails as a substitute for icons should be further investigated.

Other options to identify documents can be based on content and context. «The content of a text file is its own best name», says Jef Raskin [Ibid., p. 119]. Tools to easily perform full-text searches on a given set of files can be an efficient way to retrieve files. Some key words are often sufficient to reduce the resulting set of possible documents to a practicable number of files.
Context calls for new interface elements. Documents are related to each other in terms of tasks. For instance the creation of this present document involves several hundred files: e-mails, html pages, downloaded articles in pdf format, audio tracks, and archives to backup previous versions – and finally the layouted text documents themselves. Books, magazines, and video tapes – although not in digital form – belong to the project as well. Representing this network of information with a hierarchical folder structure is not sufficient. New concepts are needed to express such relations.

Some new objects for the desktop are proposed by Bruce Tognazzini in Tog on Software Design [Tognazzini 96, p. 196]. Intelligent folders should continuously compile the set of documents that match certain criteria. Archives look and behave like standard document icons. But they can additionally store the historical chain of versions for a document. Piles are informal clusters of documents. Tognazzini explains in Scaling Information Access [Tognazzini 98],

You start a pile by dropping one document icon on another. You may then continue to add icons to your heart’s content. The resulting pile is easy to read: click on it and drag up and down to see thumbnails of each document instantly appear adjacent to the pile.

It should be easy to collect items that might be of interest for some vague goals. Finally, a Scrapbook is Tognazzini’s answer to keep track of the flock of documents for a task related project – like the one mentioned above.

The key to develop new filing concepts, that are closer to the user, is the abstraction of the hierarchical file system. The technical layer should be concealed from the user’s perspective, as other technical aspects have been covered by the desktop metaphor before.

3.4.2 Document-Centered Design

Xerox Star and Apple Lisa are the two main systems that follow a document-centered design approach. Jeff Johnson explains the role of programs for the Star in The Xerox Star: A Retrospective [Johnson et al. 89, p. 11]:

The applications included in the system were those that office professionals would supposedly need: documents, business graphics, tables, personal databases, and electronic mail. The set was fixed, always loaded, and automatically associated with data files, eliminating the need to obtain, install, and start the right application for a given task or data file. Users could focus on their work, oblivious to concepts like software, operating systems, applications, and programs.

Despite the fact that Apple has an open architecture to load new application programs, the user experience is very similar to the Xerox Star system.

Limitations of the hardware in the early history of Macintosh computer lead to the dominant application-centric user model, that we are facing today (cf. 3.1.9 Apple Macintosh). Nevertheless, the document-centered design approach bears several advantages for the user. The most important is the fusion of document file and document window. It is simply not possible to have an unsaved document without an associated file. New documents are created with the help of stationery pads in the filing system, opposed to a new command inside of application programs. The file name is generated by the system and can become subordinate. This is coherent with the previous section on the filing system, if other means to identify documents are employed.

In a document-centered software environment, documents are not related exclusively to one application program. The document is the central element. Programs are used as tools to contribute pieces of content to the document. One tool might be a text editor, while another tool provides imaging functionality. The unrestricted composition of all those parts constitutes the document. The operating system has to define a software interface between the tools and the documents. Furthermore, it is necessary to agree on open document formats.

Such an environment would be more flexible than the application-centric model. The user is free to access tools wherever they are needed. Documents would belong to the user, and can be edited with several tools. Monolithic application programs and proprietary document formats restrain the possibilities of personal computing.

3.4.3 User Illusion

Graphical user interfaces have the power to create visual and interactive environments for abstract data spaces. The original term – before ’physical-office metaphor’ or ’desktop metaphor’ became common – was «user illusion». The term was coined at Xerox Parc in the 1970s [Tognazzini 96, p. 291]. Metaphorical concepts are useful for users to built an initial understanding of the computer system. On the flip side they narrow the perspective to functions that have no counterpart in the real world. For example, drag and drop of a document to a folder moves the object, whilst a drag and drop to a printer icon triggers a printing job. But does it also move the document icon into the printer? The system’s behavior is not guarded by the metaphor anymore. Users are also confused if items are rearranged by the computer. They expect persistence of spatial properties. Therefore a concept like Bruce Tognazzini’s intelligent folders tests the limit of the desktop metaphor.
Other interaction techniques use no metaphors at all. Point & shoot, for instance, is a technique to create hyperlinks in Adobe GoLive. A line is pulled out to connect link marker and link target with each other. It is plain abstract graphics and has no metaphorical correspondence with the real world [Müller-Prove 99].

User illusion is a more sweeping idea. The user does not have to keep in mind whether an aspect of an object is in accord with the metaphor or not. Alan Kay says, that the term has «clear connotations to the stage, theatrics, and magic» [Kay 90, p. 199]. But it should be an understandable kind of magic. The plain representation of paper on computer screens falls short compared to the possibilities of the medium. Kay argues, «it is the magical part that is all important and that must be most strongly attended to in the user interface design» [Ibid.]. Ted Nelson emphasizes this idea. In The Future of Information he suggests to consider software design as a new form of movie [Nelson 97a, p. 16]

Movies are systems of events on a screen that affect the heart and mind of the viewer. Software–even office software– is a system of events on a screen that affect the heart and mind of the participant, and interact with the participant–who is no longer a mere viewer.
That means that software is exactly what movies are, and more. Software is not just a branch, but the generalization of movies, not metaphorically but literally.

Theater, movie, and magic also point to consistency and aesthetic integrity in human interface design. The value of consistency is a gain in familiarity and predictability of software environments.
According to the Macintosh Human Interface Guidelines, aesthetic integrity means that «information is well organized and consistent with principles of visual design» [Apple 95, p. 11]. A graceful appearance is desirable for graphical user interfaces.

Table of Contents→

For a free PDF version of Vision and Reality of Hypertext and Graphical User Interfaces (122 pages), send an e-mail to:
mprove@acm.org   I’ll usually respond within a day. [privacy policy]