Research and development of computer systems have always been a dissension between vision and reality. After development has successfully implemented a working system, nobody remembers the roots. The vision and dreams that led to the product fade away. Years later the original outline might serve as a realistic specification for a new product. At least it might be fruitful to compare the plan with what has been achieved. As technology evolves over time it might have finally become possible to implement the product without restraining the original idea. But the current technology dominates the market, and the old ideas seem to be outdated.
This thesis contrasts the vision of hypertext with the prevailing situation of the World Wide Web. Many compelling aspects of hypertext systems of the last 35 years are not present in the current implementation of the Web, although they should be considered in order to understand where the Web should be improved. A discourse on the history of selected hypertext systems will point out such features and ideas, that are far from being dated.
The same approach will be taken for the field of graphical user interfaces, where graphical user interfaces like Apple Macintosh and Microsoft Windows dominate the scene. It is wise not to mix up the term graphical user interface (GUI) with WIMP interfaces, where WIMP stands for windows, icons, menus, and pointing device. WIMP interfaces are just one possible set of alternatives for graphical user interfaces. They often come along with the desktop metaphor, that has been developed in the 1970s at Xerox PARC. Other characteristic graphical user interfaces are indeed possible, even though they might be hard to imagine directly.
Each interface, that is not based on windows is a good candidate for a non-WIMP interface. Ms-Dos of course is not a WIMP interface, but it is not a graphical user interface either. MetaCreations’ Bryce, a 3D landscape editor, is an example for a GUI different from the classical WIMP world. And the entire field of information visualization explores new graphical concepts to convey deep structure of data.
What might be the reason to discuss hypertext and GUIs together? First, the history of research and development is already overlapping. And second, it is time to reconcile these two branches of information media that vie against each other on our PC screens.
All pioneers in the field of hypertext and GUI strive for an intensive and interactive dialogue between human and machine. Over 40 years ago, Joseph Licklider called it Man-Computer Symbiosis [Licklider 60]. A few years later Doug Engelbart titled his lifelong research topic: A Conceptual Framework for the Augmentation of Man’s Intellect [Engelbart 63]. The mouse is invented by Engelbart and Bill English in 1963. Interactive text editing, hyperlinking, computer-supported cooperated work (CSCW), video conferencing, among other technologies, are first developed for the system NLS on a time-shared mainframe computer at Stanford Research Institute (SRI).
In the 1970s Ted Nelson reminisces the famous article of Vannevar Bush As We May Think [Bush 45] as he entitles an essay As We Will Think [Nelson 72]. Computer Lib / Dream Machines [Nelson 74] and Literary Machines [Nelson 93] – the latter first published in 1981 – depict Nelson’s vision of worldwide hypertext, a universe of literature and personal writings where «everything is deeply intertwingled» [Nelson 74, p. dm 45].
The concept of personal computing is formulated by Alan Kay. In his doctoral thesis The Reactive Engine [Kay 69] he postulates three principles for computer systems to be used successfully in human-computer interaction. He writes [Ibid., p. 9]:
- The communications device must be as available (in every way) as a slide rule.
- The service must not be esoteric to use. (It must be learnable in private.)
- The transactions must inspire confidence. (“Kindness” should be an integral part.)
This quote shows clearly how Alan Kay anticipated the field of user interface design back in 1969. At Xerox Parc he worked on Smalltalk, and developed windows and menus for the graphical user interface as we know them today. David Canfield Smith and Larry Tesler, also at Parc, are mainly responsible for the idea of consistency and modelessness in interface design. A small generic set of operations should be sufficient to interact with the computer. Together with the conception of a physical office this makes it less complicated for the average user to remember all the commands. The metaphor helps the user to build up a mental model of the system. As long as interaction is coherent with the model interaction can happen with ease.
It only depends on the point of view whether any of those pioneers belongs more to the discipline of hypertext – in the sense of dealing with interrelated text – or to the field of user interfaces. The common vision is to enhance the possibilities for humans to cogitate about the world. They want to build tools for people who think.
The second reason to discuss hypertext and graphical user interfaces side by side is the problematic situation that we are facing on our screens since the advent of the World Wide Web. A browser window opens on the WIMP-desktop and unfolds the worldwide space of information. Inconsistency crawls in as the rules inside the browser window follow a totally different scheme of interaction techniques as the ones that apply outside the browser window. In fact a new non-WIMP graphical user interface proliferates in the middle of the desktop environment. The Web interface neither uses the desktop metaphor, nor other WIMP ingredients as its main principles. The window of the browser application just frames the Web-site on the desktop environment and can therefore be neglected as a constitutional property. Menus are also playing a minor role, because they are not part of the standard repertoire of interface elements. And icons, respectively images, are more or less just substitutes for textual hyperlinks.
Consistency is a high value in personal computing. Consistency leads to growing confidence in the tools. The user can predict the effect of a click on an icon – it gets selected. She can rely on the effect of a double click on a document icon – the document opens and displays the content in a window. The document can be edited until the window is closed. A window corresponds to a sheet of paper in the real world in such a way that it belongs to exactly one document.
Such basic WIMP interaction rules do not apply for the Web anylonger. A simple single click is sufficient to trigger a link. What follows is that the current document disappears and the content of another document is displayed at the very same place. This is the idea of browsing but it does not match the mental model that has been established with the desktop metaphor.
This is just one example to illustrate that the interference between WIMP interfaces and Web interfaces causes an awful and inelegant environment overall.
Analyzing the current situation is the first step to remove the obstacles in order to come to an interface that suits better the needs of conveying local content like personal documents as well as distant information on Web servers all over the world. Looking back into the history of modern interfaces should reveal the original ideas and principles. Only if we reconsider the visions and all the influences that have shaped these visions into existing products we can decide how to continue the development of interface technology.
The present thesis is divided into three main chapters entitled Hypertext, Graphical User Interfaces, and Beyond the Desktop. The chapters on hypertext and graphical user interfaces share the same structure. They begin with an overview of the development to provide the historical context for the presented systems. Following is a detailed discussion of the systems one by one. The focus lays on peculiarities from the perspective of the present hypertext system called World Wide Web, respectively the perspective of the WIMP desktop model of Apple Macintosh and Microsoft Windows. Before these concepts are summarized, some general theoretical approaches will be presented to obtain a common terminology. The chapter on graphical user interfaces will additionally provide some aspects of cognitive science. The closing sections Provisions for the Future of the World Wide Web and Provisions for the Future of the Desktop Model consider the potential impact of the historical visions on the current systems.
The chapter Beyond the Desktop unveils the inconsistencies between the desktop environment and the World Wide Web with respect to human-computer interaction and the utilized metaphors. The closing section Provisions for the Future argues for a fundamental approach to combine both worlds with each other. The deficiencies of each domain can be tackled by means of know-how and experience of the other side.