Hypermedia as Integration: Recollections, Reflections and Exhortations

Keynote Address, Hypertext '96 Conference
Washington, DC, March 20, 1996

Randall H. Trigg

Xerox Palo Alto Research Center
3333 Coyote Hill Road
Palo Alto, CA 94306

Copyright © 1996 Xerox Corporation

These pages contain the text of my keynote as well as the slides, converted from powerpoint to jpeg. There are two ways to review the presentation: (1) scroll through the text following links to the slides as desired, or (2) step through the slides (the relevant portion of the text is included with each slide).

Finally, I've included on this page an extended abstract which appeared in SIGOIS Bulletin in the summer of 1996.

Extended abstract

Hypermedia and its earlier form, hypertext, refer to the notion of information structured as linked networks, as well as to the building, storing, navigating, and searching of such structures. The talk starts with the observation that integration has always been a goal of hypermedia developers. At first, this took the form of large monolithic link-enhanced environments that included support for text editing, electronic mail, CSCW, software engineering, and project management among other applications. Later, the "open systems" movement arrived and the meaning of hypermedia integration shifted to connecting existing applications. With the emergence of the World-Wide Web, the world's first broad-based hypermedia system, the dream of integration now extends across sites, networks, media formats, and hardware platforms. Using a current example (extending the Netscape browser to support email management), I argue that this history is not as linear as it seems, in fact the old monolithic agendas seem still to be with us.

In the "Recollections" part of the talk, I discuss several senses of integration that have arisen in my hypermedia system development work over the last 15 years. TextNet, the system I developed during my thesis work, attempted to integrate pure network-based hypermedia ala Ted Nelson's Xanadu with the sorts of hierarchical structuring supported by Doug Engelbart's Augment system. It accomplished this by distinguishing three functions of linking: traversal, structuring, and argument representation.

In Xerox's NoteCards system (with Frank Halasz, Peggy Irish, Cathy Marshall, Melissa Monty, and Tom Moran, among others), we learned a great deal about integration from our own users. One Ph.D. student's thesis "notefile" (i.e. hypertext) illustrated the way that multiple evolving versions of the hypertext ("chronological" integration) needed to coexist over extended periods of time in order to adequately support his work. Another example involved supporting different "styles" of hypertext creation, from writing single large "document nodes" later broken up and linked, to "brainstorming" linked structures to which content was later added. Integrating these styles meant offering users easy movement between them.

The VideoNoter system (with Jeremy Roschelle) supported the integration of video in a new way. For the first time, we believe, links could be created out of video. Playing a video sequence in the system triggered links anchored in particular frames. The links' destinations were illustrative graphics or verbal transcripts which were displayed on the screen for a duration (some number of frames) dictated by the link's source anchor. Our users, largely social scientists analyzing videotapes, were pleased with this form of multimedia integration. But in the end, the system failed for lack of a different form of integration. Our users needed VideoNoter objects to link to Microsoft Word documents, rather than the emacs-like text editor supported by the Allegro Lisp environment in which VideoNoter ran.

Starting in 1990, in Aarhus, Denmark, Kaj Grønbæk and I addressed the "open systems" issue of third-party application integration directly. (We weren't the only ones; significant strides were also taken by the Microcosm team in Southampton, England, among others.) Our DHM system depended on a framework derived from the Dexter hypermedia reference model which we had found particularly well-suited to our open systems agenda. These days, the integration of systems like Microcosm and DHM with the global hypermedia of the WWW is a focus of ongoing research.

In the "Reflections" part of the talk, I revisit Frank Halasz's landmark keynote from the Hypertext Œ91 conference. There he laid out what he felt were the key issues for the future of the field. These included the problem of "Very large hypertexts," or why we had yet to achieve large-scale hypermedia. Frank's answer is especially fascinating given that the WWW had not yet "arrived." Looking back, we can consider each of the barriers Frank identified, and ask whether they were in fact overcome. I believe that most of them were not addressed by the WWW, and that we must look elsewhere for explanations of its success.

I also revisit Frank's call to move beyond the "Tyranny of links." Here I argue that the web has shown us that the problem is not the number of links, but the absence of structure. Moreover, the new structure just now coming to the web (for example, via SGML or Hyper-G) will need to be carefully integrated with browser-based navigation. Recognizing that different places on the web will support different degrees of structure, true integration will require web browsers to shift transparently from the display of highly structured to weakly or non-structured web pages.

In the "Exhortations" part of the talk, I offer recommendations to the field on directions we should be investigating. First, we need to look hard at our own history to avoid reinventing old wheels as well as repeating old errors.

Second, we need to look to neighboring fields like the emerging "Digital Libraries" for inspiration. In particular the work of David Levy and Daniel Pitti on cataloging and archiving is rich with implications (and running examples) of how structure could begin to play a significant role in web-based hypermedia.

Finally, I suggest that hypermedia has an important integrative contribution to make to activities like collaboration, mobilization and volunteering that comprise the under-valued work of community-building taking place on the web. I use an example of a low-income housing project in North Carolina slated for destruction by the local housing authority. In their struggles to gain a say in the design of their future abode, a task force of women residents were able to enlist the support and active engagement of several architects on the internet. The women's community building efforts involved both email (first to usenet lists and subsequently directly to the architects) and hypermedia in the form of an evolving collection of web pages, including pictures of the buildings and the residents as well as notes from meetings with the housing authority. I suggest that integrating such diverse approaches to community building is a worthy topic for hypermedia researchers wishing to move beyond purely commercial applications of our work.

Last modified: Mon Sep 23 13:58:53 1996
Randy Trigg trigg@workpractice.com