Conditional Links and Structure Control

 Afterpaper for
The Messenger Morphs the Media 99,
Third Hypertext Writers' Workshop
By Jean-Hugues Réty







    First of all, I want to thank Deena Larsen and all those who helped to organize the workshop. I appreciated a lot this workshop, and especially the large interactivity in discussions.

    In my forepaper, I rose up the question of the writer's control over the structure of her hypertext: links are local constraints on the reader's browsing behavior but the writer wants to maintain a global control over the link structure. The workshop shown that this is a central, very important question: all widely discussed structure related issues such as orientation or closure depend on the ability of the writer to control the structure she creates.

The need for conditional links.

    First, the standard notion of static link deeply lacks of expressiveness. Let us consider for instance the problem of ending pages. The writer may want her hypertext - or some part of it - to converge on a single ending page. But how to do so? Think about this for a few seconds and you will see that we don't even know how to realize this fairly simple, basic, organizational pattern. Enforcing a property such as the existence of an ending page will end up in very poor, often almost deterministic link structures. Static linking is not expressive enough. Conditional links provide a solution to that problem (I don't mean that other possible solutions would not be worth investigating.). The idea is to extend what has been done for a while in adaptive hypertexts with a wider range of conditions. What conditions are useful for the fiction writer? I bet this will constitute a topic for the next workshop. Conditions on the historical list of previously visited pages yet appear as a necessity. We've just agreed, Robert Kendall and me, to cooperate in developing a library for providing the writer with good (we hope) tools for writing for the web. Support for conditinal links in html is an important part of this project. I expect from our work some new light on what the structure of an hyperfiction is, and how can this structure be built up by the writer.

The need for structure control.

    Second, conditional links make it even more complicate for the writer to control the structure she produces. A large hypertext may permit thousands of different readings. Is the writer supposed to insure consistency for all these tremendous different readings? To my opinion, the answer must be yes:  since we provide the reader with several reading paths, the reader can freely choose, and from his point of view, no doubt that every possible choice must make sense. On the other hand, how hell could the writer secure consistency for all these paths? For sure she cannot test all possible readings. So what? I think that we need here some more abstract tools based on logic and/or computing science theory. This question appears quite difficult but very exciting indeed...

The Writer, the Programmer and the Scientist.

    The range of people working on hypertext goes from pure literature theorists to pure computer science theorists. At both extremes stand a very few researchers. In the middle is a large community of system designers and users. Hypertext is a trans-domain field. It involves many different domains from literature, science, human sciences, ...  This makes hypertext even more exciting, but this will make the understanding of its foundations difficult and slow.

    During the HTWW workshop and the HT conference, I realized how dramatic is the lack of literature and computer science theorists. I will focus on the latter category because this the one I know better. Indeed, as E.W. Dijkstra emphasized it a few months ago in its invited talk at the SAC'99 conference, I want here to talk about computing science rather than computer science: programming computers is a second stage of the job, we must be able to accurately define and analyse what are the computing structures we are computing with. Considering hypertext, we must note that we do not always know what we are computing with, what we are doing with computers. And the same statement holds for hyperfiction with respect to literary theory: we are writing hyperfictions, but we still do not know what hyperfiction is. In short: we all go blind.

    To conclude: more work should be done in trying to understand what hypertext - and more specifically hyperfiction  - is. On the one side, we need computing models to abstract away from technology. On the other side, we need to study what are the possibilities and outcomes of non-determinism in literature. Technology is not by itself the finality: we must know better what we are doing and where we are going.