The question of control of the reader's browsing behavior is a central issue in hypertextual literary writing. Fragmentation alone risks to breed chaos: a set of independent lexias, without any order or structure, is much likely to generate inconsistent readings. Such an unstructured set of lexias may constitute a work by itself, the writer may by striving to capture some chaotic world...  but here is not our concern: in most cases, the writer wants to maintain control over the potential readings, i.e., over the different sequences of lexias her work allows for. Static links provide an answer: a work is a directed graph whose nodes are lexias, and the reader proceeds by following links. But for many use, this still appears not satisfactory ⁽¹⁾ and the writer needs better, finer tools. Conditional links were introduced with this aim, and proved to be a useful and efficient tool in practice. However, providing the writer with more power in specifying the potential reading orders rises up another question: how can she maintain control over these readings? For large hypertexts, it hardly seems possible for the writer to be aware of all the potential readings, and she wants to have some control over them anyway. But how could she test thousands of possible paths? Indeed, whenever she defines a conditional link, she imposes a local constraint on the reader's browsing behavior. On the other hand, she wants to maintain a global control over the structure that results from these local constraints ⁽²⁾... clearly, there is no straightforward solution to the problem.

        In this workshop, I would like to discuss control related issues including the following questions:
- What kind of control does the writer want to possess over the link structure of her writing? With what tools?
- Should this control be partial or total? In other words: should the writer be able to secure consistency for all potential readings? What actually is consistency in the context of hyperfiction?

        The link structure of hypertext has strong formal, logical bases. Analysis and proof frameworks from research in computer science may be adapted. In particular, I will present at the Hypertext'99  conference a paper entitled  Structure Analysis for Hypertext with Conditional Linkage in which I propose such a structure analysis tool. I think that this paper can pave the way to practical tools and methodologies for the writer to gain a better control over the structure of her writing. I would like to discuss these issues with people experienced in hypertext writing. Actually, the practical question is: what kind of tools writers are expecting from system developers in order to help them to design and control the structure of their writings?
 

(2)  On this subject, Robert Kendall wrote in  Testing, Testing : "It's  relatively easy to ensure that the reader's options at any given point will be meaningful on a purely local level, but it's much more difficult to foresee the cumulative results of all the individual choices and to gauge whether every possible text realization will be satisfying as a whole."