Although there may be grounds for suggesting that the conversion of printed documents to hypertext format may not be as daunting as it initially appears, this issue is seen as comparatively trivial by many proponents of hypertext. This is because conversion in this fashion does not attempt to alter the logical structure of the text - the arrangement of paragraphs or sections within sequential chapters is largely maintained, even if accessing and manipulating them is very different. Writing directly in hypertext enables completely new text structures to be considered by the author, and the design problems raised by these possibilities are generally seen as the main issues in the creation of hypertext.

The design of any non-fiction text can be seen as reflecting the interaction of tool (i.e., text), task and user (as discussed in Chapter 3) and contemporary readers are sophisticated enough to readily accept a whole range of text types. A telephone directory, car repair manual and college textbook are clearly all 'books' although they are designed to suit very different requirements. This acceptance is only realised individually after many years of experience in using books, and is also the product of a cultural experience with books stretching back hundreds of years (as discussed in Chapter 2). It involves a set of expectations and skills, cognitive and manipulative, which are associated with some of the most basic conventions of book design. These conventions (e.g., contents list at the front, index at the back, and binding down the side rather than at the top of the page) are rarely violated by the designer and permit a skilled reader to approach almost any book and use it with confidence. This situation contrasts sharply with the unique organisational structures and interfaces of hypertext systems.

The various hypertext systems currently available reflect the demands of a wide range of domains and potential tasks and it is important to bear in mind the notion of 'horses for courses' when considering whether a particular hypertext system is appropriate for a given application. Computer supported systems such as hypertext are never likely to attain the level of commonality that is characteristic of books. Printing has been a relatively stable technology for 500 years; computing is not only new and rapidly evolving, but one of its characteristic features is the separation of information from storage medium. Computer text has no set physical shape – computers create virtual worlds. However, this is not to say that every hypertext system need, or should, be unique. A well designed hypertext system will not only embody principles of good user interface design but will also recognise the models that users are likely to bring to the system - models which may well be based on experience with print technology, as we discussed in Chapter 4.

Marchionini and Shneiderman (1988) recognised the breadth of strategies that could be supported by hypertext systems, from semi-directed browsing to systematic, iterative strategies using Boolean operators and a thesaurus of terms. They optimistically propose that the solution is to build