HTML 4.0 Sourcebook
(Publisher: John Wiley & Sons, Inc.)
Author(s): Ian S. Graham
ISBN: 0471257249
Publication Date: 04/01/98

Chapter 4
The Design of Web Collections

Chapters 1 and 2 provided a gentle (I hope!) introduction to HTML and to good design habits for creating HTML documents, while Chapter 3 outlined important issues associated with Web page graphics. This chapter takes a broader approach and looks at the issues involved in designing collections of Web documents and associated resources. In the printed world, there is a difference between designing a single page and designing a complete magazine, book, or library. Such document collections require organizational and design elements that are neither necessary nor apparent from the perspective of a single page. The same is true of hypertext collections, although here the required design elements are quite different from those needed in the purely printed world.

Why such differences? The reasons lie in the nature of the presentation media: Books are spatial, physical, static collections, with a fixed, linear structure, while hypertext is nonspatial, nonphysical, possibly dynamic, and often nonlinear. Good hypertext design must embrace these differences, while preserving the easy navigability of printed books. This chapter looks at some ways of accomplishing this goal and provides references for additional reading on this subject.

Paper and Books

The easiest way to appreciate the main issues is to start with the familiar example of a book. This lets us introduce, using a familiar model, the ideas behind structured document collections. The issues that arise in hypertext design can then be introduced and analyzed with respect to this more familiar case.

In simplest terms, a book is a collection of related, printed pages. Of course, there is much more to a book than that! A large collection of unbound and unnumbered pages is, to say the least, a cumbersome format (rather like the floor of this author’s office as he sits writing this chapter). Given a pile of printed pages, a reader cannot distinguish between pages arising from different books or documents (should there be pages present from more than one collection), and cannot, even within a collection of associated pages, determine the proper reading sequence without explicitly checking for page-to-page continuity. The problems are essentially navigational: There is no easy way for the reader to figure out how to read the pages as a coherent whole.

Book or magazine design solves such problems by giving the pages a uniform design (top and bottom page banners, typeface, and so on), so that pages have a distinctive look; by numbering the pages to give linear order to the collection; and by binding the pages together to enforce the correct order. If there are many pages, or if there are organizational requirements, there is often a table of contents listing the page numbers of important starting pages and perhaps an index providing references to other significant locations. By convention, such content-listing or indexing tools are placed at the beginning or end of the volume (the exact location varies according to linguistic and national conventions), to make them easy to find. Additional cross-referencing is possible through internal page references, footnotes, bibliographies, and so on, while additional indexing components are present within specialty books, such as dictionaries. Indeed, the organizational technology of printed material is very sophisticated, covering everything from simple pamphlets to multivolume encyclopedias. This is not surprising, given that this technology has been refined over 500 years of practical experience.

Linear Documents

Books and other printed media are all, essentially, linear. By linear, I mean that they have an obvious beginning and end as well as a fixed sequence of pages in between. Indexes, tables of contents, or cross-references exist superimposed on this linear framework—they provide added value and are often critically important—but they do not change the underlying structure. In fact, they depend on the underlying linear structure, page numbers, and so on, to provide reference points within the book.

The reasons for the near universality of this linear model are both physical and psychological. Physically, the only reliable way to organize printed pages is as a bound, linear entity—it is hard to create a book as a random collection of connected but nonlinearly accessible documents! Psychologically, a linear, well-defined structure is comfortable, familiar, and convenient, since the result is easy to read, easy to reference, and easy to communicate to others. The goal of all publishing is communication, and a book is a robust collection that can be reliably communicated to others (through duplicate copies) and reliably referenced and compared (through page number references), since everyone with the same book has the same information, at the same location within the book.

It is also important to note that the physical nature of a book provides a psychological comfort zone to readers, by letting them know both the exact size of the book and where they are within the book. This makes it easy to browse a book, for example by jumping from the table of contents to a selected location or by simply selecting pages at random, all the while retaining a sense of location with respect to the beginning, end, table of contents, or index.

Other traditional media, such as music, video, and film, are also linear in this sense, being predetermined presentations of sounds or images, created by a musician or director. This, in part, reflects the temporal nature of these media—music and film move dynamically (and usually forward!) in time in a linear way. This also reflects the technical limitations of the media, as it is almost impossible to make nonlinear presentations with traditional film, video, or audio technology, just as was formerly the case with printed text.

Figure 4.1 illustrates both the structure of a book and how the table of contents and index merely provide referencing on top of the underlying linear structure.