Book title: Hypertext and the Technology of Conversation: Orderly Situational Choice

Greenwood Press, 1993, ISBN 0-313-28962-X

Andrew Dillon

This item is not the definitive copy. Please use the following citation when referencing this material: Dillon, A. (1995) Review of S. Gray Hypertext and the Technology of Conversation: Orderly Situational Choice, Greenwood Press, Westport CT. LISR, 17(2), 188-189.

Hypertext has been varyingly portrayed as the technology of liberation and the panacea for information storage. Liberation is claimed by those who view traditional text (read paper) as inherently linear and thus constraining on the part of the reader who must use it. Information storage and retrieval problems are supposedly cured by the ability to semantically link and follow related material through a network of nodes, selecting and rejecting information on the move. While the full potential for hypertext applications in our information technologies has not yet been tested or realized, it is apparent to anyone who uses hypertext for real tasks, or who keeps an eye on the serious research that examines the design of more usable technologies, that there are some real user problems associated with navigating large electronic spaces.

The present volume is an extended analysis of the use people make of hypertext information. In particular it is a study of how novice users of hypertext technology actually react when faced with a hypertext information resource and are asked to find information. Seen through the interpretative lens of conversation analysis, the author advances the argument that, sociological perspectives on meaning construction and the analysis of the sense that users try to make of their interaction with the device, is likely to be useful for guiding our designs of interactive technologies. She contrasts this approach with what she casts as the traditional evaluation approach of the HCI community, that of error counting and retrieval speed.

The book has three main parts. In the first, the author sketches out her argument for this alternative perspective. Interestingly, she pulls together a substantive supporting cast of references and quotes to justify the approach which perhaps raises questions about the characterization of the "traditional" that is offered. As a traditionalist in this domain, I found little familiar in her characterization of the tradition or much of which I needed convincing in the supposed alternative. This section gives way to the main body of the work which is an analysis of the interactions, divided across chapters looking at specific issues such as navigation, categorization difficulties and indexicality (the context-dependence of meaning). Concluding remarks draw the work together and test the idiosyncrasy of the materials used.

In all, the entire book is based on the extended analysis of only 10 users, naive both in the use of hypertext and the material they were exploring, over a maximum of two hours of interaction. Readers were required to think-aloud while interacting and to sketch their impressions of the structure of the text at the outset, the middle, and the end of their interaction. The resulting text is peppered with their rough outlines of nodes and links, mostly partially formed and scant in detail, which the author uses to justify her interpretations of the readers' thoughts, knowledge state, perceptions and metaphorical representations of the information being explored. While this is no doubt interesting work, no reliability checks are taken on the interpretation of data by other raters. Instead, the author "assumes an overlap in reasoning between the author and reader that, if quantified, would indicate reasonably high inter-rater reliability" (p.37). Reasonable for this author is equated with 72% agreement found in other researchers' unrelated studies, a rather sweeping assumption one might think, given that such a figure really only accounts for 50% of the possible variance between raters and there is no reason to assume these different data sets are even comparable.

The general approach leans heavily on the work of Suchman, who performed similar sorts of analyses of users, and brought the perspective of situated action and meaning construction to a larger audience in the HCI field. However, there is little of the originality that made Suchman's work so impressive and at the end of this book one is left with little added knowledge about hypertext design. Instead one gets the feeling of being given a lesson in ethnomethodological analysis for its own sake. While this might be useful for those new to evaluation, or students considering this as an area of inquiry, a cynic might conclude that the methods described here tend more towards endless interpretation than substantive insight.

In all, the book is interesting though irritatingly biased at times (the equation of empiricism in HCI generally with positivism in science is particularly so). It tells us nothing about hypertext that we did not know before but might offer some clues to investigators considering ethnomethodological analysis in this arena. At 280 pages, one might be forgiven for thinking the author took material sufficient for an article and extended it unnecessarily for a book but in her defense, the nature of the material requires space to be considered fully, it is just that one wonders if quite so much space was required. At nearly $60 it is also not likely to be snapped off the shelves by hungry students. However, in a domain where so little effort is expended analyzing users, and more given to predicting the wonders this technology will yield, we must be grateful for any small measure of real data, and that is precisely what the present book provides.