Browse Books

Reviewer: Hilary D. Burton

… we were struck by a discontinuity. People interested in hypertext, electronic libraries, and information retrieval, the very audience influenced by Memex, knew little or nothing about Bush as an engineer and pioneer of computing machines. People who knew Bush as an engineer and statesman had written little about Memex. We set out to determine what was known about Bushs Memex, to better understand the context from which it emerged, the ideas it represented, and to evaluate the impact it has had on the computer and information sciences (p. ix). The editors address several interesting questions in this book. This intriguing collection of essays includes two commentaries by the editors: “A Machine for the Mind: Vannevar Bushs Memex” and “The Idea of a Machine: The Later Memex Essays.” The remaining chapters are Larry Owenss introductory piece, “Vannevar Bush and the Differential Analyzer”; seven essays by Bush, including the full text of both the Atlantic Monthly and Life versions of “As We May Think”; and eight other discussions: Colin Burkes “A Practical View of Memex: The Career of the Rapid Selector,” Doug Engelbarts “Letter to Vannevar Bush and Program on Human Effectiveness,” Theodor Nelsons “As We Will Think,” Linda Smiths “Memex as an Image of Potentiality Revisited,” Norman Meyrowitzs “Hypertext—Does It Reduce Cholesterol, Too ” Tim Orens “Memex: Getting Back on the Trail,” Gregory Cranes “Aristotles Library: Memex as Vision and Hypertext as Reality,” and Randall Triggss “From Trailblazing to Guided Tours: The Legacy of Vannevar Bushs Vision of Hypertext Use.” The essays address the questions stated in the introduction. Smiths essay, which updates an earlier study she published in 1981, documents the impact, or lack thereof, of Bushs “As We May Think” in the computer and information science literature. The essays by Owens, Kahn and Nyce, and Burke provide the context from which Memex originated. But Engelbarts 1962 letter to Bush requesting permission to include extensive quotes from “As We May Think” in his Air Force project report (also included) on computer-based means to augment human intellect starts to explain the discontinuity. More than anyone else, except perhaps Ted Nelson, Engelbart appreciated the implications of a Memex-type approach to information handling. Yet Bush never replied to Engelbarts letter, and Engelbart himself has never received the acknowledgment or appreciation from the information science community that his early work deserves. The discontinuity is extended by Nelsons claims that Bush is misunderstood—that Memex has little to do with information retrieval and that Bush rejected indexing and settled instead on “interwoven documents.” The primary reason that Memex has never been implemented, conceptually or otherwise, is not that the hardware does not exist to do it. Rather, we have not yet learned how to index so that hetero geneous sets of users can be equally well satisfied. We have not even learned how to index well enough to satisfy a homogeneous group of users without training the users in retrieval techniques and query formulation. Hypertext is one more approach to supporting user-supplied indexing via links and trails and webs. Like Bushs associative indexing, it is a useful approach but not sufficient for all types of infor mation retrieval. It is not the lack of a large corpus of online information but the lack of appropriate indexing for that information that causes us problems. Triggss comment that we have no “large corpi of heterogeneous on-line information” (p. 364) ignores the existence of the thousands of online databases and the work being done to establish protocols for retrieval and data exchange. Even when this work is accomplished, however, the basic problem will still exist. When users can search system Y using system Xs commands, or when they can sit down at their workstations or PCs and wander through hypermedia, what they retrieve will be dependent on the underlying indexing, and we still do not know the best ways to do that. Bushs Memex conceptualized the problem and one kind of solution—the first mechanical, personalized information system. Many other solutions have been tried since then, none of them wholly successful. The essays in this book, taken as a whole, give us a picture of why this is so.