A kind of library?

From the name alone, it is implicit that a digital library is some kind of library. What is at stake in this implication? Are we merely invoking a library metaphor? What are essential properties of libraries? Levy and Marshall (1995: 82), for example, argue that the World-Wide Web is not a library because (1) it is more essentially part of a potential library infrastructure, (2) its contents do not constitute a collection organized for a particular clientel, (3) it lacks institutional services, such as collection development and cataloging, and (4) there has been little attention to integration of digital and nondigital material. These points entrain a good, strong definition of library, but also relegate the term to a tiny niche in the landscape of contemporary library technology.

I urge that we avoid this niche concept approach. Who is the particular clientel for the collection of the New York Public Library? Traditional libraries have paid little attention to the integration of digital and nondigital material: sound recordings, videotapes, physical specimens are usually are placed in annexes, apart from the main item (bound books). It is not likely that digital libraries will turn out merely to be libraries of digital materials. We need to discover and assess new possibilities and new concepts.

Many new possibilities were discussed at Allerton: Bishop talked about an online art gallery as "a novel medium for creating art" -- creating art, not just storing and retrieving art. Cochrane talked about libraries as educational settings, of reference librarians as "instruction librarians." Janes addressed the "impact of information systems in people's thoughts, work, lives, and communities." Levy described the "collaborative use of collections of heterogeneous materials." Reich observed that "libraries are important as community entities and as spaces for communication and companionship." Twidale talked about "collaborative browsing" in libraries.

Scenario-based design

How can we flexibly leverage existing library concepts, but still remain open to new possibilities? Scenario-based design provides one approach (Carroll, 1995). The basic idea is to explicitly "represent" design possibilities in terms of the use people could make of them. Thus, an online art gallery would be depicted as a set of stories about art gallery activities from the view points of various stakeholders (artists, art lovers, art dealers, software designers). Perhaps their most important characteristic is that they are both concrete and rough: They allow design deliberation to be very specific about implications for use, and yet to avoid premature commitment to specific concepts and implementations. They are a "language" accessible to all stakeholders, encouraging and facilitating broader participation in design.

Many aspects of a design can be represented as scenarios: Requirements can be motivated and described in stories about problems in current usage situations that should be addressed and improved. Prototypes can be envisioned and initially specified as stories of alternative usage situations (showing how requirements can be addressed, but also emphasizing issues that remain to be addressed). The entities, relations, and behaviors in scenarios provide initial object models. And so forth.

Scenarios can be attested or inferred. Formative evaluation scenarios are typically observed episodes of actual use that are used to discover new requirements and specifications for redesign. Envisionment scenarios are hypotheses about the use situations that might obtain if a particular design idea were developed. Scenario-based design keeps the spotlight on human activities that will be facilitated, obstructed, and transformed by new technology.

Digital libraries for information and communication

Many examples of retargetting digital libraries through scenario-based design involve the discovery that digital libraries promote communication at least as dramatically as they do information retrieval and presentation. The Computer Science Department at Virginia Tech has been working in the last two years to pervasively incorporate Web-based resources into its curricula (supported by an NSF Educational Infrastructure grant, directed by Edward Fox; see Fox & Barnette, 1994, and http://ei.cs.vt.edu/EIproj.html). Initially, our emphasis was on digitized readings and other materials, class management, and classroom presentation. Obviously, many advances can be made in this arena.

In my view, one breakthrough in our work occurred when Stuart Laughton and J.A.N. Lee decided to explore digitizing an ethical debates activity we include in our Computer Professionalism course (see Laughton, 1996, and http://ei.cs.vt.edu/~cs3604/). They adapted the paradigm of Web Interactive Talk (WIT) to create a forum for structured asynchronous discourse: Students *created* the information in these debates; they did not just read it. The activities enabled by Laughton and Lee's debate forum have rippled through our department's approach to Web-based courses. Several courses have now implemented discussion forums, and this capability has become part of our model of digital libraries for education.

A second example in my experience is the Blacksburg Electronic Village (BEV, see http://www.bev.net). The BEV is a highly advanced community network in several senses: it has a relatively high rate of penetration in the Blacksburg community, it incorporates more high-speed network infrastructure than is typical of community networks; it has an extremely high rate of participation by community groups, local businesses, and the town government. Curiously, however, it was somewhat underdeveloped with respect to computer forums: the BEV development group maintains an official mailing list, some individuals run their own mailing lists, there were a couple of fairly sluggish newsgroups (and of course everyone exchanges 1-to-1 e-mail).

This is curious because the history of community networks is largely a history of electronic forums like newsgroups and other conferences. Brian Amento, Jonathan Kies, Michael Mellott and Craig Struble (1995) decided to explore supporting real-time social interaction within the BEV. They built MOOsburg, a multi-user domain geometrically modeled on the town of Blacksburg. This was an overwhelming success and drew participation from throughout the BEV community (though it was heavily weighted toward Virginia Tech undergraduates, a typical demographic fact about MUDs and MOOs). Apparently, social interaction is a category of activity that the BEV community wanted more of; recently, the BEV development group was inspired to create a community chat room.

The BEV history swallows the BEV

Also in the context of the BEV, Andrew Cohill, Edward Fox, Mary Beth Rosson, K. William Schmidt, Neill Kipp and I are building a digital history of the Blacksburg Electronic Village, a database of e-mail, old Web-pages, design documents, audio recordings of BEV pioneers, NBC Nightly News video clips, and so forth. The database of these documents can be browsed as graphical and chronologically sorted views of events, searched by keyword, and contributed to by members of the BEV community via a Web-based browser. We are using the Postgres database management system, and cgi-forms for the front-end. We refer to the ensemble as the BEV HistoryBase (see Carroll, Rosson, Cohill and Schorger, 1995, and http://history.bev.net/bevhist/).

Our initial objective in this project was to study organizational memory and design rationale. We wanted to provide the BEV development group with access to intact and unreconstructed records of what was done and why to see how this could support and change their routine work. For example, someone in the BEV group might be seeking a press release used last month for Gannett so that it can be edited for the Washington Post, or trying to recall how a particular design issue in a community group's homepage was settled in the past, so that the approach can be reused.

We quickly identified two further objectives. Documenting the history and rationale for the BEV project with a Web-based information system created the opportunity to make the BEV "model" more accessible to developers of other community networks. This also created the opportunity to integrate history-building into the BEV itself: We decided to support collaboration and community in the BEV through the collective activity of building a design history.

To date, our principal method has been archiving by hand: we have collected historically significant documents, adding them, or summaries of them, to the HistoryBase one-by-one. These include vision statements, manuals and brochures, interviews with BEV pioneers, press reports, and so forth. We see this chiefly as remedial; we need to preserve and organize these materials before they are lost or corrupted.

Shared goals and collective effort create human communities. Thus, we see history-building work -- including work on the HistoryBase -- as potentially constitutive of the BEV as a human community. We believe that a sustainable HistoryBase must be *owned* by the community. However, we also appreciate that people are often narrowly task-oriented. This makes it difficult for them to allocate time and effort for tasks directed at reflection, integration, or more generally, at the collective good. This phenomenon has been widely noted in post mortems of recent CSCW applications.

To successfully integrate history-building into life in the BEV, we need to closely integrate the effort of history-making with task-oriented activities. Within the BEV development group, we have tried with fair success to establish history-building procedures. Some of these are programmatic (the "message of the day" is automatically saved in the HistoryBase), others are practices (the HistoryBase is the primary repository for press releases). Through these means, important technical activities are logged into the HistoryBase as they occur.

We are now trying to extend this approach to all members of the BEV community by rendering first-approximation history-making as a by-product of routine activity. We developed a Web-page authoring tool that makes it easier for community groups to create their own BEV home pages from templates of typed fields. These structured group home pages programmatically archive group information in the HistoryBase when it becomes "past."

For example, the advance notice for a picnic posted by a church group would automatically be logged in the HistoryBase when the date of the event became "past". Thus, the task-oriented effort directed at creating and maintaining a group's presence in the BEV yields structured historical information without requiring any further effort. The automatically generated material can be edited by the event became "past". Thus, the task-oriented effort directed at creating and maintaining a group's presence in the BEV yields structured historical information without requiring any further effort. The automatically generated material can be edited by the owner of the page.

This approach reconceives the community network itself as a community working memory, one of whose functions is to consolidate current activities into shared memories. Currently, we are working with community groups to facilitate their adoption of our Web-authoring tool in order to study how the groups manage the HistoryBase documents generated by their activities in the BEV.

Driving technology with use

Are the CS debates a digital library? Is MOOsburg? Is the BEV community chat room? Is the BEV HistoryBase a digital library? Perhaps another term is all that is needed. I think in these projects we have been *discovering* what digital libraries are about by allowing explorations of use to drive the evolution of our digital library systems. This has led us to novel technology concepts -- like the concept of a community-MOO and an instrumented group homepage, all the time driven by scenarios of use.

References

Amento, B., Kies, J.K., Mellott, M., and Struble, C. 1995. MOOsburg: Experiences with a community-based MOO. (submitted for publication).

Basalla, G. 1988. The evolution of technology. Cambridge University Press, New York.

Carroll, J.M. (Ed.) 1995. Scenario-based design: Envisioning work and technology in system development. John Wiley & Sons, New York.

Carroll, J.M., Rosson, M.B., Cohill, A.M. & Schorger, J. 1995. Building a history of the Blacksburg Electronic Village. Proceedings of DIS'95: First ACM Symposium on Designing Interactive Systems. (Ann Arbor, MI, August 23-25), ACM Press, NY, pp. 1-6.

Fox, E.A. & Barnette, D. 1994. Improving education through a computer science digital library with three types of WWW servers. Proceedings of WWW'94: Second International Conference on Mosaic and the Web. (Chicago, IL, Oct. 17-20).

Laughton, S. 1996. An ethnographic study of Internet-based applications in education. Ph.D. Dissertation, Department of Computer Science, Virginia Tech (in progress).

Levy, D.M. and Marshall, C.C. 1995. Going digital: A look at assumptions underlying digital libraries. Communications of the ACM, 38:4, 77-84.