Table of Contents 
Previous Chapter
John M. Carroll
The WWW offers many possibilities for the support of computer science research and the dissemination of technical and administrative information by NSF. Few of these possibilities are technologically routine, and all would entail significant change in the established practices of computer scientists and NSF personnel. Thus, it is critical that the NSF develop an explicit plan to monitor and analyze the usability of various elements of WWW technology as they are deployed.
We recommend support of the development of example task-analyses and information-needs analyses for PI homepages. A basic element proposed in this report is to encourage all PIs to develop appropriately designed and linked homepages. However, currently we do not have an explicit understanding of the characteristic tasks and needs of the users of WWW homepages: What are people trying to do, how are the trying to do it? What are the problems? How do these tasks, needs, and problems vary as a function of the type of user (a computer scientist, an NSF Program Manager, a newspaper reporter)? Of course, such analyses are coupled to the state-of-the-art: the use of and requirements for WWW information will have to be tracked dynamically as innovations in WWW technology and practices are introduced.
The most pertinent task-analyses and information-needs analyses will, of course, be empirically grounded. Today, many appeals to "use" of the Web refer merely to large access logs. It is important to know that Web pages are accessed, of course, but they can be accessed for many different reasons - including misleading links and navigational errors. We need to promote a detailed understanding of what people are doing with the Web, how they are using it, and what their real needs and problems are.
For NSF's immediate concerns, some subset of PI homepages should be instrumented to create a database of usage for subsequent analysis and modelling. This, in turn, raises many research questions regarding methods and instrumentation for remote evaluation, and policy issues regarding informed consent. An important example of remote evaluation methodology is Georgia Tech's Graphic's Visualization & Usability Center's extensive research with surveys
<URL:http://www.cc.gatech.edu/gvu/user_surveys/User_Survey_Home.html>. Another example is Virginia Tech's adaptation of the critical incident method <URL:http://hci.ise.vt.edu/cgi-bin/wwwproj/story>.
The result of this effort should be an explicit set of Web page requirements. These requirements should be exemplified in prototype page designs: this will ensure that the requirements are operational and facilitate their application in the short-term, and will allow the requirements themselves to be directly evaluated empirically in the longer-term.
Understanding and facilitating new uses of the WWW can be integrated through a focus on highly visible "model farm" projects that would be extensively instrumented, closely documented, and fully accessible to the World-Wide Web community. Such projects would provide an beacon both to the PI community and to the public at large with respect to NSF's efforts at modernizing its information dissemination. They would convey concrete models of the process of managing and utilizing Web information resources. The best candidates would be NSF projects that involve substantial project management, for example, through the collaboration of several institutions or groups -- the six new digital library projects are excellent candidates for this.
In the Introduction, Foley refers to the magnitude of potential change unleased by the Web. It is essential for us to keep reminding ourselves that these changes may in fact be historic - the technologies we use every day become mundane to us very quickly. The institutions of science are clearly transforming themselves, and this transformation may in the end be quite profound. At the same time, because this transformation pertains to communication and collaboration across electronic networks, the tokens and patterns of the change are easily captured and organized. A systematic effort should be undertaken to build a history of the World-Wide Web with respect to its role in the management and conduct of science.
Table of Contents 
Next Chapter