Building a Campus Information Culture i 1 CAUSE '92 Building a Campus Information Culture Robert J. Brentrup Dartmouth College Hanover, NH 03755 Abstract Delivering information resources is not merely a technical challenge, but an organizational one as well. An integrated information system benefits all parts of an institution by meeting individualÕs information needs through an accessible and familiar interface. This paper describes the experiences of the author and his colleagues in augmenting the campus computing culture through the development and deployment of a distributed campus-wide information system. The Dartmouth College Information System (DCIS) based on a client/server architecture is presently in widespread use across the institution and is experiencing rapid growth. Learning to work with the institution through its individual members and distinct communities of interest has been as important as the technical design and implementation of the system. This analysis of our venture should benefit others pursuing similar goals. Building a Campus Information Culture Computing at Dartmouth The model of computing at Dartmouth was established by the pioneering work of John Kemeny and Tom Kurtz. Dr. Kemeny believed that computing services should follow the example of the library, that is, service is provided as a public utility. At Dartmouth, computing is centrally funded and provided at little or no charge to all members of the community. Services are aimed at reaching everyone in the institution. In the days of the Dartmouth College Time Sharing System (DCTS), BASIC was the common language of computer discourse and sharing computer programs was easy. Although use of DCTS is coming to a close, everyone on campus is still automatically given an account on DCTS when they join the Dartmouth community as faculty, staff, or students. The expectation was and still is that everyone is encouraged to use the services. The standard account has been replaced by an entry in the Dartmouth Name Directory (DND), a database of all students, faculty and staff. A DND entry provides your electronic mailbox and is your identity in DCIS. Computing services have been planned by a central computing organization with much input from the entire campus. The basic services have been centrally developed and maintained. This has minimized duplication of effort and reduced inter-operability problems in order to maximize the extent to which individuals are able to interact easily with each other. By building on this common foundation, the different departments have been able to pursue their varying interests more directly. They have in many cases developed for themselves a wealth of beneficial applications. Special computing offices in engineering, the humanities, and the social sciences develop applications targeted to their respective disciplines. Computing technology is readily available and can easily be applied wherever appropriate in all disciplines. The Network The Dartmouth network has its beginnings in DCTS. Terminals were provided in increasingly remote locations from the computer center. In the effort to support large numbers of simultaneous users on a relatively modest central computer, the network engineers distributed processing power in front of the main system. These devices evolved into the present system of network routers located in the building wiring hubs. The system terminals acquired additional Z80 microprocessor boards to support interactive editing. Currently all dormitory rooms, offices, and meeting rooms have network ports (explaining the motto Òa port for every pillowÓ). The wiring is mostly unshielded twisted pair. Each building is a zone on a very large AppleTalk network. The building hubs are connected by an Ethernet. A second Ethernet interconnects the current collection of mini/mainframe computers. Additional Ethernet segments connect workstations in the computing center, the computer science department and the engineering school. Ethernet links are now starting to replace selected desktop connections to some of the more than 6,000 Macintosh computers in use. Network Services Using personal computers in place of the DCTS intelligent terminals was a natural next step. The idea to select a standard personal computer for the entire campus grew out of the desire to maintain the ease of data and program interchange that was a primary feature of DCTS. In 1983 when this was being planned, the finalists in the system selection were the IBM PCjr, the DEC Rainbow, and the Apple Macintosh 128K. The Macintosh was chosen to be the standard. This decision has obviously had a significant impact on the current state of computing services. The introduction of the Macintosh was accompanied by the local development of network services. The first project was a network stream protocol and a terminal emulator to replace the DCTS intelligent terminals. In a relatively short time (1986), public file servers, public laser printing, and Macintosh-based electronic mail became available to everyone. These network services support the daily use of about 4,500 students, 825 faculty, 2000 staff members and about 5000 people at the Dartmouth Hitchcock Medical Center. The internal implementations change but the original services are still available. For example the electronic mail server was originally hosted on DCTS and today runs on a network of UNIX workstations. Compatibility with older ways of doing things is always considered when introducing a new service. The electronic mail system offers an interesting example. The default mail destination is paper, that is, it may be printed and distributed through the campus mail system. Users of the electronic mail system are thereby guaranteed that they can communicate with everyone on campus, though at varying speeds. Individuals not using electronic mail may find out about it in this way; they receive slightly unusual-looking memos in their campus paper mail boxes. Adoption of standards has simplified providing breadth of service. Software developers can concentrate on a single platform knowing the entire community will be able to use their product, a significant advantage. This narrower focus provides the opportunity to pursue greater depth of service. Instead of investing effort in providing duplicate services on multiple platforms, services can attain greater sophistication and attention to detail. The emphasis of Dartmouth computing remains on creating services that everyone on campus can and will use. New services start as demonstration projects but they are designed and developed to the point of universal availability. DCIS is the next step in this evolutionary process. While the first network services, like mail, are content free, DCIS development has concentrated on providing access to actively collected, content-laden information resources. The network infrastructure is a essential base on which to build a system like DCIS. Need for an Information System The utility of a campus-wide information system can be seen in every department and office. In the academic departments, the use of electronic information sources in research has grown rapidly. Researchers routinely pursue electronic sources on the global Internet. Through the Internet they can communicate easily with their peers regardless of their physical locations. Increasingly widespread electronic information access is providing an immense collection of valuable resources available from oneÕs desktop workstation. Every administrative office has both information they need to manage their operation, and information to share with the community. Each office faces the need to accomplish these tasks with greater accuracy, more efficiently, with less cost. Consumers of the information want to use it easily and in conjunction with their other computer tools. For them the computing technology must remain a secondary consideration, a tool in their work and not the focus of their efforts. DCIS Project Organization The DCIS project is a joint development of the central computing organization and the library. It was apparent in the initial stages that such a partnership was natural and essential. The library and computing services each supply critical skills in making the project possible and successful. Co-principal investigators oversee the operation of the project. Developing this service centrally eliminated much duplication of effort. In practice the team is so small it is essential to collaborate. The Dartmouth library, like Computing Services, has been a leader in the development and use of information technology. Development of an electronic card catalog began in 1979. The entire collection has been cataloged for many years. The physical card catalog has been closed since June 1991. Some new branches of the library never had a card catalog. Material acquisitions and circulation are automated and integrated with the catalog. Daily operations are heavily dependent on these systems. In the DCIS partnership, the library functions as a central clearinghouse for the selection, acquisition, and maintenance of electronic reference resources. The library staff provide content expertise and are advanced users of the information services in their everyday jobs. They also have frequent contact with the students and faculty using these services and provide important insights and feedback to the software developers. The computing center installs, operates, and maintains the central database servers. It provides the physical network and the logical data transport between clients and servers. The DCIS team are the information system software developers. An array of software clients and servers have been developed to deliver a variety of information. Development team members also act as consultants and intermediaries for other departments seeking to develop their own information resources for campus-wide delivery. Documentation and training have been joint projects of both the central computing organization and the public services of the library. The Dartmouth community has served as subjects in the experiment and in return has gained access to many new tools and resources. DCIS Objectives DCIS is designed to address a number of interrelated objectives. The most obvious objective is to simplify access to information. The software tries to eliminate as many manual steps as possible from the process of locating and using electronic information resources. For example, DCIS makes connecting to a database as simple as pointing and clicking. Each step in lowering the barrier to entry has brought more members of the community into daily use of these resources. It also seemed obvious that access could be greatly improved by harnessing the power of the personal computers available, using them as more than terminals to the mainframe systems. Local processing power has been harnessed in a client-server design that provides more functionality with less user tedium. More sophisticated presentation has been developed than was previously possible. Faster response to requests is being provided with fewer server resources. The ability to leverage individualsÕ existing knowledge of the MacintoshÕs standard program interface is an important goal. The DCIS system fits in with other Macintosh productivity applications. Skills learned in one generally apply to others. Data interchange between different applications is greatly enhanced. For example, collecting citations and formatting a bibliography for a paper are very easy. The viewer formats them and they can be copied and pasted right into your paper. The ability to deal with many resources through the same user interface is another important objective. The information retrieval programs or ÒviewersÓ are able to access a number of functionally different servers in a manner almost transparent to the user. These different sources of information do have similar content. For example both bibliographic and full text resources can be accessed through the same viewer. Another viewer supporting statistical functions has been written for numeric databases. The effort invested in learning to use one of these viewer applications benefits the user many times on many different databases. Another objective was to help users find out what resources are available. To answer the question ÒwhatÕs out there?Ó a system-wide directory pulls together resources from all around the campus. Users can search and browse through the resource collection, similar to the on-line catalog, helping them to discover useful electronic information sources. We wanted to help users locate, acquire, and maintain the necessary client software on their workstations. It is often frustrating and difficult to keep up with the latest versions of so many application programs. DCIS provides a mechanism to update its components as it is developed. Users learn about updates to the software as they use it, with the system delivering the new version automatically if desired. In this way, new versions have propagated quickly through the community. Finally, the system is designed to be very modular and yet provide an integrated feel and operation as much as possible. Many standard Macintosh applications can be used as a viewer. In this way, we have been able to incorporate previously existing systems. This is as essential as providing new ones. This design also provides an minimally disruptive upgrade path as various parts change. Current State of DCIS Project The DCIS project has been in development about 3 years, starting in earnest in January 1990. The first applications were publicly available on the campus network in April 1991. The software has been distributed to all members of two incoming classes (1995 and 1996) as their Macintoshes were delivered. The library staff is now switching their daily usage from the earlier library system, as DCIS has equaled and exceeded the earlier systemÕs capabilities. Well into the second year of production use, we are trying to cope with substantial growth in usage and demand for more information resources. Over 50 local databases and hundreds of external sources of information are reachable by the system. Five central computers running about a dozen different servers currently provide information services to hundreds of daily users. The distribution of the servers on a number of different computer systems allows us to add capacity incrementally with no user-visible system changes. Six new client applications have been developed and several preexisting ones have been integrated into DCIS. The DCIS Navigator application and its system-wide directory server provide a central point of contact with a wide variety of resources. This directory is a way to ÒadvertiseÓ the contents of the system and automates connecting to a resource. It also provides the software delivery and update features. The Online Library application provides sophisticated search and retrieval capabilities to the collection of bibliographies and full texts. Complex searches, local manipulation, caching, and graphical display of results are among its features. The CLASS application provides a similar function for numerical databases. An image viewer has been developed to display illustrations and photographs. Several protocol-translating gateway servers are also available. They provide access to WAIS and Z39.50 services at other campuses and businesses using the same DCIS Online Library application. Collection Development The project started with scholarly electronic data databases at the library, primarily the on-line catalog, indexes to minor special collections, and bibliographies listing journal articles. A number of electronic general reference works have been added including an encyclopedia, dictionaries, and full texts of standard works. These sources have tended to be highly structured, read-only works with infrequent updates. They were selected because of their broad use and previous development work. The directory of resources was then expanded to include existing sources of information available on campus life, such as bulletin boards for events, jobs, housing, and network news. Some of these resources were and are still accessed through terminal emulation, although DCIS has eliminated several steps in using them. These sources have tended to be less structured, read-only works with frequent updates. Improved interfaces for more of these resources are being planned. Using the same terminal emulation approach, selected resources provided by other institutions have been added to the DCIS directory. This makes them more accessible, although each supplies its own unique user interface which makes using a lot of them complicated. As news of the system spreads, a variety of administrative offices have been suggesting and developing new resources. The success of such projects are essential to broaden interest and increase usage of the system. Time and again we have seen someoneÕs casual interest (ÒthatÕs pretty neatÓ) change to active involvement (Òwhere can I get a copy?Ó) when a resource of personal interest to them is made available. Plans are now being developed to incorporate personnel and financial systems into the mix. These sources bring with them a number of complex problems. For example, access control must often be implemented down to the level of specific fields in the database. On-line update of the data, while simultaneously supporting searching, is very desirable if not essential. Local Information Resource Development As mentioned earlier, most administrative offices have information to share with the community, and DCIS has provided a new, more powerful method of providing that service. To assist offices in developing an information resource, the DCIS team usually arranges a discussion meeting to understand the resource and its audience. The data content and its size generally determines the complexity of the project and its update needs. Insight into the officeÕs existing process and work flow are often helpful in determining an approach. We try to involve the office in evaluating possible solutions from the tools available, highlighting the differences and tradeoffs. The DCIS team helps to set up the initial system. The prototype is made available on a limited basis, feedback is gathered and used to refine the implementation. Finally we bring the system up in the DCIS directory and public servers, and hand off update control of the information to the originating office. We usually then mount an announcement effort to bring the new resource to the attention of the appropriate community of interest. Developing some of these projects has illustrated to us how important it is to distribute the maintenance of the information sources. Unlike the central software development, if the originating office cannot largely maintain its own information resources, the information system developers will be consumed by maintenance tasks. Experiences with Deployment We have used a Òripple-outÓ strategy in deploying the new systems. We have sought to identify and utilize early adopters in a department or office. Developing such an ÒinsiderÓ relationship can be a valuable asset. The early adopter often becomes a product champion, distributing information and channeling back other feedback. Individuals in the work group tend to be influenced more by someone in their existing group than by the DCIS development team. It is important to draw experimental users into the process at the proper time. The system under development must be stable and powerful enough to be used for real work at this stage. It must provide equivalent facility when competing with an existing system. It is important to watch out for missing features that people have taken for granted and then realize they miss. To make the most of the opportunity, we are reactive to the tester's feedback and follow up on complaints and requests quickly. Gradually, more people are added to the group of testers as refinements are implemented. Finally, a public announcement is made through all available channels. The switch over period is especially important. It is essential to support overlapping dual systems for some period time. The dual systems have been a problem for the staff; they needed to know both methods for some time and to accommodate various special needs. Substantial functional reasons to switch to DCIS have generally been available and these help to overcome a userÕs natural reluctance to change a routine. To assist the adoption of a new system, we have tried to recognize and address the variety of individual approaches to learning and the computer abilities of persons in the community of interest. Training and documentation attempt to support the range of those who Òjump inÓ and experiment (the early adopters), those who like to read about how the whole thing works as they try it, and those who need a demonstration of it before theyÕll begin to use it. In response, we have provided on-line help in each application, printed extensive user guides, and run ÒminicoursesÓ and hands-on training workshops. The Macintosh interface has enabled individuals already used to the Macintosh to use the system with less instruction and training than earlier mainframe systems. Part of this effect is generational. The students are more likely simply to try the software, and they manage to get useful results. Of course, new students have not invested time in learning the previous system. The faculty and staff require more assistance and need to make comparisons against the previous system. Conclusions We think the DCIS project has been quite successful. A number of similar systems have been in development for longer periods of time and others have been released and subsequently withdrawn. We think one of the keys to this success is to stress performance and quality in the software. At the beginning, we had to trade these against quick results. Performance and quality allow others to use the software as a tool in the performance of their ÒrealÓ jobs. The use of Òindustrial strengthÓ tools and design techniques, especially object-oriented programming technology, has leveraged our development efforts in a very visible way. By being able to draw from a collection of reusable components, our progress has accelerated. Common components help provide underlying levels of common features that look and act the same in different applications. DCIS is designed to change incrementally as it is developed. This allows great flexibility in integrating preexisting resources. A large number of resources have been assembled in a short time. In some ways, our strategy to implement a Òseamless feelÓ with separate components has worked too well. New users have been confused about which application they were interacting with. Their mental model sometimes does not keep up with where the system led them. The advantage of having a ÒstandardÓ workstation has been substantial. It has been difficult enough to account for the variety of models and software versions in one line of development. DCIS software is starting to influence equipment purchases. Workstation performance became noticeable to those who had entry-level models with some deciding to upgrade. People now ask for equipment recommendations because they want to have a workstation that can take best advantage of DCIS. At the same time, moving much of the computation to the workstation has allowed the same central computer to serve more people than before. But the library's central computer has also been upgraded to keep up with increased demand. The client/server nature of DCIS can make the system appear more responsive during some high usage times. Since the client workstation is handling operations like keyboard input and scrolling instead of having the server interacting with a terminal, users don't notice slightly longer pauses at times they normally wait for a few seconds. Spreading the news is still a significant problem. Certain pockets of the campus are still not aware of what is available. Providing a new resource for a particular community of interest has been the most successful way to add new users. Word of mouth has been the next most effective device. Printed and electronic publications (that is, newsletters and electronic bulletins) tend to reach the same group of regular readers. By working closely with the various organizations that have information to distribute, we have tried to ensure a stable base on which many more resources can be provided. The partnership we have assembled seems to provide an efficient way to develop and maintain such an information system. As people have started to use what is available we have received many requests for other resources. Users of DCIS have reported enhanced productivity in teaching and research. The electronic mailbox we setup to collect bug reports and suggestions has received a number of complimentary notes. The most apparent effect has been large increases in the number of people using the system. Usage has more than doubled from summer to fall in 1992. Building a campus information culture is an evolutionary process. Each individual approaches the use of information technology a little differently. By lowering the barriers to entry by several more steps and by trying to cater to a wide variety of experiences and needs, DCIS has been a significant step forward in library and computing services at Dartmouth College.