Co-Chair: William J. Mitchell, wjm@mit.edu
Co-Chair: Michael L. Dertouzos, mld@mit.edu
Harold Abelson, hal@mit.edu
John W. Belcher, jwb@space.mit.edu
Timothy J. Berners-Lee, timbl@w3.org
Peter Child, child@mit.edu
Peter S. Donaldson, psdlit@mit.edu
Julie Dorsey, dorsey@mit.edu
Anne L. Drazen, adrazen@sloan.mit.edu
M. S. Vijay Kumar, vkumar@mit.edu
Richard Larson, rclarson@mit.edu
Steven R. Lerman, lerman@mit.edu
Nicholas Negroponte, nicholas@media.mit.edu
Alex Pentland, sandy@media.mit.edu
Bruce Tidor, tidor@mit.edu
Rosalind H. Williams, rhwill@mit.edu
John S. Wilson, jswilson@mit.edu
Ann J. Wolpert, awolpert@mit.edu
The product of eight years of research and development, the Athena Computing Environment provides computing resources to over 16,000 users across the MIT campus through a vast system of 1300 computers in more than 40 clusters, private offices, and machine rooms, all connected to MIT's campus-wide network.
Athena users have access to software to help them write papers, create graphs, analyze data, communicate with their colleagues, play games, register for courses, and perform countless other tasks, as well as access to software designed specifically for classwork.
Athena has pervaded campus life. At last count, all of MIT undergraduates and 90% of MIT graduate students had Athena accounts. On a typical day, over 6,000 different users access their personal files and various software packages on the system.
The Athena Computing Environment provides widely distributed, client-server computing for education at MIT, with a focus on undergraduate education. Access to central Athena facilities is available to MIT students, faculty, and on-campus staff at no cost to the user. Access to departmental Athena facilities is generally more limited, but consistent with the principles and ends described below.
Athena therefore is designed and operated to maximize the availability of computing for education at reasonable cost to the Institute. MIT-wide optimization requires considerable standardization of systems, and a distributed client-server environment configured to minimize support requirements and downtime. The MIT Information Systems organization (IS) follows one overarching principle in allocating Athena resources: the Athena Computing Environment should contribute to the continuing improvement of education at MIT. Among opportunities for educational improvement, IS attempts to support those that achieve as many of the following four ends as possible:
1. Serve many students, rather than few.
2. Serve the core of the MIT curriculum rather than its periphery,
including General Institute Requirements and large introductory subjects
in popular departments.
3. Seek innovation and creativity, rather than simple automation
of traditional approaches to education.
4. Increase technological equity among departments, rather than
decrease it.
The Athena Computing Environment includes about 700 workstations for general
and departmental use, about 300 private workstations on faculty, teaching-assistant,
and staff desks, and about 100 servers. It is supported at an approximate
annual cost of $6 million in MIT general funds. These funds are used for
staff, software, hardware maintenance and hardware renewal.
Athena uses the distributed client/server model extensively. Only a few essential pieces of system software are actually kept on the Athena workstations—everything else is delivered and maintained remotely. This greatly reduces the burden on the local workstation, and the local machine is consequently freed up to apply almost allrses. From modest beginnings in &q of its computing power to the specific applications being run. More significantly, the client/server model allows centrally-managed services to be concentrated so that a relatively small staff can support the system. Client and server workstations are widely distributed geographically, and extensive security of software on workstations in Athena clusters is not needed. For example, none of the Athena clusters has assigned attendants, but when a machine gets totally disabled, it is typically restored and rebooted almost instantaneously. There are fewer than 20 people who operate and watch over the entire Athena environment of 1,000 client workstations. Athena's design also features the elements that contribute to its robustness and usefulness as an educational computing environment. In particular, the uniform use of the TCP/IP protocols, the X Window system, and systems security through Kerberos allow an extensive installation of networked Athena workstations to have a common interface regardless of architecture, enabling them to share programs, data, and a variety of network services.
The Athena Computing Environment has seven key components:
1. Networked workstations in general-use clusters that are virtually
always open to all of the MIT community (four electronic classrooms are
more restricted).
2. Similar workstations in departmental clusters and other more
private locations, whose accessibility is generally more limited and whose
costs are borne jointly by IS and academic departments.
3. Basic network services (the network "Commons") comprising
a full suite of communications and information-retrieval services, including
electronic mail, World Wide Web services, and selected bibliographic and
reference databases.
4. Commercial tools for analyzing, displaying, manipulating, and
otherwise processing information, including symbolic mathematics, statistics,
text, and graphics.
5. Locally developed curriculum materials associated with specific
MIT subjects.
6. Shared file storage that is accessible transparently from any
Athena workstation.
7. Support for faculty, students, and other users of the system
in the form of training, documentation, faculty liaisons, network help,
and consulting.
A subset of Athena-like services, especially the network Commons and selected
commercial software, is available to personal computers and workstations
connected to MITnet, which extends to all undergraduate dormitories and
independent living groups, to all graduate housing on campus, and to most
offices and other workspaces at the Institute.
Hardware Renewal Strategy
As is to be expected with rapidly advancing technology, Athena workstations, like other computers, start falling behind the current performance/price ratio very soon. Current software often works poorly or becomes unavailable for older equipment, even when that equipment continues to function perfectly well with its original software. Moreover, annual maintenance costs rise sharply as workstations age, often approaching half the cost of new workstations. For these reasons, the goal is to have no Athena workstations more than four years old. As Athena workstations are retired, new workstations replace them. Each spring IS surveys the market, consults with faculty, evaluates hardware and software costs, explores support requirements, and identifies workstations suitable for the Athena Computing Environment.
Software Available on Athena
The Athena system presents its users with a large library of ready-to-use tools and application packages, including specialized software for courses, Internet access, and programming, as well as general software for typical user tasks such as document preparation and electronic communication. All of the software is available to any user logged in at a workstation in an Athena cluster. An increasing number of Athena applications are becoming accessible to users accessing Athena indirectly (e.g., from a dorm room or living group, or from a campus office).
In general, IS attempts to provide centrally on Athena any commercial tool software that is educationally useful for education in more than one department. The current suite of commercial tools includes document preparation, symbolic and numerical mathematics, statistics and graphing, design and graphics, and reference. IS makes a more limited suite of commercial software available for students' personal computers connected to MITnet. IS provides limited help with licensing, installation, and sometimes central storage for curriculum materials and commercial software used only within a single department.
Athena capabilities (in the form of network enabled Athena workstations) have been installed in several classrooms so that educational software may be incorporated directly into the teaching of a subject. Athena's effectiveness as a comprehensive environment to support education is enhanced by resources beyond the physical installations and the set of core services. This includes software and services such as:
1. NEOS (Networked Educational On-line System also known as Turn-in/Pickup)
which allows students to turn in, and their instructors to grade and return,
assignments electronically.
2. OLTA (On Line TA) a system by which students in a course may
consult electronically with their TAs while logged in and working on an
assignment.
3. OWL (On-Line With Librarians) is a system by which patrons may
consult electronically with reference librarians from the various MIT Libraries.
4. Tools for programmers.
5. Access to a suite of other electronic library services, databases
and reference tools.
6. Assistance through the On Line Consulting (OLC) service from
the Athena Consultants; support for faculty, students, and other users
of the system in the form of training, documentation as well as consulting
and project assistance from faculty liaisons.
Major Limitations of Athena
The current Athena system has served the community well for more than a decade and has clearly demonstrated the tremendous value and even greater potential of using high-end technologies for education and for research. The following shifts in technology and the expectations of our faculty and staff need to be accounted for in planning any transition:
1. It will soon be possible to affordably deploy networks throughout
the world capable of transmitting real-time video.
2. The Internet has become the nearly universal digital communications
medium around the world.
3. There has been an explosion in the interest and use of computers
and network access resulting from the development of the World Wide Web
and the associated standards being developed for information transmittal.
4. The development of authoring tools now makes it feasible for
non-experts to produce multimedia teaching lessons.
5. The desktop computing environment for the vast majority of the
computer users is dominated by personal computing hardware and software,
leading to the creation of a huge amount of software for the Windows and
Macintosh platforms and more extensive price competition among hardware
vendors in these markets.
MIT's plan for its information infrastructure should retain the strengths of the current Athena system: relatively low cost of operations per user, widespread availability, an orientation towards supporting MIT's educational mission and a cadre of talented staff to help select tools and develop educationally valuable materials. However, it should, also add new capabilities and functionality.
Among the most crucial limitations of Athena, as we look towards the future, are:
1. Limited bandwidth and lack of multimedia capability. In its present form, Athena does not support effective distribution of large quantities of high quality images, audio, and video, and it does not support high-end videoconferencing. To most locations it does not provide the necessary bandwidth, and most of its workstations have limited graphics and video capability.
2. Limited off-campus use. Although there are some provisions for remote access to Athena, these do not suffice to support high quality distance education. This creates an undesirable technological distinction between on-campus and off-campus communities.
3. Inability to make the most effective use of commercial software. The relatively complex process of integrating commercial software with the Athena environment required to ensure the secure and reliable availability of software, occasionally creates a barrier to quick, effective exploitation of the latest products—some of which potentially have great value to us—and sometimes requires us to continue to support homegrown applications when commercial products could now do the job more effectively and inexpensively. Operating systems and applications on Athena are frequently a generation behind to accommodate the integration with software resources and the file system, thereby making it difficult for some useful applications to make it to Athena.
4. Limited flexibility and limited capacity to support bottom-up efforts. If we want to tap faculty and student initiative, energy, and creativity in the most effective way, we need more Web-like capacity for quickly and easily linking individual, bottom-up efforts into a larger structure, and less reliance on top-down integration strategies.
5. Limited support for privately owned personal computers. The majority of the faculty, students and staff that have their own computers (purchased either personally or with MIT funds) are not fully integrated into the Athena computing environment. While these machines are typically connected to the MIT network, they do not have access to the full spectrum of software which is available on Athena and they cannot easily access the shared file storage Athena provides.
MIT currently produces publications in a wide variety of modes and formats. It has established and continues to develop mechanisms—both formal and informal—for collecting, evaluating, and distributing intellectual property. And its libraries, archives, and museums play the role of preserving, indexing, managing, and consulting on the use of it.
Informal Products
MIT faculty members and students produce a huge amount of intellectual property in quick, informal ways—in the form of course outlines, spoken performances, chalk-and-talk, demonstrations, lecture notes, reading lists, compilations of readings, laboratory notebooks, problem sets, examinations, student papers and designs, theses, and, so on. Much of this is of great value, but it may go unrecorded; if it is recorded, it is often done so in rough and impermanent ways, and it is rarely systematically compiled and indexed for wider, longer-term use.
The World Wide Web has very quickly done much to change this by providing a fast and inexpensive way to mount materials online. Furthermore, the associated search engines and indexing services provide a remarkably effective, low-cost way of sifting through vast amounts of this material to find exactly what you need at a particular moment. So it is now commonplace to find course materials, drafts of papers, and the like on the Web, and informal Web publication is likely to play an increasingly vital role.
As consumers of this sort of information, members of the MIT community will need access to the most powerful available browsers, search engines, selection and filtering software, and so on. Even more important, as producers, they will need convenient server access and authoring tools that make it very easy to capture material, format it, apply any desired access controls, and mount it online, under security if appropriate.
So far, educational material informally mounted on the Web has mainly consisted of text and scanned images. In the future, however, the growing sophistication of the Web environment, increased server capacity, and increased bandwidth should allow a much wider range of materials to be captured, preserved, and distributed in this way—in particular, videos of lectures, laboratory demonstrations, and industrial processes, simulations and animations, and interactive games and virtual experiments. This will create a demand for widely distributed and easy-to-use videorecording capabilities, for convenient editing and authoring tools, for server capacity on a much larger scale (plus the ability to backup the servers), and for enough bandwidth to move this material around without diminishing its quality or generating unacceptable delays. If this demand can be satisfied, the payoff will be a growing informal archive that allows students to resolve schedule clashes by downloading videos of lectures and other events that they cannot attend live, that allows revisiting important experiences, that functions as an ongoing history of MIT's day-to-day intellectual activity, and that provides a source to mine for information and inspiration.
Primary Sources
A key purpose of the Web, as it was originally conceived, was to provide convenient online access to primary scientific data. This remains important to researchers, and as Web access becomes ubiquitous, it can play an increasingly prominent educational role. Instead of relying on limited textbook examples, instructors can—where appropriate—illustrate concepts by operating on archives of primary data, and set students to work on problems and projects based on these data. This is a potentially effective way of bringing the excitement of the research lab into the classroom.
These primary sources can take different forms in different disciplines—experimental data in the physical and life sciences, statistical data in the social sciences, the U.S. National Library of Medicine's Visible Human Dataset, satellite images and GIS databases in urban planning and civil engineering, comprehensive text and document libraries such as the TLG in the humanities, online slide libraries in art and architectural history, film libraries for theater and film studies, libraries of CAD models in architectural and mechanical engineering design, and so on.
Where these primary databases are relatively small, where they do not require a lot of ongoing maintenance, and where the intellectual property issues can be resolved, they can be maintained on servers at MIT. Where these conditions are not met, these databases are more likely to be maintained at remote sites—typically those of the original producers or of specialist maintenance organizations. To make use of them, we will need to have effective ways of identifying members of our community, managing our pointers (something better than hotlists of URLs) and having connections fast enough to make convenient remote access possible—a particular problem in the case of huge, graphics-intensive databases such as the Visible Human.
Formally Published Teaching Materials
Universities consume prodigious quantities of teaching materials. In the past, they have often taken the form of printed textbooks, monographs, compilations, handbooks, and reference works. In recent years, so-called course packs—customized readers compiled from multiple published sources—have become increasingly popular with faculty. Students have purchased their own copies, consulted copies in the library, and/or borrowed copies from the library. To a lesser extent, there have been instructional films, videos, interactive CD-ROM and audio recordings; these have had particularly important niches in professional and extended education, in teaching foreign languages, and in music and theater. There is growing interest in publishing teaching materials in the form of CD and online multimedia productions. It is too soon to say how rapidly, and to what extent, digital multimedia instructional materials will supplant print-on-paper, but there is little doubt that they will ultimately play an extremely important role.
Universities also participate extensively in the production of teaching materials; faculty members write and consult on textbooks, contribute to reference works, write monographs, and sometimes participate in film, video, and multimedia educational productions. Over time, fairly standard arrangements have been worked out for doing this. Typically, faculty members write the material and publishers provide editorial and design services, manage production, and handle distribution and sales. The publishers create brand names, take most of the risk, and get most of the profit. Authors get royalties. In most (although not all) cases, universities do not directly profit at all.
With the shift to digital media, we must consider the question of whether the traditional arrangements still make sense. Should we attempt to take a direct role in financing, producing, and distributing digital multimedia productions? Should we create the multimedia equivalent of a university press? Should we form strategic alliances with major media corporations to produce and distribute ambitious digital multimedia productions? If we were to form such alliances, who would control the resulting intellectual property, and how would faculty members be rewarded for their participation? Should we become a national and international educational materials brand name? Can we reduce costs to students and to libraries by taking a production and distribution role?
It is unlikely that we will find easy or quick answers to these questions. But it does seem clear that an increasing number of faculty members will want to be involved in producing digital multimedia educational materials. They will need more than desks and word processors for this. They will require facilities, tools, and support staff to shoot video, record audio, produce sophisticated graphics—both still and motion—create storyboards, and edit multimedia material. If we do not want all of this activity to shift to off-campus studio and postproduction locations, we will need to provide some level of professional-level facilities on campus or nearby, and we will need to generate income to pay for these facilities.
Refereed Scholarly and Research Journals
Refereed scholarly and research journals have traditionally performed the tasks of disseminating original work produced by faculty members and graduate students and of creating permanent archives of this work. Editors, editorial boards, and publishers take responsibility for the quality of content—thus providing some level of guarantee that this content is reliable and useful. Because contents are refereed, and space is usually scarce, publication records in respected refereed journals play a crucial role in promotion and tenure processes.
Journals are essential, but the problems of traditional print journals are notorious and increasingly troublesome. The process of getting a journal paper into print is usually a slow one, which creates a tremendous problem in fast-moving fields—so much so that these fields tend to find alternative distribution channels. Subscriptions are increasingly expensive, which is creating a crisis for libraries as, simultaneously, journals proliferate and budgets diminish. And from a research university's viewpoint, the business model seems wrong; we pay faculty members, who write papers and provide them at no cost to journal publishers, who then sell them back to our libraries at very high cost!
Online journals potentially provide an extremely attractive way of overcoming many of these problems. Although they may do little to speed the refereeing process, they eliminate many steps in the production process, allow paper-by-paper rather than issue-by-issue production, make distribution much quicker and easier, and reduce the need for shelf space. Thus, in principle, they can provide a quicker and cheaper means of distribution than print. Furthermore, they lend themselves to efficient, automatic indexing and searching, they allow cross-linkage of content, and they can potentially be accessed from any workstation anywhere—they are never checked out.
Given that libraries are highly motivated to find ways to overcome their problems with journals, and given that journal publishers are highly motivated to find ways to survive in the digital electronic era, it seems likely that we will eventually see a massive shift away from paper and toward the online distribution of articles. But some difficult issues will have to be settled first. Libraries with major journal collections will have to attend to new sets of issues: to ensure that electronic archives of back issues are secure, stable, and affordable; to raise the alarm if the rights of educational fair use in the classroom and for nonprofit research are threatened in the digital environment; to manage pointers and finding aids rather than bulky inventories of physical objects. The functions of editing, designing, refereeing, and marketing will still have to be performed, and publishers will have to find ways of paying for these functions out of reduced gross revenues. Servers will have to be managed—either by publishers or libraries—and this will require skills not traditionally possessed by these organizations.
Whatever the difficulties, online journals are already appearing. They will be an important part of our future research and teaching activities, and we must have effective ways to contribute material to them, manage them, and access their contents.
Role of the MIT Press
The MIT Press is MIT's principal (although by no means only) publishing arm. Traditionally, it has published monographs, reference works, journals, and trade books on subjects related to the core concerns of MIT. It has particularly strong lists in architecture and visual arts, economics, linguistics, cognitive science, and computer science. It has wide distribution and a well-known, highly respected, international brand name.
Among the leading university presses, MIT has probably been the most aggressive and innovative in pursuing the possibilities of online electronic publishing. It has pioneered the publication of online books with City of Bits, Moths to the Flame, and Hal's Legacy. It has a growing list of online journals, including The Chicago Journal of Theoretical Computer Science and Journal of Contemporary Neurology. And it is embarking on a very ambitious project to develop CogNet—a comprehensive Web site to serve the cognitive science community.
The MIT Press has growing experience and expertise in electronic publishing, a developed marketing capability, and a valuable brand name under which to market electronic as well as traditional products. Its management has expressed eagerness to participate actively in Institute-wide initiatives to take the lead in applying advanced educational technology, and the press is very well positioned to do so.
Role of the Center for Advanced Educational Services
The Center for Advanced Educational Services has its roots in video and multimedia, and in extended education, rather than print publication for the research and scholarly communities. As such, it brings a different and complementary set of capabilities to the task of capturing, managing, and distributing MIT's intellectual property.
CAES defines its role as one of creating and distributing educational products and services worldwide. It focuses its production and distribution efforts on interactive multimedia, the Internet, the Web, videoconferencing, satellite TV, and MIT Cable, as well as more mature delivery mechanisms such as videotapes and books. It has studio and editing facilities and associated staff. It incorporates an applied research arm, the Center for Educational Computing Initiatives (CECI) that investigates the uses of new technologies in education. And it regularly offers nondegree short courses to industry audiences.
CAES has MIT's largest concentration of video and multimedia production facilities and staff, it is modernizing these facilities, and it has an entrepreneurial mission to provide these facilities to the MIT community and to promote distance learning.
The MIT libraries envision a future world of scholarship and research that provides seamless access to the world's relevant literature and data from any individual's desktop. While faculty at MIT are deeply knowledgeable about the sources and forms of scholarly information of their own disciplines, and indeed will usually have built personal libraries and human networks of unique and awesome strength, interdisciplinary interests often require that faculty delve into areas of information with which they may not be so familiar. Students and junior faculty will lack the in-depth personal libraries and contacts of senior faculty, and will inevitably rely on the judgment and expertise of others as they gather and evaluate needed information. It is important to remember that any individual's need for and use of information will be a function of the dominant literature of the discipline, the nature of the research specialty, career stage and research training, access to "local" resources (this once meant physically local, but it now means accessible locally—as in full-text databases), and availability of human helpers and agents. The personal productivity of individual faculty and enhanced productivity of research teams remains the driving force behind the demand for locally accessible information resources.
The availability of a world-class networked environment, especially one that readily supports a wide variety of formats, platforms, and media types, is essential to the fulfillment of a vision of enhanced local access to resources in many media. The successful execution of the vision will require MIT to address a number of difficult issues. These issues are common to all large, complex organizations that face the prospect of integrating the digital world into their traditional ways of managing and organizing information. Obviously, any activity in the areas discussed below is as useful in a distance learning or remote campus environment as it is here in Cambridge.
The libraries will focus on four critical issues in this new world:
1. Integrated content. Knowledgeable individuals who are familiar with the quality and substantive coverage of Web sites, databases, and other forms of digitized data will have to identify useful resources, negotiate favorable licenses for such resources where necessary, create meaningful front-end menus, and otherwise facilitate access to the information of disciplines and/or research interest communities. For the near future, researchers will work and students will learn in an environment of mixed media. The large-scale retrospective conversion of the print record in any area other than high-value serial or report literature will not be financially feasible for many years, if ever. New methods and approaches to revealing and accessing printed material must be developed to integrate information sources formatted in traditional media with sources in the digital environment if the value and utility of these resources is to continue. In many fields, it is difficult to imagine credible work that excludes all work older or other than that which has been digitized. This is as true for corporate intellectual property as it is for scholarly information.
2. Skilled support. Data and information available on the network frequently assume a preexisting level of subject familiarity, intellectual understanding, and/or technical ability that the student or researcher may not have—particularly if the area of interest is outside an individual's primary discipline. For example, locating and downloading numeric data, identifying and viewing medical research results, or locating and rotating molecular structures all require knowledge of the subject, the database structure, and the downloading environment. New forms of network-based training and support to facilitate the cost effective use of new media must be developed. Computer-based "helpers," Web-based FAQs, and electronic "librarians" are among the possibilities to explore. Companies as well as educational institutions would benefit from the creation of new models of skilled support for the networked environment.
3. New media. Photographic images, music, computer models, video clips, and other media amenable to digital management and network distribution will become a standard aspect of "library collections." Whether through creation, conversion, pointing, purchasing or licensing, the financial and service advantages of assigning responsibility for the appropriate management and leveraged reuse of acquired (especially purchased) new media information assets will need to be addressed. Organizational techniques for identifying new media by relevant retrievable fields must be developed for the network environment if the investment in technology is to have broad application and result in products that can be leveraged across the institution. For example, a student might be able to search an image file to retrieve the architectural drawings of two concert halls, then search an audio file for sounds recorded from different locations in the two halls. Another aspect of new media relevant to the libraries is the formats used in theses. The libraries are currently launching a pilot project to archive MIT theses in digital format. Digital sound, images, and so on could potentially become standard features of MIT theses. A network that supports the authoring and production of multimedia theses would be essential.
4. Curriculum support. Whether on or off campus, technical or artistic, faculty are exploring new ways of conceptualizing and teaching courses. From modest beginnings in "electronic reserves" to more sophisticated support using advanced Web technologies, libraries continue to develop new approaches to supporting the educational process. A ubiquitous, functional, broadband network is essential to these initiatives. A second aspect of curriculum support essential to the digital future is the development of a set of core competencies that all educated individuals should expect to possess. A basic familiarity with bibliographic research techniques was once assumed to be an attribute of the properly educated college graduate. MIT can utilize the availability of a high-quality network to define and ensure equivalent expectations for graduates who will spend their adult work lives in a digitally formatted and network-based work environment. If the new communications requirement for undergraduate students advances on schedule, many additional opportunities for network-based curriculum support will be presented. One can envision supporting such a curriculum with videos "on reserve," network-based support for assignments on a "just-in-time" basis, and course-specific, multimedia home pages.
Each of these four issues has broad interest outside MIT. In addition, industry and higher education alike are intensely interested in understanding how the economics of information change in these new environments, and how the learning experience itself benefits from new technologies.