Collaboratories: Doing Science On The Internet

Richard T. Kouzes West Virginia University James D. Myers Pacific Northwest National Laboratory William A. Wulf University of Virginia The success of many complex scientific investigations hinges on bringing the capabilities of diverse individuals from multiple institutions together with state-of-the-art instrumentation.

We are all aware of the tremendous impact computers have had on science and engineering in the past 50 years, but the impact in the near future may be far greater. A 1993 National Research Council study suggested that

The fusion of computers and electronic communications has the potential to dramatically enhance the output and productivity of U.S. researchers. A major step toward realizing that potential can come from combining the interests of the scientific community at large with those of the computer science and engineering community to create integrated, tool-oriented computing and communications systems to support scientific collaboration. Such systems can be called "collaboratories."¹

...center without walls, in which the nation`s researchers can perform their research without regard to geographical location - interacting with colleagues, accessing instrumentation, sharing data and computational resource, and accessing information in digital libraries.

COLLABORATORY PROTOTYPES
    To facilitate scientific work, collaboratory systems must support the sharing of secure data, analysis, instruments, and interaction spaces.^3,4 Several systems incorporate these basic components.⁵
    An advanced example of a data-driven collaboration is the Worm Community System, one of the original collaboratory projects sponsored by the NSF in 1990.⁶ This system supports researchers studying the nematode C. elegans, which is a harmless, soil-residing worm of little human significance. The Worm Community System provides a repository for data about the nematode, ranging from genome to behavior level, and ties this data to the literature. Everything known about C. elegans and everyone contributing to this knowledge is accessible through the system. These capabilities elevate the Worm Community System from a simple tool for sharing data to an electronic forum.
    Providing access to scientific instruments from distant locations is another common focus of collaboratories. Early collaboratories focused on the sharing of large, expensive instruments such as astronomical telescopes, particle accelerators, oceanographic instruments, atmospheric observatories, and space research applications. The Upper Atmospheric Research Collaboratory, another NSF-funded project, is an example. UARC provides six institutions access to instruments in Greenland for solar wind observation. (For more information, see http://www.si.umich.edu/UARC/HomePage.html). UARC collaborators exchange and archive multimedia information from the instruments and the measurements analysis. Other collaboratory projects share smaller devices, such as electron microscopes, scanning tunneling microscopes, and nuclear magnetic resonance instruments.
    Developing a shared interaction space across several laboratories is the ambitious goal of a new initiative from the US Department of Energy. The DOE's Distributed Collaboratory Experiment Environments Program involves four major projects, briefly described in the "Distributed Collaboratory Experiment Environments" sidebar. These are the first efforts of a major program to develop the technology for a virtual laboratory system that encompasses the scientific resources of the national laboratory system. The goal is to enable scientists around the world to participate in solving DOE`s science and technology challenges.

SOCIOLOGY OF COLLABORATION
    Facilitating collaboration among a widely distributed scientific community is highly complex. although a collaboratory is potentially nothing less than the village square of the Information Age, it is a synthetic place requiring social adaptation.
    Is such a place socially sustainable? The requirements for a technological system's success seem contrary to the sentiment expressed in the motto for the 1933 Chicago World`s Fair: "Science finds, industry applies, man conforms." Today we expect technology to adapt to the user. Some people contend that "technology is incompatible with a gentle and humane society,"⁷ but we believe that technology implemented with an awareness of human needs can facilitate the collaborative process.
    Asserting the social acceptability of a synthetic "place" does not make it so, of course. Thus, in addition to purely technical issues, the research agenda for creating collaboratories must address fundamental psychosocial questions as well. Is it possible to electronically create a suitable sense of place that permits, and enhances, the successful cooperation of dispersed individuals toward common goals? How can we support communication that permit human cooperation even when the evolutionary social mechanisms that depend on proximity are absent?

Asserting the social acceptability of a synthetic "place" does not make it so, of course. Thus, in addition to purely technical issues, the research agenda for creating collaboratories must address fundamental psychosocial questions as well.

The silent language of body motion and spatial position are central to human communications and social control; can this richness of human interaction be provided? Indeed, which aspects of this richness are critical and which are not?
    Collaboratory developers must consider psychosocial issues such as autonomy, trust, sense of place, and attention to ritual. Autonomy, which describes how an organization is governed or regulated, is implemented through informal communications, acquaintances, and associations. Collaboratory developers must embed autonomy into the virtual organization in a considered manner. Trust, which is established among collaborators through shared experience, is implemented over time through informal means such as meeting face to face and working together in the same place. In a collaboratory, trust will have to be established "through some special means. A sense of place, which allows people to feel comfortable in their surroundings, provides security so that people can feel creative. If a collaboratory can harness some of the design strategies that have been so successful in physical group settings, it can also create a sense of place and purpose among its dispersed members that will engender an enduring sense of affiliation and cooperation toward its goals. The mechanisms of ritual, which moderate our interpersonal interactions, must find a place in the synthetic surroundings of a collaboratory.
    Technology solutions abound, but often fail to find a human problem to solve. Groupware applications, like those for meeting scheduling, group decision support, joint authorship, and distributed management, have had mixed success. The failure of groupware to gain wider acceptance is due to its primitive technology and its insensitivity to social and political issues in the workplace. Groupware applications for a collaboratory will have to be selected and implemented with a clear understanding of the social and political concerns that characterize joint scientific work. Among these are issues of authorship, acknowledgement of contributions, esteem of peers, and recognition by professional role models. Without such characteristics, they will not find acceptance!

COLLABORATION TYPES
    We can learn much about the process of scientific collaboration from discussions with scientists themselves. As part of the development of DOE's Environmental Molecular Sciences Laboratory project, researchers were asked about the nature of their current and future collaborations in order to understand what types of communications an electronic collaborative environment must support. Scientific collaborations span a wide range in terms of group size, collaboration style, and focus (experimental, theoretical, computational). The focus at EMSL is basic scientific research, undertaken by as many as 300 researchers.

Distributed Collaboratory Experiment Environments     The US Department of Energy has initiated a series of four major collaboratory projects known as the Distributed Collaboratory Experiment Environments Program.     Argonne National Laboratory and Northeastern University are building and testing LabSpace: A National Electronic Laboratory Infrastructure. They are implementing a shared space with persistence and history in two application testbeds. The first is the Telepresence Electron Microscopy project, which is designed to allow the remote use of the Advanced Analytical Electron Microscope and the Analytical Scanning Electron Microscope. The second involves a collaboration with CERN, the European high-energy physics center, which will exercise the LabSpace version of a collaboratory in a larger collaboration over an international data link.     Lawrence Livermore National Laboratory, Oak Ridge National Laboratory, the Princeton Plasma Physics Laboratory, and General Atomics are pursuing the Distributed Computing Testbed for a Remote Experimental Environment. The experiment focuses on fusion energy R&D, an archetype of research that must be carried out at a few large central facilities with national and international participation. The experiments also are designed for steady-state operation, for which interactive, real-time experimentation becomes important. Remote operations will be conducted at the D-IIID tokamak fusion facility, a fusion research instrument located at General Atomics, San Diego, California. This project demands not only real-time synchronization and exchange of data among multiple computer networks, but also the presentation of sufficient auditory and visual information associated with the control-room environment so that remote staff at multiple sites can be fully integrated in operations.     Pacific Northwest National Laboratory is working on Collaboratory Development in the Environmental and Molecular Sciences. The Core testbed (described in the main text) is based on instrumentation being developed for the Environmental Molecular Sciences Laboratory project at PNNL. This includes two unique nuclear magnetic resonance spectrometers, which are large, highly shared items, and some small instruments used by a limited number of researchers in molecular-beam reaction dynamics. Thus the characteristics of two related yet distinct scientific cultures, working with two quite different kinds of machines, are being examined.     The University of Wisconsin-Milwaukee will remotely operate a sophisticated synchrotron-radiation beamline in The Spectro Microscopy Laboratory at the Advanced Light Source project. This collaboratory development project will provide remote access to three analytical tools at the Advanced Light Source located at Lawrence Berkeley National Laboratory that provide spatially resolved chemical information at length scales ranging from one micron down to the atomic scale. The collaboration that uses these instruments is fairly large and geographically distributed, with investigators from nine institutions, so the potential for saving the time and expense of training, staffing, and travel is considerable. The ongoing growth trend of synchrotron-radiation applications will provide a large and welcoming audience for the results. One particularly interesting target audience for remote usage of this and similar facilities is the semiconductor industry, which has a critical need for sample inspection.     For more information on these projects, see http://www-itg.lbl.gov/~jtcjew/DCEE_Overview.html.

    On the basis of this feedback, we identified four broad categories:

• Peer-to-peer. Some collaborations involve researchers in the same field sharing an instrument. for example, a remote researcher might contribute to the design of a new detector and then use the instrument to study systems of interest. In this type of collaboration, the researchers share a common scientific vocabulary. The most important aspects of the collaboration are shared instruments and raw, unanalyzed data. Remote instrument control and direct data access important in this type of collaboration.
• Mentor-student. Some collaborations involve senior scientists and their more junior partners, such as students and postdoctoral fellows. In these collaborations, the mentor may use prepared materials and live demonstrations to teach data acquisition, analysis techniques, and scientific principles. The mentor must then observe as the student demonstrates mastery of the new concepts by using them appropriately. The necessary real-time interactions between mentor and student go far beyond standard conferencing: A mentor and student must be able to work collaboratively and interactively. In this type of collaboration, access to many types of archival information - data, notes, results, and so on - is also important so that the student can revisit the material.
• Interdisciplinary. Some collaborations involve scientists who are doing complementary studies of the same system. For instance, a theorist may calculate molecular structures of clusters while an experimentalist uses laser spectroscopy to make experimental structure measurements. Researchers in such collaborations may not share a common vocabulary and so must often translate their results into each other`s terms. Here, researchers might alternate between the roles of mentor and student as they seek to synthesize their findings. In these types of collaborations, it is less important to provide direct access to instruments and raw data; more important is access to summaries and analyses, perhaps recorded into an electronic notebook, and support for the discussion of unfamiliar concepts so that misunderstandings can be corrected.
• Producer-consumer. Some collaborations also involve researchers in different disciplines, but in this type one researcher or research team provides input to another. For example, a mass spectroscopist might determine the sequence of a protein for a biologist, or a surface scientist might provide reaction rate data to a geologist who wants to model the subsurface transport of hazardous wastes. An extreme form of this collaboration type involves an analytical laboratory that works on a fee-per-service basis. Here there is often a wider gap between the disciplines and motivations of researchers: A scientist may want to study a new physical phenomenon, while the engineering collaborator wants to reduce the cost of a clean-up effort. Collaborators in this type of relationship have little chance for professional contact. They place the most importance on being able to receive a sample and transmit results to the other party. However, if these researchers can communicate more closely, they might be able to generate new ideas and approaches. To foster close communication among basic and applied scientists, laboratories hold seminar series, workshops, and pizza dinners. This approach suggests that such collaborations may become more complementary when researchers are provided with readily available tools for electronic discussions.

    It is important to note that although we present these classifications and examples as distinct types, a single collaboration may actually contain elements from several styles, either in parallel or as the collaboration evolves. Nevertheless, these categories do help to show the varying communications needs researchers have as they work in different modes and how an individual`s needs may change as the task or nature of the collaboration changes. The fact that researchers may switch collaboration styles frequently as they work through various tasks in an experiment implies that an electronic collaboratory environment should not impose a particular mode. It should instead provide a wide range of capabilities that can be quickly and easily selected and configured for the task at hand. Such flexibility addresses some of the social barriers inhibiting collaboration.

FIGURE 1. Tools provide varying functionality. Some are synchronous, while others are asynchronous. Some work well for more static applications, while some are inherently dynamic.

COLLABORATION TECHNOLOGY
    Electronic collaboration must occur in an environment that lets collaborators work intimately with one another.
    Current implementations use an integrated set of cross-platform tools such as electronic notebooks, video-conferencing systems, electronic whiteboards, shared screens, information-access tools, and instrument-control tools. Figure 1 illustrates how different tools provide varying functionality in interactions depending upon the static or dynamic nature of the information exchange as well as upon the synchronous or asynchronous nature of the session.
    Electronic mail supports collaboration via a time serial dialog. Videoconferencing supports real-time discussion and, with the addition of graphics and whiteboard capabilities, presentation and brainstorming.
    Because the collaboratory concept brings all the scientific resources used by researchers into the mix, both real-time work and asynchronous collaboration are possible. The effect of having all scientific resources available to all researchers moves a remote collaborator from the role of part-time consultant to co-worker.
    Full support for this vision requires substantial additional work, but progress is being made on prototype systems and on the robust, secure, scaleable architecture required for production systems. Network infrastructure is vital for supporting collaboratory-style interaction and linking the high-performance computing system, experimental equipment, data-acquisition systems, and the scientist's desktop workstation into a unified research tool.
    As part of the DCEE program, the EMSL has developed Core, a prototype collaboratory that provides a loosely integrated set of Internet capabilities that appear as extensions to the Web. Core provides a one-click method to start or join multi-tool collaborative sessions from a Web page. The Core Session Manager and Desktop Executive launch and track active sessions, participants, and tools, letting users pick capabilities appropriate for their work without having to be aware of the connection syntax of individual tools, port numbers, firewalls, or Internet addresses.
    Core and the tools it uses are under development at PNNL and other national laboratories and universities. The tools have been chosen because they provide a wide range of cross-platform functions that let researchers interact with remote colleagues in a rich, in-process, style. The tools include:

• Audio/video conferencing. Collaborators can see and hear each other and monitor instruments and laboratories. The main tool used is network-based software such as the Mbone.

  FIGURE 2. A sample electronic notebook page.

• Chatting. Collaborators can exchange text messages.
• Shared computer display and whiteboard. Collaborators can view and interact with any program running on the shared display. They can also use whiteboard-style annotation on top of a live image and remotely control the shared application. This tool turns noncollaborative applications into collaborative ones. Electronic whiteboard functions are also provided.
• Shared electronic notebook. Collaborators can share this electronic version of a traditional paper laboratory notebook. Electronic notebooks provide distributed access to data, as well as automated data entry, searching, and other information processing not possible in a paper notebook. The current EMSL notebook creates a dynamic Web page that can be queried, display experiment information, and accept multimedia user annotations. The Web page provides the data files, an image, or live Java-based graphical summary of the data in each file, and information about each file (such as instrument parameters used, operator`s name, date). Information from instruments that is sent to the enterprise database/archive system is queried to provide automatic updates to the electronic notebook. Users may query to select a subset of files to be displayed by sample, date, and owner name. They can also easily add text, picture, and file annotations to the original information. Figure 2 shows a sample notebook page.
• File sharing. Multiple collaborators can transfer files with a drag-and-drop facility.
• On-line instruments, computation, and visualization. Data acquisition, analysis, computation, and visualization software written for a single local user can be modified for collaborative use in an environment such as Core. One of the first on-line instruments in the EMSL, a radio frequency ion-trap mass spectrometer, can be launched via Core and saves data directly to the electronic notebook. The instrument software also includes pan and tilt controls for a laboratory video camera that can be included in a videoconference. Similarly, anaylsis, modeling, and other types of scientific applications can be enhanced to let multiple users to simultaneously view and interact with them.
• Web browser synchronization. Collaborators can use Web material for their lectures or discussions. When one user goes to a new URL, all linked browsers automatically follow. Lecture (only the leader`s browser is echoed) or discussion (peer-to-peer) modes are possible.

FIGURE 3. A collaboratory software environment uses software tools such as whiteboards, electronic notebooks, chat boxes, televiewers, information browsers, and videoconferencing to facilitate effective interactions between dispersed scientists. This present tools set, a realization using today's applications, is rapidly evolving as more is being learned about facilitating the collaborative process.

BARRIERS TO ADOPTION
    The barriers to implementing these environments are both technical and sociological. Existing tools are often immature, unintegrated, hard to support, and costly to maintain.
    The adage "build it and they will come" is disproved daily in the computing industry. A technology often exists without being used because it is perceived as adding little or no value. While we contend that the time is right for a collaboratory solution to the needs of scientific interaction, we are challenged to make it a viable necessity for scientific progress, as well as psychosocially acceptable.
    We can make glowing predictions about the value collaboratories bring to scientific inquiry; we can make equally valid projections of why they will fail. We know how difficult it is to see the future clearly. We remember what the father of radio, Lee De Forest, said about television in 1926: "While theoretically and technically television may be feasible, commercially and financially I consider it an impossibility, a development of which we need waste little time dreaming."
    Videoconferencing has been slow to catch on partly because of its cost, hardware restrictions, lack of standards, and poor audio and video quality.⁸ The perceived benefit of videoconferencing is not sufficient to overcome the problems of using available systems.⁹
    Mbone-based freeware applications, which we use in several DOE projects for videoconferencing, originated only in 1992.¹⁰ Mbone is now used extensively by a small class of Internet users for videoconferencing and broadcasts. The Unix-based Mbone video applications provide frame rates of only a few per second, while consuming about 200 Kbytes/sec of network bandwidth. Mbone is now providing one-way videoconferencing connectivity to Mac and PC platforms. Cross-platform whiteboards, electronic notebooks, and shared screens use less bandwidth than video but remain in an early state of development, especially with regard to interoperability.
    Collaboratory tools need several more years of research until they will be mature enough to be acceptable to end users.

WE PREDICT THAT COLLABORATORIES will be part of our future, that they will be rich telepresent environments, and that virtual laboratories will proliferate. We also conjecture that complexity will drive the need for increased collaboration in scientific endeavors and new research funding models. As educators, we will be faced with training our students to work in a multidisciplinary world of complex problems, which will cause us to adopt new educational strategies. The collaboratory concept is a qualitatively different way of using communication and information technologies. It has the potential to remove the walls around departments and organizations, and it will lead to the creation of a metalaboratory with capabilities that far exceed those available in any single laboratory. In the next few years, a growing community of scientists from multiple universities and national laboratories will be conducting serious scientific research in cyberspace.

Acknowledgements

References

1. V.G. Cerf et al., National Collaboratories: Applying Information Technologies for Scientific Research, National Academy Press: Washington, D.C., 1993.
2. P.S. Malm, "The unOfficial Yellow Pages of CSCW, Groupware, Prototypes, and Projects," in Classification of Cooperative Systems from a Technological Perspective, Groupware in Local Government Administration, doctoral disseration, University of Tromso, Norway, 1994.
3. R.T. Kouzes, "Creating the Cyberspace Laboratory," in The World & I, The Washington Times Corp., Washington, D.C., 1995, pp. 190-197.
4. R. Pool et al., "Beyond Databases and E-Mail," Science, Aug. 1993, pp. 841-872.
5. R.T. Kouzes, "The Collaboratory: Creating R&D Laboratories Without Walls," in Electronic Laboratory Notebooks and Collaborative Computing in R&D: Social, Legal, Regulatory, and Technology Issues, Rich Lysakowski, ed., Team Science Publishing, Sudbury, Mass., 1996, to appear.
6. B.R. Schatz, "Building an Electronic Community System," J. Management & Information Systems, 1992.
7. G. Levy, " The Gods Must Be Crazy," American Laboratory, Aug. 1995, pp. 6-8.
8. E. Garland and D. Rowell, "Face-to-Face Collaboration," Byte,Nov. 1994, pp. 233-242.
9. W. Machrone, "Seeing Is Almost Believing," PC Magazine, June 14, 1994, pp 233-251.
10. H. Eriksson, "MBONE: The Multicast Backbone," Comm. of ACM, Aug. 1994, pp. 54-60.

Richard T. Kouzes is the director of program development for science and engineering, and professor of physics, at West Virginia University. He has worked in the field of collaborative computing and is a principal investigator on a DOE Distributed Collaborative Experiment Environment project. He is the originator of the PNNL collaboratory effort. He has been a leader in the field of computerized data acquisition and was a founder and past chair of the IEEE Committee for Computer Applications in Nuclear and Plasma Sciences. Kouzes received a BS in physics from Michigan State University and an MS and a PhD in physics from Princeton University.

James D. Myers is a senior research scientist in the Computing and Information Sciences department of the Environmental Molecular Sciences Laboratory at PNNL, the EMSL Collaboratory project leader and a principal investigator on a DOE Distributed Collaborative Experiment Environment project, working to design, develop, deploy, and understand the use of the EMSL Collaborative Research Environment software. He is the recipient of a 1996 Associated Western Universities Distinguished Lectureship for his work in promoting the collaboratory concept for scientific research, undergraduate research and education, and K-12 education. Myers received a BA in physics from Cornell University and a PhD in chemistry from the University of California, Berkeley.

William A. Wulf is the AT&T professor of engineering and applied science at the University of Virginia, where he is revising the undergraduate computer science curriculum, researching computer architecture and computer security, and assisting scholars in the humanities to exploit information technology. Wulf chairs the Computer Science and Telecommunications Board of the National Research Council and is the interim president of the National Academy of Engineering.

Contact Kouzes at rkouzes@wvnvm.wvnet.edu; Myers at jim.myers@pnl.gov; and Wulf at wulf@viper.cs.virginia.edu.