Investigators today face new dimensions in information dissemination. It is no longer acceptable to delay the publication of important new research findings until they can be published in printed journals. For the same reason, it is no longer acceptable to rely solely on printed journals for the latest in scientific findings. The Internet has created a new world of research in two important ways: by reducing the time required for new findings to be made available, and by minimizing the separation of geographic distance to a minor inconvenience in collaborative research. In this new world, scientists must develop new tools to enhance the opportunities afforded by this collapse of the time and space dimensions to an electronic bubble. We are learning how to do so, but we have not yet fully exploited the opportunities available to us.
This paper describes a tool developed originally to enhance learning in all forms, from lecture to distance learning. As we developed these tools, faculty in all disciplines pointed out that the tools developed to enhance interaction between learners and instructors were in fact the very tools needed to facilitate collaborative research. We therefore describe our tool, which we name Remote Technical Assistance (RTA for short) in its current form and discuss ways in which it can be applied to collaborative research among distributed partners in scientific investigations.
As we reflect on the way in which our own research in recent years has been affected by the Internet, we recognize a number of changes in the patterns of our investigation. Taken together, these changes present opportunities and challenges which today's technology can most certainly meet, but the needs will not be met without careful planning. Our research design must begin with the question "what is needed to enhance distributed collaborative research?" rather than the more often posed question, "How can technology serve?" By reversing the forcing function, we are implicitly stating that, once conceived, technology can indeed meet the requirements imposed, but if we instead start with a new tool and look for a solution (as some say holography has been abused), we end up giving pre-eminence to the technology rather than the underlying needs of the researchers.
From this initial assumption, we select the following design criteria as the basis for implementing new support tools for distributed research.
* Interaction between humans has always been the greatest missing element in collaborative research. Ways must be found to enhance human interaction. Further, that interaction should be available to pairs or to groups, without regard to their geographic location with respect to each other.
* That interaction should allow the most natural forms of communication between distributed researchers. It should permit easy incorporation of written text, spoken words, and images drawn from all manner of sources. It should allow these objects to be annotated and manipulated as though the researchers were physically together in the same facility.
*There should be no limitations imposed on forms of notation in this interaction. In text, one should be able to use complex scientific notation as one would with pencil and paper. Images should be capable of being rendered and analyzed as they are using modern filters and other investigative tools, and those results available in timely manner for collaborative study.
* That interaction should also recognize the need for both synchronous and asynchronous forms of communication, linking the two in a continuum that is both seamless and easy to use and switch back and forth.
* A record should be kept of the interaction, so that the researchers can return to those records as they might to a laboratory notebook to review findings, suggestions, and conclusions reached along the way.
* The medium by which the interaction is accomplished should be insofar as possible independent of the specific hardware resources of the investigators.
* The tools developed should be malleable, capable of adjusting to researcher's needs in different forms of investigation, emphasizing those components favored by one group, and minimizing features that, while attractive, have little bearing on that particular investigation.
* The management of this tool must be straightforward for investigators whose expertise lies in fields other than computers.
* Finally, the system should be open-ended, allowing active use to pose new criteria for inclusion in creation of the ultimate tool.
These, we submit, are valid, though perhaps incomplete, criteria for design of a collaborative support tool.
Our particular research has focused in recent years on technological support of the learning process in all forms. Our analysis of the learning process derived a nearly identical set of criteria for the needs of the distance learner. To the list above, we added the need for learners to interact, not only with their instructors, but also with other learners. In fact, the systematic exclusion of human interaction of all forms in the learning process has plagued the learning process throughout history. (A student recently remarked, "If you want to know what distance learning is, try sitting in the back row of a lecture hall seating 450 students.") The remaining components are, however, common both to learning and to research, which makes sense when we consider that novel research is inherently a learning process.
A secondary but important criterion emerging from instructional use of this tool was upwards scalability. While the research domain is most likely to involve small groups, instructional needs today may well require interaction involving hundreds or even thousands of learners. The tool must be capable of serving groups of this size.
Given these criteria, we embarked over two years ago on the development of a tool that would meet as many of these criteria as possible and at the same time be open-ended to the inclusion of new needs that we had not yet envisioned. Our work was done initially only with support from our institution, the University of California, who owns the rights to this product in its entirety. Subsequently we were able to obtain a three-year grant from the Fund for Improvement of Pos-Secondary Education, which has expanded the research effort without changing the ownership.
We called our project Remote Technical Assistance, of RTA, rejecting the term Remote Teaching Assistant because it was too limiting in scope. We recognized from the outset that RTA could serve equally well for learning, remote consultations, or collaborative scientific investigations.
RTA has evolved significantly over the past two years and is still in its infancy in terms of its full potential and its use in various forms of human interaction. It is now at the point, however, where it can serve as an effective tool even as its capabilities are being extended to new features and new applications. The principal design components that resulted from our consideration of the needs analysis cited above are the following:
* RTA was conceived as a platform-independent system. The minimal requirements are access to the Internet, the availability of a server, and the availability of clients that can connect remotely to that server.
* RTA would be multimedia, allowing for text, images, sound, and all forms of resource sharing in an interactive environment.
* RTA would be a moderated interaction, with communication going between end users through a communications system that would store all necessary components of the interchange and at the same time allow systems analysis of the interactive process.
* RTA would be open-ended, incorporating functionality central to its interactive purposes while providing transparent links to other resources and other functionality.
These were the most important components of our design. As the tool took form, they translated into more specific decisions, and these decisions let in term to solutions that themselves posed new questions and offered new opportunities. As we began to share our tool with others, we immediately learned that each new demonstration of RTA served to raise a number of suggestions for new features to be incorporated into future versions. That process, we now believe, will continue for years to come.
We have built an operational tool which we are using in second language acquisition and will soon use in courses in computer science and medicine, clinical consultations between distributed health care centers, and a wide range of other activities. Simultaneously, we are beginning to use RTA in our own collaborative research. The current version of RTA incorporates the following basic characteristics:
RTA uses a server which moderates all communication between clients, storing whatever needs to be stored, controlling access through usercodes stored on the server, and providing links to other tools as required. Data are converted from client-based formats to a standard format on the server.
Our server is currently a UNIX system. While other platforms could be used, the UNIX server is robust, widely available, and we have not yet had a need to adapt it to other operating systems. The advent of UNIX functionality on PC platforms further extends the useful domain of this decision, although other platforms could be used in the future.
The server is distributable within a given work group. We keep lists of all groups (e.g., class lists) on all servers within a given domain. Connecting with any server will enable a user to link to the one on which group information for that user is available.
The server moderates two forms of multimedia interaction: synchronous and asynchronous. In synchronous mode, pairs or groups can be formed for live discussions and collaboration. Messages can be broadcast to all group members, to the first "staff" person to respond, or to selected individuals within a group.
Management tools available through Web access enable owners of groups to define "staff," "students," and resources linked to that group. The owner can also view and manipulate all materials generated through group interaction.
The clients are currently Macintosh and PC, with a UNIX client partially completed. Support of second language acquisition has led us to focus on the Macintosh client, but we are now concentrating on bringing the PC client up to full parity with the Macintosh version. A user connects through a client to a server, selects a group, types usercode and password and is then connected to the group.
Once connected, the user can connect with another user in paired interaction or else create or connect to a group, identified by subject. The interactive process includes the following features:
* Dialog: time-ordered display of comments made by group members identified by username. The dialog may use non-ASCII characters sets: currently Spanish, Russian and Japanese are accommodated, but the capability is open-ended, permitting incorporation of scientific notations as alternate "languages." Users connecting late to a group have the entire history of the interactive session downloaded to their client.
* Whiteboard: Users can select a portion of their screen for snapshot, or they can select an area from a blank screen for true whiteboard notation. Annotation of the resulting window can be color-coded by user, allowing lines, circles, rectangles and text to be added to the image. The image is downloaded in standard format and adapted by each client for display on that system.
* Sound: A user can submit a sound message which is stored on the server and downloaded to each client which converts the standard sound format to local display format (WAV on the PC, SND on Macintosh, AU on UNIX workstations). The length of the message is parameter-driven, currently defaulting to approximately 20 seconds.
* Web URL: Anyone in a group can invoke a URL which will appear on the browser of all clients in the group. No further control of the other users' screens is included, allowing users to explore sites once downloaded without interference from remote sites.
* Textpad: a shared collaborative writing tool has been incorporated into RTA. This tool enables any group member to take control of a text document, add or edit it as desired, using ASCII and non-ASCII character sets as appropriate. Other group members may request control which enables them to make further changes. The dialog box and sound are used in conjunction with the textpad to further enrich the collaboration. Textpads can be saved, edited off line, the retrieved for further collaborative development.
* File transfer: This is an open-ended feature allowing transmission of any file from one user to others. Viewing of the transmitted file is through use of browsers independent of RTA, but discussion of the file may be through RTA modalities.
The client also permits asynchronous creation and viewing of messages, which include all features of the interactive process with the exception of Web page URLs (which are not needed in this context). Messages may be annotated; new sound messages added, a graphic image further annotated, again using different colors for clarity, new or edited files attached, and the dialog added to. The modified message can then be returned to the originator and optionally forwarded to other users, who can review them at their own leisure.
Finally, the client has a "Content" link, in which a group owner can identify any form of resource (text or other file, multimedia quicktime or other graphic, sound, etc.). This feature is again open-ended, linking directly to Web sites as appropriate.
RTA has been operational in beta versions for over a year. It has been used experimentally linking second language students at Berkeley with students at the same level at Davis, matched in pairs and developing joint research assignments in the target language. It has been used in intermediate Spanish and Japanese classes at Davis for paired research requiring collaborative text-pad document generation based on live and asynchronous messages. It is being used during the current academic term for other sections of those two languages, and in a computer security course. Plans are being laid for extended use to collaborative research, and to clinical consulting and medical courses. We are investigating other uses and would welcome suggestions from interested groups.
The experience to date has been highly positive. While it is difficult to assess specific learning achieved, or to compare learning achieved with other forms of instruction, it is clear from attitudinal surveys of RTA users that it is welcomed and sincerely appreciated. Japanese students are enthusiastic in noting their ability to use this means of enhancing their ability to write in that rather difficult character format. Interactive breakthroughs through negotiated dialogs are readily apparent in review of the records preserved by the moderated server process.
The barrier to wider use up to this time has been the lack of a PC client with functionality comparable to its Macintosh equivalent (textpad, one of the most powerful tools in live interaction, was conceived and added to RTA in early 1998, at a time when resources limited its implementation to the Macintosh client). With that barrier now removed, we anticipate an exponential growth in the use of RTA both in numbers and in variety of use.
The design criteria that led to the success of RTA are not new, nor are they proprietary. Others will develop functionality comparable to RTA in time. At present, however, it appears to be unique in its multiple capabilities, and since it is available without charge to educational institutions, we envisage a continued growth based on these concepts in the short term future.
We intend to link RTA to some other key collaborative tools. In particular, while we regard stored sound as essential, we recognize the value of live sound, not stored by the server. This is readily accomplished in our current uses of RTA in experimental collaborative research by the use of speaker phones, but we are actively working with other investigators to link audio components of the ACTS Toolkit with RTA, deriving maximum benefits from each. We are also working with the Emory University-based project: Collaborative Computing Framework for Natural Sciences Research, to combine the features of their project with RTA. These are only two examples of a natural joining of two powerful collaborative research tools. We are hopeful of establishing other links, partly through interaction at this conference.
RTA is available at the Website http://escher.cs.ucdavis.edu/ from which it is possible to download clients, test RTA in remote sites using the UC Davis server, and view documents amplifying on its capabilities. We welcome opportunities to establish closer ties, seek joint funding for further development, and accept suggestions for further enhancement of the RTA concept.
CCF: http://emily.mathcs.emory.edu/ccf ACTS Toolkit: see DOE2000: http://www.mcs.anl.gov.DOE2000/index.html RTA references: see http://escher.cs.ucdavis.edu
This research was initially funded by the University of California through several different programs. A donation by Apple Computers, Inc of a Macintosh faciliteated development of clients on that platform. In 1997, the US Department of Education supported the project with a three-year grant under the Fund for Improvement of Post-Secondary Education with matching support from the University of California. Active participants in the project besides the named authors have included Robert Blake David Fahy, James Gallant, Adam Karp, Jennifer Pelletier, Eva Barela, and Sarah Chavis.