Science of Collaboratories logo - link
An alliance to advance the understanding of collaboratories
Science of Collaboratories
Workshops : Comparative Investigation : Final Summary : GLR CFAR

Great Lakes Regional Collaboratory for AIDS Research (GLR CFAR)

Presenter: Stephanie Teasley
Participants: Dr. Steven Wolinsky, GLR CFAR Director, Jamie Drew, the head administrator of the GLR CFAR
Discussants: Jim Herbsleb, Prasun Dewan

The Great Lakes Regional Center for AIDS Research (GLR CFAR) was the first distributed CFAR sponsored by the NIH. This collaboratory, which built on the successes achieved in the SPARC/UARC project, bridges the geography of the four member institutions, Northwestern University, University of Michigan, University of Minnesota, and University of Wisconsin with off-the-shelf information technologies. The scientific goal of the GLR CFAR is to provide:
   1) complimentary technological or expert-based services, and
   2) educational opportunities for all members of the participating institutions.
Membership in the collaboratory is open to any AIDS researcher at the GLR CFAR member institutions. To date, there are 110 members.

Information was gathered by conducting user interviews and observations to select the appropriate technologies for the collaboratory. Communication, data access, shared authoring, and distance education were identified as functions the technology needed to support. Microsoft NetMeeting was selected for communication and shared authoring to support activities such as a pathologist and clinician discussing a tissue sample and two clinicians negotiating the final details of a clinical protocol. PlaceWare, a java-based conferencing tool, was selected for distance education, to support a lab seminar series with national experts presenting pre-published research. Xerox DocuShare was selected for data access and sharing to support activities such as the distribution of completed developmental award applications to the GLR CFAR Scientific Advisory Board.

Collaboration readiness of the participating scientists was also assessed in the preliminary interviews and augmented with a web-based survey administered to 10 GLR CFAR members, who are referred to as the "founding scientists." In terms of collaboration technology readiness, all participants had at least one year of email use and none of the participants had prior experience using other computer-mediated communication, i.e., phone and fax were the primary communication methods used for long distance collaborations by these scientists. In terms of social collaboration readiness, all the participants reported reasonably high levels of trust among other participants and there were 4 pre-existing within-site collaborations, 4 pre-existing cross-site collaborations, 3-5 anticipated new collaborations, one of which is a new cross-site collaboration.

Two cases of use were reported:
   1) a case study of the founding scientists and
   2) a general study of the 110 members of the collaboratory.

The case study of the founding scientists and general study included interviews, observations, citation and grant award analysis, and web-based surveys. In the case study, focus was on the founding scientists':
   1) satisfaction with the IT tools,
   2) within and cross-site collaborations, and
   3) impact on scholarly work.

The scientists quickly adopted technology, and two scientists began using the technology for other geographically distributed collaborations outside the GLR CFAR. By year three, there were 8 new grants between the 10 scientists, including five grants across two sites and one grant across three sites. Two of these eight grants involved scientists who had never collaborated together and three of these grants applied a clinical-bench approach.

In the general study, the focus was on:
   1) satisfaction with tools and
   2) impact of the collaboratory on scholarly work.

Seventy-five percent of the total GLR CFAR membership participated in at least one of the collaboratory's educational activities, the virtual seminars. Members reported having an average of eight GLR CFAR collaborators. Members also reported having two hundred distinct within-site collaborations and sixty-eight between site collaborations. Finally, there was a 64% increase in the NIH funded research for the CFAR members.

There were two indicators of success, couched in terms of the overall scientific goals of the collaboratory. The first was the development of a clinical protocol by two clinicians at two different sites who had never worked together, using MS NetMeeting. According to the clinicians, NetMeeting enabled this task to be completed faster than their previous attempts at geographically distributed protocol development. The study based on this protocol was subsequently funded and produced two high quality publications. The second indicator of success was that 8 of 9 developmental awardees went on to secure competitive and prestigious RO1 grants.

Two questions face researchers of distributed centers; What influences participation and how is participation assessed.

Comment from the CFAR Participants

The Director and many of his colleagues have integrated the technology into their everyday ways of communicating, i.e., the collaborative technology is no longer considered something novel.


Jim Herbsleb and workshop participants
The GLR CFAR was a noteworthy collaboratory because it was inherently interdisciplinary, bridging clinicians and bench scientists, and appeared to enable scientists to complete tasks more quickly. In spite of these positive features, there are some fundamental questions that need to be resolved (in no particular order):
   1) Do the founding scientists affect cross-site collaboration?
   2) What is the scientific impact of the collaboratory?
   3) What was collaboration like among scientists before the collaboratory?

There are three hypotheses about time concerning collaboratories:
   1) distance matters less in collaboratories over time,
   2) there are huge impediments to starting,and
   3) different tools are needed to support different phases of a collaboratory.

Tool selection should not be simply based on what the scientists report wanting, but should also include observations of the work. One of the most important aspects of work is understanding the dependencies of tasks. The three areas of research that can help understand work:
   1) information processing theory;
   2) stability of environment, and
   3) soft dependencies.

Prasun Dewan and workshop participants
Research efforts regarding collaborative tools should be divided not by domain, but by application features. That is, it should focus less on the content of work of the scientists being supported and more on the application features that support the scientific work. We should develop a common model of interaction to help system designers build tools to support collaboratories.

Based on the GLR CFAR data presented, off-the-shelf tools e.g., NetMeeting, Timbuktu, and Virtual PC, seem to work fine for application sharing, but collaboration requires more than one tool. What would the ideal single tool be, especially for also doing "beyond being there" collaboration? Projects should be defined not by domain but by points in dimension space.


Distance and Organizational Structure
Physical and organizational distances are significant and must be considered in design and evaluation of technology in collaboratories. Organizations reorganize themselves to become parts of different references groups, which can bridge some of these differences. The blue ribbon panel on cyber-infrastructure found that computational capabilities can allow scientists to tackle larger problems, but it requires a larger group of participants. It is not clear how to organize them.

Technologies to support collaborative science
Focus on user experiences and work practices in designing or selecting tools to implement. We should not be chained to off-the-shelf technologies when users are working in ways that are not supported by them.

Funding Agencies
It is difficult to show value added of collaboratories to scientific practices. The problem for funding agencies is how to help scientists make these large-scale projects work and produce high quality science.

back to the Final Summary

  Home | About SOC | Workshops | Resources | News & Events  

University of Michigan Logo

University of Michigan

School of Information Logo

School of Information University of Michigan