Presenters: Tom Finholt
Collaboratory Participants: Bob Clauer, Peter Knoop, Gary Olson
Discussants: Jason Owen-Smith, Jim Meyers
The Upper Atmospheric Research Collaboratory (UARC) was one of the first collaboratory projects to become operational. Funded in 1992 by the National Science Foundation, UARC aimed to provide remote access to geographically isolated instruments used in upper atmospheric research. In 1997, UARC was reborn as the Space Physics and Aeronomy Research Collaboratory (SPARC).. Initially, UARC enabled scientists using the system running on NeXTstep to monitor the incoherent scatter radar at the Sondrestrom Observatory in Sondrestrom, Greenland. The collaboratory later included many more instruments of various types, including incoherent scatter radars, all-sky cameras, interferometry instruments, magnetometer chains, riometry instruments, satellite imagery, shuttle imagery, sounding rockets, and results from numerical simulation. Operation of the collaboratory was transferred to NCAR in 2002.
The initial UARC end-user system was built on the NeXT platform and migrated into the NeXTstep operating system. The NeXT platform provided a powerful environment for rapidly prototyping and deploying tools and interfaces, but its obscurity and the size of the UARC client software meant that using UARC required a significant investment in computing hardware. In order to reduce this barrier to use, the collaboratory took advantage of the developing world wide web and deployed a web-based system built using Java applets (MURAL) with the release of UARC v.6 in 1996. Realizing that the Java-based environment was still an undesirable thick client, SPARC was built as a thin client and is currently migrating to a CHEF-based system (link) (portal framework).
Throughout the lifecycle of the project, the collaboratory client provided access to real-time and stored data from the instruments that were integrated into the collaboratory framework, enabled remote control of some features of the instruments, facilitated communication between users of the systems and provided a variety of ways of visualizing data from the instruments and simulations.
The use of UARC included several campaigns (periods of special activity observed by many people at the same time) that demonstrated the potential of the collaboratory as a scientific and educational tool. Major scientific campaigns included:
- the first simultaneous real-time display of an incoherent scatter radar chain in April, 1997,
- the first combined real-time simulation and observation campaign in April, 1998,
- the first archival campaign in 1993, and
- simultaneous views of observational and simulation data over the lifetime of the project.
A major educational campaign took place in February, 1995 when students from a university in Florida participated in a campaign, working together with scientists at two locations in California, in Greenland, and in Michigan, engineers in California, and the site crew in Greenland.
A small number of individuals dominated the use of the UARC systems. The NeXTstep-based UARC v.5 had a total of 27 scientific users, with a peak simultaneous usage of 8 users. Eighty percent of the system's use is accounted for by four users. Similarly, the Java-based UARC v.6 had 100 users, with peak usage of 40 simultaneous users, but 80% of the use of the system is accounted for by six users. This finding is consistent across both use and chat logs.
Investigators at Michigan conducted a longitudinal survey over the 8 year period from 1993-2001. This survey focused on communication choice, use of web resources and outcomes connected with those factors among a sample of UARC users and a matched control sample. The survey showed a change in methods of communication over the period, with a difference between groups, changes in the level of web use with a difference between groups, and a relationships between both communication choice and web use to funding. Key findings from this survey include evidence of a substitution of email for other communication modalities, which was more dramatic in the control group, and increased use of other web sites. The survey also suggested some use of data intensive web sites, but did not find web use to have a universal impact on the field as no predominant web site emerged. Use of collaboration tools and online reading of scientific journals was also low.
UARC/SPARC is an interesting illustration of the lifecycle of a collaboratory project. The system began as a shared instrument, but evolved into many other things, notably a community of practice and a community data system. SPARC's use as a community data system was hampered by the lack of standardization of the database. The use of SPARC as a community of practice adversely affected the collaboratory itself as users directed each other to other sites and other instruments when things were going on.
The success measures that the evaluation team applied to UARC/SPARC are narrow and leave a large number of questions unanswered. For instance what are the undocumented and unobserved social consequences of the collaboratory with regard to education and outreach? On another level, the existing evaluation of the impact of UARC/SPARC on the science leaves a lot of room for further investigation.
The data that were collected by the collaboratory evaluation team, on the other hand, has tremendous potential. The information in the use and chat logs can provide insight into the flow of communication within the social network, they type of discussions the collaboratory facilitated, and the impact of technology on the field.
Mostly independent of the impact UARC/SPARC had on the space physics community, the collaboratory is a model that other fields have adopted. The success of SPARC cannot ignore the number of collaboratories it has inspired. To this end, UARC/SPARC may be a victim of its own success in that the type of things that were never attempted before the collaboratory was used are now being done outside the collaboratory more quickly and more easily.
Another feature of SPARC and collaboratories more generally is that they may allow science to grow to its natural size. Forcing collocation places a strong upper bound on both the size and complexity of a collaboration - removing this constraint allows the involvement of many more people and at many different levels or participation.
There also exists a hierarchy of collaboration, akin to Maslow's needs hierarchy. A key barrier to collaboration is the tradition of competition in the sciences. One way of cementing a collaboration is to institutionalize it as a research group - how do you create a research group around a virtual system?
Q: How transferable were data cleaning and analysis routines?
A: Not at all. Expertise regarding data cleaning is not heavily used. There is no incentive to clean other people's data and it is very difficult to do so.
Q: How important were access controls?
A: Initially access control was not very important, but when cleaned data and analyses began to be available, it was more important.
Q: Did the user base ever expand beyond the size of a good chat community? Are there limits on the size of a virtual community?
back to the Final Summary