 |
 |
 |
 |
| |
|||||
|
|
|
 Challenge Problems Guide the Future of Collaboration Research By Rod Holland MITRE's collaboration research program, comprised of multiple teams supported by diverse sponsorship, has amassed significant accomplishments in the collaboration arena. MITRE research has elucidated some of the basic technical and organizational principles of the construction and use of collaborative mission environments. This has helped stimulate new operational thinking in our sponsor communities. The various agencies affected are issuing new requirements, and vendors are trying vigorously to deliver commercial products to satisfy them.
In this environment, it is legitimate to pose the question, "What will be the focus of future collaboration research?" Should we declare this area mature and move on to agent technology? Is MITRE's role in collaboration technology now confined to advice on acquisition decisions and similar operational matters? On the contrary, I believe that much remains for MITRE to do in the field of collaboration research, but only if we continue to break new ground. Breaking new ground in collaboration research requires that we step away from the problems that have become comfortable -- work group- and small-enterprise-scale collaboration -- and focus on a set of new challenge problems. These challenge problems are technically difficult, are important to our sponsors, and are not necessarily going to be solved by the "waiting for COTS" algorithm. Many of the challenge problems are interrelated; multiple investigations across the gamut of difficult topics are required to lay a secure foundation for the next generation of collaboration technology.
Current systems (e.g., Collaborative Virtual Workspace, GTE's InfoWorkSpace ™, Odyssey) have credibly demonstrated a capacity to serve communities of up to several hundred users simultaneously. The whole Department of Defense (DoD) and the IC (Intelligence Community) can be considered together as an extended community with a head count in the low millions, and an organization count in the thousands or tens of thousands. While individual work groups and organizations may be well served by existing solutions, the liaison and echelon relationships within the larger community are not. Current systems have a centralized client-server architecture, which is unlikely to scale appropriately, and raises issues both of control and of catastrophic failure. Server federation has been proposed as a possible solution for both CVW and InfoWorkSpace, but remains to be demonstrated. More radical approaches, such as fine-grained distribution of collaboration services -- a collaboration fabric -- may be more promising in terms of scalability and robustness. The COTS experience with very large deployments of instant messaging services needs to be studied.
The rapid deployment of the current generation of collaborative environments to the sponsor base has resulted in the creation of what MITRE's Cindy Kabat has termed "islands of collaboration," in which different organizations deploy different systems, thereby compounding the problems already presented by scaling and distribution. If one considers that uniformity by bureaucratic fiat (whether desirable or not) is unlikely to succeed so early in the evolution of a technology, then solving the problems of cross-representation, intercommunication, and document interchange among heterogeneous collaborative environments becomes urgent. More subtle is the problem of collaboration policy automation at organizational boundaries: if my agency is collaborating with your agency, what shall I let you see of my virtual environment, what are you willing to let me see of yours, and how can we arrange for the automatic enforcement of our respective policies so that we can initiate a collaboration quickly and confidently?
The interaction between collaboration technology, which seeks to make information available to all who need it, and information security, which seeks to ensure that information is provided only to those authorized to have it, is an extremely rich research topic, which should sustain a variety of investigations. For example, anybody trying to provide access to a common collaborative environment across multiple firewalls in a way that satisfies both the operators of the firewalls and the participants in the collaborative environment will appreciate the difficulty of the problem and its immediate relevance.
Existing implementations of collaboration systems, with their centralized server architectures and expectations of relatively benign network environments, are not well suited to the demands of tactical and mobile users. In a tactical environment, servers may come and go -- sometimes never to return. Network Quality of Service parameters will typically range from very inhospitable to bizarre. Adequately supporting tactical users requires research in network services, distributed computing, and collaborative Human-Computer Interaction (HCI) domains, and the construction of fieldable prototypes for testing in tactical environments.
Collaboration technology is ineffective if it is not used. Barriers to use may derive from user interfaces, user interaction policies, or user interaction modalities that clash with individuals' workstyles. Other barriers may derive from a mismatch between group culture and practices and the characteristics of a newly-deployed collaborative system. Investigations in HCI for collaborative environments can address the first problem; organizational behavioral research can shed light on the second. MITRE has made important contributions in these areas and should continue to do so.
Meaningful, reliable, and reproducible evaluation methodologies for collaborative systems must be developed and employed to stimulate rapid progress in collaboration technology research and to help organizations measure the operational effectiveness of deployed solutions. These evaluation methodologies must be supported by an array of instrumentation techniques for collaborative environments. MITRE's strong presence in these research areas provides us with many opportunities to aid our sponsors.
The question of how one builds a collaborative environment, or a collaborative mission application, has numerous answers. For example, the question of how one most effectively uses component architectures for both client and server construction is a topic of current investigation. As new technologies (e.g., Jini ™) become available, novel collaboration architectures will suggest themselves and should be investigated. Some may represent real advances over earlier solutions; they may also contain hidden traps for government users. Timely investigation could bring to light problems inherent in new architectures and, perhaps, suggest remedies. For more information, please contact Rod Holland using the employee directory. |
Solutions That Make a Difference.® |
|
|
Problem
1: Scaling and Distribution
Problem
2: Heterogeneity
Problem
3: Security
Problem
4: Tactical/Mobile Collaboration
Problem
5: The Human Factor
Problem
6: Evaluation and Instrumentation
Problem
7: Architectures