COORDINATION THEORY - Human-Computer Interaction and Management Information Systems: Foundations - page 115

Theoretical Factors
Martens identifies two theoretical factors for the success of a theory: multiple uses for theory, and
fit with other theories or approaches. Again, CT does well on these criteria. First, CT can be applied
to several purposes, such as modeling a process to understanding how it works to suggesting
improvements, or for designing new coordination mechanisms, particularly those that rely on
information and communications technologies. A number of articles build on the modeling
approach proposed by Malone and Crowston (1994). For example, Tolksdorf used the definition
and modeling concepts from CT to develop an XML schema for describing business process
models (Tolksdorf, 2000) and a dependency markup language (Tolksdorf, 2003).
Second, CT has been combined with other process modeling techniques. For example, Crowston
and Osborn (2003) developed a six-step technique for documenting organizational processes by
embedding CT in Checkland's (1981; 1990) soft systems methodology. They propose two general
heuristics for identifying dependencies: dependency-focused analyses (identify dependencies
then search for mechanisms) and activity-focused analyses (identify mechanisms then search for
dependencies). Examples of each approach are provided in the form of a small marketing services
company. Barbuceanu, Gray, and Mankovski (1999) offer a four-level model of social interaction
and behavior into agent-based programming tools. In this model, an agent's request consists of a
set of “obliged and forbidden behaviors.” Social laws, expressed as obligations, are positioned at
the highest level of this model. Below this level is a layer that considers each agent's unique pri-
orities and authorities. The lowest levels of this model consist of scheduling and executing
adopted behaviors, respectively.
Kim (2000) uses CT as a basis for organizational process change by applying the Massachusetts
Institute of Technology Process Handbook (Malone et al., 1999) with a simulation technique. The
handbook modeling approach extends the framework of CT described above by adding the notions
of inheritance and decomposition. A process can be decomposed into smaller steps, which in turn
can be decomposed further. Once the process has been decomposed, the dependencies among the
steps can then be analyzed. A novel feature of the Process Handbook is the use of inheritance,
adopted from object-oriented programming. A process in the handbook can be viewed as a spe-
cialization of a more general process (e.g., the processes for selling in a store or selling on an
e-commerce Web site are both special cases of selling; the process the company Lands' End uses
to sell using a Web site is still more specialized). This specialization can be done at any level of
decomposition. The combination of the two approaches provides great flexibility in composing
novel processes. Kim (2000) gives an example of this approach in a case of an organizational
change in a hospital.
Several authors have developed new coordination mechanisms, particularly those in distrib-
uted artificial intelligence. For example, Decker in his PhD thesis designs and evaluates a “fam-
ily” of coordination mechanisms for cooperative computational task environments. Chen and
Decker describe seventeen coordination mechanisms to handle the dependency relationships among
multiple agents' tasks. They describe these as components that can enable agents to adapt to
changing environments. Each mechanism was catalogued into one of eight groups (p. 6):
• Avoidance
• Reservation schemes
• Simple predecessor-side commitments
• Simple successor-side commitments
• Polling approaches