Large, inherently complex, socio-technical systems, such as the electric power grid, depend heavily on the flow of information to maintain productivity levels. Hi-tech manufacturing firms attempt to improve operations by integrating Information Systems (IS) into existing processes. To examine these efforts from a systemic viewpoint, I initiated a field research project in 1996 at a utility's electric power grid dispatch center. California's deregulation of the electric power industry offered a unique opportunity to examine the increased use of IS in a complex domain that is evolving due to deregulation pressures. The field research revealed how numerous institutional, economic, and policy constraints emerged at all levels of governing authorities and organizations. These constraints translated into system requirements, specifications, and ultimately a series of IS implementation efforts aimed at making changes to the Nation's, and in particular, California's power grid.

The IS projects I observed at grid dispatch depended heavily on integration and were driven by domain constraints. That development approach presents a high risk to successful IS implementations. To understand and facilitate these complex development efforts, I created a conceptual framework. The framework is a synthesis of several different approaches drawn from literatures including IS Architecture research, Software Architecture research, and complex adaptive system theories. A software tool based on the framework should help designers develop requirements from emerging integration constraints. Therefore, I designed and developed a software tool suite as a proof of concept to demonstrate how to codify or map the framework into a system. The tool is an analysis aide for designers who must consider complex systemic relationships when considering domain specific constraints; e.g., those arising from the economics and policies found in the electric power industry.

Using the tool, a designer builds an architecture prototype that can be executed within a simulated domain in order to gain important feedback. Stakeholders can use the tool to address resource constraints incrementally at varying levels of fidelity to bridge the gap between ideal and actual system configurations. The execution of the prototype architecture generates data to be analyzed during the design process. Furthermore, resource manipulation within the executing prototype makes the cost-effective exchange of resources or services possible. Thus, the designer compares system component behavior. In fact, actual (not just modeled) services can be inserted into a distributed architecture prototype allowing the incremental evolution of the architectural representation into the actual system.

Future Research:
The conceptual framework and tool that I have developed provides a springboard to pursue three areas of inter-related research.

• First, and particular to the domain, our evolving electric power infrastructure and specifically the power grid will necessitate the exploration of new models, technologies, integrated designs, and various market configurations. I intend to greatly expand the current domain description encoded in the present catalogue (i.e., dependent on my own constraints and resources available, especially regarding possible collaborative efforts). This includes additional use of resources from the web; cataloging them for reuse introduces several questions involving supply chain management, service provider models, etc.
• Second, the conceptual framework's underlying theory can incorporate adaptive (software) agents that could evolve the architectural configuration based on an internal algorithm. For example, it would be interesting and fruitful to explore agent usage in direct negotiation or online auctions for resources such as electricity exhibiting expected power quality characteristics. Specific architectures to support that interaction need to address a broad variety of constraints. The framework and tool will enable users to set constraint priorities based on broad high-level abstractions of the domain.
• Third, I envision additional research on taking the tool and developing it into a full featured Computer Aided Software Architecture (CASA) suite that will increasingly lend itself to use by interdisciplinary project teams including non-technical stakeholders. These collaborative opportunities are possible because of the distributed computing and work design employed by the tool. The tool's flexibility provides a wide range of degrees of freedom regarding future research that I intend to explore fully.