johnson@vaxa.isi.edu (Lewis Johnson) (09/06/90)
Call for Participation AAAI Spring Symposium Design of Composite Systems March 26,27, & 28, 1991 Stanford University The goal of this symposium is to discuss issues in formally representing and automatically reasoning about design problems that involve a mix of physical, software, and human components, i.e., what we will call composite systems. In particular, we are interested in designing an artifact that will be added to an existing environment to achieve new goals (or old goals more efficiently). In such systems, goal achievement can be viewed as a joint problem solving process. The artifact specified and produced will be embedded in a rich and complex physical and social context. In essence, the difficulty might not be in the specification of the artifact itself, but instead in reasoning about the complex interaction of the artifact with the environment. For instance, the artifact by its very presence may place restrictions of the freedom of action of human agents, change the pattern of human-human interactions, etc. A designer is thus defining new roles that agents in the environment will play when interacting with the artifact. We propose that there is a common set of problems for specifying and designing composite systems irrespective of software/hardware/human distinctions. At least one purpose of the symposium is to bring together researchers from various fields to explore this claim. We wish to determine whether techniques developed for the design of software, hardware, interfaces, organizational processes, etc., can be adapted to support composite system design, or whether new techniques are needed. The symposium will examine the following issues in composite system design: Theoretical Issues - Determining the division of responsibility for system-wide goals among the various components. - Motivating components to abide by their responsibilies; designing for tolerance of error and of malicious irresponsibility. - Detecting and resolving conflicts between goals of the different components of the system. - Designing the interface between components. Pragmatic Issues - What are the limits of this approach? Is it effective? What tools do we need? - When is it appropriate (cost-effective) to use this design methodology? What new burdens does this introduce, and how can these be mitigated? - How can we ensure the acceptability of the artifacts that we design? Can acceptability be treated as another goal to be achieved? - Do we understand the conceptual foundations and techniques well enough yet to make the approach practical at this time? If not, how should we go about developing them? Prospective participants are encouraged to contact a member of the symposium committee to obtain a more detailed description of the symposium goals and issues. Participants should then submit a 1 to 4 page position statement explicitly addressing one or more of the symposium issues. Descriptions of experimental and empirical results are especially encouraged. Submissions that identify the practical or theoretical limits of automated design of composite systems are also welcome. Submissions should be sent by electronic mail to johnson@isi.edu by November 16th. All such submissions will be promptly acknowledged. If electronic mail is impossible, four paper copies should be sent to arrive by Nov. 16 to: Lewis Johnson USC / ISI, 4676 Admiralty Way, Marina del Rey, CA 90292-6695 or fax a copy to (213) 823-6714. Program Co-Chairs: Martin Feather (feather@isi.edu), Stephen Fickas (fickas@cs.uoregon.edu), Les Gasser (gasser@pollux.usc.edu), Lewis Johnson (johnson@isi.edu).