A New Construct for Systems Modeling and Theory: The Kind

Abstract

Our primary research goal is the development of theories and technology to facilitate the design, implementation, and management of complex systems. Complex systems, in this context, are any systems which exhibit "interesting" behavior including, but not limited to, nondeterminism, collective or emergent behavior, and adaptability. We can claim to understand a system only when we can describe how it works (e.g. provide a specification) such that, if it is a constructive system, another can build it. This notion is our constructive peer of the traditional scientific method: repeatability of results is equivalent to repeatability of construction. Abstraction is recognized as a key to understanding complex systems. While increasing our abstraction level results in a more complete metamodel (i.e. we can talk about more systems because we can talk about more complex systems), it also means a more complex metamodel. On the other hand, we don't want to create theories and systems that only an expert can use. We need abstractions that are useful, comprehensible, and manipulable by humans (modelers, simulators, designers, developers, tool builders, etc.) and systems. In our experience, the highest-level abstractions in use today (e.g. classes, objects, types, subjects) can not model the systems we are interested in exploring. A higher-level abstraction missing: an "ubertype" of sorts - a syntactic and semantic bridge between types. We call this new abstraction a "kind". This paper will briefly describe kinds and provide several examples of their use.