Chaos in Models, Life Cycles, and Strategies

Introduction

I use chaos as a metaphor in two ways: to see the structure of a piece of code (the Chaos model) and to see the sequence of events within a process (the Chaos life cycle). I use these metaphors to show why the middle levels of the project are important (the Complexity Gap) and to discuss the need for a systematic strategy (the Chaos strategy).

Chaos Model

The Chaos Model defines a chaotic structure within software processes. We identify many levels of problem solving in a project.

At the top levels of a project, developer write and maintain whole a program.
At the upper levels of a project, developers write and maintain components.
At the lower levels of a project, developers write and maintain objects and functions. Developers decide which function or object to create, they create the object, and then test and integrate the function into the program.
At the bottom levels of a project, developers write and maintain individual lines of code. Developers first decide which line of code to write or edit next, they change the source using an editor, and then they see if the code looks correct and elegant enough to keep.

Adjacent levels affect each other strongly. For example, the meaning of a function depends very much on the meaning and position of each line of code. Distant levels affect each other only loosely. The meaning of a component depends very little on the meaning or position of any one line of code. If developers had used different data structures or a different program organization, completely different lines of code would have been written.

Chaos Life Cycle

The Chaos Life Cycle defines the sequence of events within a process.

Complexity Gap

The Complexity Gap extends the ideas from the Chaos model and describes the important middle levels of a project.

The lowest levels of a project are well understood. Any programmer worth his or her salt can write a few lines of code, almost without thinking, but writing anything larger becomes very difficult.

The highest levels of the project are also well understood. The highest level is the whole program. Stepwise refinement works well at the top levels to break the whole program into large components, but below that, it degenerates into a morass of arbitrary decisions.

The trick is to understand the middle levels of the project, where the high-level goals transmute into low-level solutions. We call this gap between the highest and lowest levels the "complexity gap."

Chaos Strategy

The Chaos Strategy helps to answer the question, "what line of code should I write next?" It is a "middle-out" strategy that presumes that the process has a chaotic complxity.

Webliography

The following site discusses chaos in software engineering and related ideas.

Richard Botting Can One Size Fit All?
Ron Burback (1997) Software Engineering Methodology: the Watersluice

Following sites discuss chaos in general terms.

Bibliography

Lawrence Bernstein (1996) Better Software Through Operational Dynamics, in IEEE Software, 1996, IEEE Press.
Lawrence Bernstein (1996) Software Investment Strategy, in Software Engineering Notes, Volume 21, Number 1, Pages 47 to 53, March 1996, ACM Press.
Dennis J. Frailey (1993) Concurrent Engineering and the Software Process, in Proceedings 2nd International COnference on the Software Process, Pages 103 to 114, IEEE Computer Society Press.
Peter Kokol, Janez Brest, and Viljem Zumer (1996) Software Complexity - An Alternative View, in SIGPLAN Notices, Volume 31, Number 2, Pages 35 to 41, February 1996, ACM Press.
Benoit B. Mandelbrot (1983) The Fractal Geometry of Nature, Freeman.
Dave Olson (1993) Exploiting Chaos, Van Nostrand Reinhold.
L. B. S. Raccoon (1995) The Chaos Model and the Chaos Life Cycle, in Software Engineering Notes, Volume 20, Number 1, Pages 55 to 66, January 1995, ACM Press.
L. B. S. Raccoon (1995) The Complexity Gap, in Software Engineering Notes, Volume 20, Number 3, Pages 37 to 44, July 1995, ACM Press.
L. B. S. Raccoon (1995) The Chaos Strategy, in Software Engineering Notes, Volume 20, Number 5, Pages 40 to 47, December 1995, ACM Press.

Go to L. B. S. Raccoon's home page.