Design Rules: The Power of Modularity v. 1 [Hardcover]

Opening the "black box" of technological and industrial progress, Baldwin and Clark introduce the notion that technology is a set of tasks and the organization mirrors the design of the artifact that it produces.

The authors model builds upon the work of John Holland, Stuart Kauffman and Brian Arthur (from the Santa Fe institute) on Complex Adaptive systems (CAS). CAS have four properties:

1. Each of these systems is a network of many agents acting in parallel. The control of these agents is highly dispersed.

2. The CAS has many levels of organization, with agents at any one level serving as the building blocks for agents at the higher level. Furthermore, CAS are constantly revising and rearranging their building blocks as they gain experience. Baldwin and Clark carefully document four layers operating in the computer industry, The global financial system, the markets for goods and labor, organizations, and the design and production of computers. In Addition, the authors describe the six "modular operators", the complete set of options that can be used by agents to modify the system that can be used at any level.

3. All CAS anticipate the future. The various models, whether implicit or explicit assumptions, are constantly tested, refined and rearranged as the system gains experience. Baldwin and Clark assume that designers "see and seek" value, with value being measure in the global financial system.

4. CAS typically have many niches, each one exploited by an agent adapted to fill that niche. Moreover, the very act of filling a niche opens up new more niches. Thus, there is no equilibrium in these models, it is not about a "punctuated equilibrium". The process is a constant search for an improved fit with the environment. Moreover, the clock speed of the process should match the environment.

This book has deep implications for practitioners and scholars interested in understanding the "new" economy. I highly recommend the text.

This is an important book. John Holland, who is heavily referenced, foreshadowed its publication when he linked the concepts of real options and complex adaptive systems in a talk at the Santa Fe Institute in late 1998.

The authors clearly and persuasively explain how modular design adds a tremendous amount of value through the creation of real options. Furthermore, modularity allows for the evolution of both design and industry.

In the 1960's, IBM created the System/360, the first modular family of computers. As a result, IBM launched an industry -- and lost control over the tremendous value it stimulated.

"Design Rules" was recommended me by one of the authors' colleagues, who thought that I'd "eat it up." I did, and I'm hungry for Volume 2.

This is one of the most useful and practical business books I have read. At my companies we apply it at two levels.

We use the modularization rules in the design of systems and businesses. These rules (i) define what is needed to modularize a system - interfaces, communication rules, aggregation rules and (ii) a set of operators that can be used to evolve modular systems. It is best practice to design all of these (interfaces, communication rules and data exchange, aggregation and disaggregation rules and the modularization operators) into business systems. This is even more important than ever in a time of rapid cycles and transformational change.

It is also useful, especially when combined with the work of Clayton Christensen on disruptive inovation, to understand the evolution of industry ecologies. I have heard that the next volume (if it ever comes out) will go further in this direction. One needs a point of view on value-web evolution to make strategic bets. Of course one will often be wrong, the future will always surprize us. The real purpose of strategic thinking is to help us realize we are wrong more quickly so that we can respond and shape the emerging change.

Design Rules also serves as history of evolution and differentiation in the mainframe industry (there are rich veins of historical data), introduces or reminds us of variable analytical tools, and it is physically a lovely book, well designed, well bound.

Software people reading this review may think 'so what' we know all this stuff about information hiding and modularization. But I wonder, how many software systems are designed to evolve? Most decay with change. A system that evovles needs to be self similar across levels of organization, allow the modular operators to operate, represent itself to itself ... Even if object and aspect programming catch much of this, few software engineers are able to apply it at an organizatoinal or value-web level, and so their systems are detached from the environments that will select them.

I hope the authors and publishers will get the second volume out soon!