Title: Technology as a complex adaptive system: evidence from patent data
Author: Lee Fleming and Olav Sorenson
Scope: 4 stars
Readability: 2 stars
My personal rating: 4 stars
See more on my book rating system.

Topic of Article

The authors draw comparison between technological innovation and biological evolution via natural selection.

My Comments:

This article is definitely not easy reading, but it gives compelling statistical evidence that technologies evolve just as plants and animals do. In addition, it shows why broad-thinkers tend to be more innovative than specialists.

Key Take-aways

• Innovation is about combining previously existing technologies together to form new technologies that solve a new problem.
• To be innovative, one must search broadly to find technologies that are currently being used to solve different problems.
• Extreme specialization hurts innovation because specialists rarely look outside their immediate domain for ideas. Specialist are excellent at perfecting existing technologies, but it requires knowledge in a wide range of fields to innovate consistently.

Important Quotes from Book

This paper develops a theory of invention by drawing on complex adaptive systems theory. We see invention as a process of recombinant search over technology landscapes. This framing suggests that inventors might face a ‘complexity catastrophe’ when they attempt to combine highly interdependent technologies. Our empirical analysis of patent citation rates supports this expectation. Our results also suggest, however, that the process of invention differs in important ways from biological evolution.

We consider an invention to be either a new synthesis of existing and/or new technological components or a refinement of a previous combination of technologies (Henderson and Clark, 1990; Fleming, 2001). This framework allows us to think of invention as a process of recombinant search for better combinations and configurations of constituent technologies.

At first glance, technological evolution differs from evolution in biological systems in at least one important respect: the agent of recombination (Basalla, 1988). In natural evolution, recombination occurs primarily through haphazard sex. In contrast, inventors can purposely combine elements in technological evolution. Because inventors have a much higher degree of intelligence than the automata that navigate Kauffman’s landscapes, they can move beyond simple search patterns, such as hill-climbing or random combination. Understanding the impact of more realistic search heuristics strikes us as an important subject for future research.

Similarly, the ability of humans to move beyond blind search algorithms might systematically alter the nature of technological evolution over time. Intelligent actors can develop conceptual models, not just of the topography of the landscape, but also of the forces that create that terrain. Over time, inventors might become increasingly ‘foresighted’ and able to predict the outcomes of previously untried combinations. The development of broader scientific and technological knowledge might also improve the search process. landscapes… Thus, the effective difficulty of using interdependent technologies declines with time and experience.

Cognitive limits constrain such technological expertise to narrow fields. For this reason, people concentrate on particular subjects in the course of their schooling and work.

Inventors begin the process by trying completely new combinations of components. They then discard the obvious failures and re-organize the most promising combinations and interfaces between components. This modularization of the initial synthesis decreases the effective interdependence between components. Without new syntheses, however, the modularization process eventually exhausts creative potential. Thinking of invention as a process of recombinant search over an interdependent landscape provides a more complete and causal explanation of the technological sources of the life-cycle, from birth by synthesis, growth and productivity through initial modularization, eventual exhaustion from complete modularization, and rebirth through new interdependent syntheses.