Musings on Entrepreneurship and Innovation
Saturday, October 07, 2006Deductive Tinkering and the Evolutionary Algorithm
This is the second in a series of posts on my understanding of an evolutionary model of innovation. In my previous post, I introduced the concept of the fitness landscape in the context of physical technologies - devices that fulfill a human purpose. The goal of this post is to explain how evolution can be thought of as a general class of algorithm for effectively searching fitness landscapes.
The fitness landscape shows us visually...where the good designs are located. We can think of a good design as high fitness peaks, and our problem of finding good designs in the near infinity of design space can be reconceived as finding high peaks in the fitness landscape.
Eric Beinhocker, The Origin of Wealth
Evolution can be thought of as a recursive algorithm for searching design spaces for fitness peaks. The recipe is straightforward:
- Differentiate to create variety in order to search a sufficiently broad subset of the fitness landscape given limited resources;
- Select those designs that are the most fit, and amplify or replicate those designs; and
- Repeat.
In biology, variation is achieved through the mechanisms of random mutation and sexual recombination. Recombination also plays a critical role in generating sufficient differentiation among physical technologies:
...recombination is the fundamental mechanism behind creative insight...there's no such thing as the immaculate perception: People don't suddenly come up with new ideas; they piece them together from what they already know.
Andy Hargadon, How Breakthroughs Happen
However, random recombination - throwing spaghetti on the wall to see what sticks - isn't likely to yield much that is useful, at least in the timeframe relevant to business. As Robert Axelrod and Michael Cohen note in Harnessing Complexity:
...most of the variants introduced into orderly systems by [random] processes are deleterious - with occasional small improvements and a sprinkling of very rare spectacular advances. Exploring for new possibilities by nearly random variation can therefore be expensive.
Consequently, Beinhocker underscores the role of deduction in combination with experimentation as a mechanism for variation, something he calls deductive-tinkering:
A critical feature of Physical Technology evolution is that the fingers of deductive exploration have a higher probability of hitting high fitness peaks versus experimental tinkering or purely random search...The impact of science has essentially been to dramatically increase the hit rate of deductive insight. Deduction thus suddenly began to play a much bigger role in the deductive-tinkering mix.
Furthermore, many have noted the importance of social context to the generation of useful new ideas and have highlighted the impact of technology brokers:
...brokers are critical to learning and creativity. People whose networks span structural holes have early access to diverse, often contradictory, information and interpretations, which give them a competitive advantage in delivering good ideas. People connected to groups beyond their own can expect to find themselves delivering valuable ideas, seeming to be gifted with creativity. This is not creativity born of deep intellectual ability. It is creativity as an import-export business.
Ron Burt, The Social Origins of Good Ideas
So, "inventors" play an important role in the evolutionary model of innovation. But, rather than being the product of lone geniuses who generate deeply original insights, most invention derives from individuals' (often outsiders) ability to connect otherwise disparate realms of thought and practice and to engage in deductive-tinkering. Nevertheless, as Hargadon notes:
...the reason recombinant innovations are relatively rare is that the flow of people, ideas, and objects among these small worlds is relatively rare.
So, if technology brokerage and deductive-tinkering are differentiation mechanisms, what serves as the mechanism for selection? Beinhocker offers two candidates: Big Man hierarchies and markets, and he elaborates on how Big Men have been more common historically and how markets are more effective. Effective selection enables an environment in which "Successful designs tend to be amplified as they are copied, attract more resources, and spread" (Beinhocker).
There's a potential catch, however, as Axelrod and Cohen point out:
The advantage of strong selection pressure is that it exploits success...The disadvantage is that it can quickly destroy the variety...that is needed to explore for even better outcomes in the future.
Fitness is dynamic. A physical technology that is at or near a peak of high fitness today may find itself in a valley of low fitness tomorrow as a consequence of a morphing of the landscape. Thus, advantage is temporary and the search for higher fitness relentless (though some thinkers such as John Hagel III and John Seely Brown assert that businesses can proactively shape the fitness landscape as well as adapt to changes). In other words, the process of innovation is a competitive race without end.
In my next post in this series, I'll take a stab at explaining Beinhocker's thesis that business plans are the product of the co-evolution of physical technologies and social technologies and why that's important to inventors (who create differentiated physical technologies) and entrepreneurs (who create differentiated business plans).
