The industrial evolution

June 14, 1996

Can Darwinism be applied to technological innovation? John Ziman examines the analogy between biological and cultural evolution.

Go to a technology museum, and look at the bicycles. Then go to a museum of archaeology, and look at the prehistoric stone axes. Finally, go to a natural history museum, and look at the fossil horses. In each case, you will see a sequence, ordered in time, of changing but somewhat similar objects. The fossils, we know, are sampled from the history of a family of biological organisms. They are similar because they are related by reproductive descent. They have evolved over time because they have adapted, by genetic variation and natural selection, to their changing environment. Can technological innovation be explained in similar terms? Do all cultural entities evolve by essentially the same mechanism?

The analogy between biological and cultural evolution has often been remarked. One need not accept the principle of "evolutionary epistemology", which interprets the whole story of human social, intellectual and material development as the continuation of organic evolution by other means. One can simply note the immense diversity of artefacts that are invented and put on the market, and the superior utility of the few that eventually survive.

This analogy can be developed in considerable detail. First of all: what are the structural analogies between biological and cultural processes? It is easy to think of situations where material artefacts - indeed, also, less tangible cultural entities, such as scientific theories, social customs, laws, commercial firms and so on - undergo variation by mutation and recombination of characteristic traits. These traits can be associated with "memes" - elementary concepts that replicate themselves and shape the artefacts, rather like genes. The entities that survive are differentially replicated and diffuse through the population. Mutualistic relationships are very common, as between pens and inks, or bombers and radar systems. Isolated sub-populations - "demes" - may separate for long periods before recombining. And so on.

What is more, the history of technology provides many episodes that are remarkably similar to well-known biological phenomena. Very cryptically (and in no special order), these include: evolutionary drift; adaptation to environmental change; developmental lock vestiges; niche competition; diversification; speciation; convergence; punctuated stasis; emergence; extinctions; co-evolutionary stable strategies; arms races; ecological interdependence; increasing complexity; self-organisation; unpredictability; path dependence; irreversibility; "progress". This list is too long to be decoded and interpreted in detail. Suffice to say that it strongly suggests the possibility of transforming the notion of "technological evolution" from a vague metaphor into a well-formed model.

But before trying to set up such a model, we must look at the flip side of the comparison. Unfortunately, as many students of the subject have pointed out, there are many "disanalogies" to take into account. Technological systems are not like biological systems in a number of important ways.

In the first place, it is very convenient to talk about memes as if they were rather like genes. But memes are not operationally equivalent to the indivisible entities discovered by Mendel. The characteristic features of an artefact cannot be analysed uniquely into precisely defined design elements that endure unchanged for long periods. Thus all bicycles have wheels, but these are so varied in design and construction that they cannot be regarded as manifestations of a "wheel meme" that persists from type to type.

Again, in technological evolution, memes from distant lineages often recombine and multiple parentage is the norm. No biological organism is like, say, a computer chip, which combines basic ideas, techniques and materials from a variety of distinct fields of chemistry, physics, mathematics and engineering. Its cladogram would look more like a neural net than a family tree!

The most serious disanalogy is that novel artefacts are not generated randomly: they are the products of conscious design. Inventors learn by experience and experiment, and visualise their creations before they make them. Their inventions thus acquire characteristics that are deliberately handed on to the next generation.

Such "Lamarckian" factors are forbidden in biology. But their presence would not necessarily make an evolutionary process perfectly predictable. Their main effect is to introduce linkages along the time axis. The range of feasible variants at a given moment is not limited solely by present circumstances, such as the materials and tools available: it is also conditioned by memories of past circumstances, such as unsuccessful configurations and ideas, and by mental images of future circumstances, such as a novel device in action. That still leaves a vast universe of possible variants that might or might not work.

Biological theorists make much of the relationship between the phenotypes that are subjected to selection, and the genotypes that encode them. This is much simpler and more precise than the relationship between artefacts and their "memotypes". The blueprint of a novel artefact is usually analysed and revised many times before any engineering work begins. Technological memes can be transmitted, stored, revived, varied and selected independently of the artefacts to which they might apply. It would be possible, for example, to construct a workable modern "penny-farthing" bicycle solely on the basis of an old photograph or patent specification.

Thus, in addition to the market selection of artefacts, technological innovation incorporates two other "non-biological" systems of variation and selection. Industrial research and development is a systematic process where virtual artefacts - for example, preliminary designs - are subjected to theoretical analysis, computer simulations and practical tests before entering the market system. Further back, research and development itself depends on the knowledge system of techno-science, where technological memes often evolve as generic concepts, disconnected from their practical utility. The "linear model of innovation" treats these as three separate stages in the overall process. In reality they are very closely linked. What is more, technological artefacts are cultural objects, and thus co-evolve with industrial firms, and other entities in their socioeconomic environment.

A realistic evolutionary model of technological change would thus be even more complicated than its biological counterpart. It would have to incorporate "design" as well as "selection".

What, then, are we to make of the striking analogies between biological and technological change? Perhaps these phenomena are not really sensitive to the structural details of the system. Perhaps they are common to all systems that evolve by mechanisms that include stages of partially random variation, selection and replication.

This line of argument is confirmed by the results of computer simulations on very simple models. The burgeoning literature on artificial life, genetic algorithms and so on contains instances of almost all the phenomena common to technological and biological evolution. For example, some forms of artificial life clearly exhibit "punctuated stasis".

Thus, instead of lumping technology and biology together into the same species, we should treat them as distinct members of a larger genus of "complex systems". The general theory of such systems then suggests some interesting research questions. For example, does a "Lamarckian" factor necessarily alter the nature of a "Darwinian" mechanism? Might it not just improve the efficiency of the search for a higher peak in fitness space? Could it perhaps facilitate self-organisation and damp out some of the random fluctuations as the system approaches the edge of chaos?

Improved understanding of technological change is not irrelevant to evolutionary biology. Evolutionists tend to take biology as the standard model, as if all evolutionary processes had to conform to its peculiarities. The exploration of an alternative system throws into relief those features specific to biology, such as nearly permanent genes and sexual reproduction, and suggests limits to their evolutionary functions.

A question that might exercise historians, ethnographers and prehistorians of technology is whether there has been a progressive move away from selection towards design in the invention or improvement of artefacts. How much systematic thinking went into the production and selection of stone axes? How much random trial and error will really be needed in the creation of the next generation of "designer drugs"?

Technological innovation is of such enormous social importance that it is worth pursuing such questions further into the interstices of scientific, industrial and commercial life. What is the relationship between "selection" and "design" in the research and development divisions of industrial firms? Does this relationship differ from firm to firm, or from industry to industry - and if so, why? Is the overall process speeded up by more feedback between the various stages, or does each system evolve at its own characteristic rate? What are the real selection criteria in various types of market? An evolutionary perspective should yield new insights into such practical matters.

Technological evolution is one aspect of cultural change in general. Material artefacts are meaningful objects. They can be preserved unchanged for centuries, and are not dissipated by close scrutiny. Their evolutionary trajectories ought to be similar to those of other cultural entities, such as social institutions, scientific theories or patterns of behaviour, but much easier to investigate and understand. The study of technological innovation, seemingly so marginal to the humanistic endeavour, could eventually lead right into its centre.

John Ziman is emeritus professor of physics at the University of Bristol and convenor of the Epistemology Group.

An open Forum on Technology and Evolution, arranged jointly by the Epistemology Group and the London School of Economics Centre for the Philosophy of the Natural and Social Sciences, will take place on June , 3pm-6.30pm at the New Theatre, Houghton Street, London.

You've reached your article limit.

Register to continue

Registration is free and only takes a moment. Once registered you can read a total of 3 articles each month, plus:

  • Sign up for the editor's highlights
  • Receive World University Rankings news first
  • Get job alerts, shortlist jobs and save job searches
  • Participate in reader discussions and post comments

Have your say

Log in or register to post comments