Order this book
Department of Biological Anthropology, University of Cambridge.
This book is the result of meetings between the contributors, who were brought together by John Ziman to form The Epistemology Group, explicitly for the purpose of exploring evolutionary theory as applied to the process of technological, rather than biological, change. The book title would suggest that evolution is particularly amenable to the study of innovation, but thankfully the chapter authors do not restrict themselves to this aspect of the technological process.
The chapters are highly heterogeneous in theme, perspective and style, as might be expected when their authors come from such a wide range of disciplines (including physics, engineering, computer science, biology, economics, social anthropology, business management, and education). For example, we have W. Bernard Carlson's historical investigation of Thomas Edison's sketches for the telephone, Gerry Martin's study of stasis in the production methods for Japanese swords, David Turnbull on the social nature of changes in cathedral design during the Middle Ages, Alan Macfarlane and Sarah Harrison's comparison of Japanese and European technologies in their dependence on human power, Walter G. Vincenti's case histories of bridge and aircraft design, Gerard Fairtlough on how businesses can organise to innovate successfully, Eva Jablonka on Lamarckian inheritance systems in biology, Geoffrey Miller's review of genetic algorithms and Joan Solomon on school children's perceptions of technology. While interesting in their own right, many of these chapters do not contribute significantly to the book's primary goal of answering the question: "Can technological change be adequately and fruitfully described as a process of variation and selection?" (p. 3) The remaining chapters are more explicitly devoted to this task. But even here, the varied conceptions among the contributors of how this question should be answered present problems in linking them together into a unified whole. The sense one gets is that, despite the considerable efforts of the editor and contributors, a synthesis of biology and technology remains beyond reach, with some of the participants doubting whether it can ever be achieved.
The primary difficulty is that, at present, the application of evolutionary thinking in technological studies remains at the level of analogy. This means that problems begin with fundamentals, such as finding the basic unit of analysis. Nearly everyone agrees that to explain technological advances, we must look beyond the artefacts themselves. The ideas behind them (which can be separate from them), as well as the institutions that support and make them, also seem to play crucial roles in technological change. What then actually evolves? Artefacts themselves (e.g. Basalla 1988), the technical knowledge required to make them (e.g. chapters by Joel Mokyr and Edward Constant) or some combination of these? For example, James Fleck argues (in this volume) that the unit of technological analysis is the interface of artefacts and ideas in technological practices. By contrast, Nelson and Winter (1982) argue that it is the organisations (such as firms) necessary to produce artifacts. Disagreement abounds even with respect to this central issue.
Rikard Stankiewicz points out in his chapter that while considerable attention has been paid to the problem of defining or isolating the unit of technological evolution, almost no one discusses what is perhaps a more important question for an evolutionary analysis: how does heritability occur in technological systems? That is, how do technological units (whatever they may be) carry their information forward through time? Establishing the mechanism through which similarities between subsequent generations of artifacts are produced is essential to any evolutionary approach. Such lineages of descendants are the true focus of evolutionary theory (at least according to David Hull, the philosopher of biology responsible for the dominant conception of evolution, which distinguishes among replicators, interactors and lineages). So besides a unit of analysis, evolutionary thinkers applying their wares to technology will have to identify lineages of artifacts in which members are descendants of one another, not mere sequences of machines fulfilling a common purpose. For example, quite a few successions of artefacts serve the function of "playing music", for example - records, magnetic tape, DVD, MP3 and so on - but may not represent a proper evolutionary lineage, with each successor developing out of earlier models.
If, as I argued earlier, evolutionary analyses of technology largely remain at the level of analogies to the organic process, then disanalogies are bound to appear when the same set of concepts is applied to both domains. This problem becomes obvious when contributors argue that artefacts are designed, and are therefore defined by their purpose or function, not just by their existence or form. This appears to make technological innovation teleological or Lamarckian in nature, because then foresighted modifications seem to enter directly into the "germ line" of later models. But if this is true, then the Darwinian artifice of random selection and variation must be abandoned. As the editor John Ziman admits, this is a potentially crippling consequence for anyone who wishes to take the evolutionary approach seriously. Thus, the contributors collectively conclude in their final chapter that technological change is a bit like Darwinian evolution. But this is the primary reason evolutionary approaches have not been taken very far in the past: evolution may be useful as a heuristic, but is not formalised, and perhaps cannot be formalised. Is there a way forward?
Jumping headlong, as most of the contributors do, into the typical analogy - which compares market competition between brands of artifacts to ecological competition between alternative genes - seems to be the wrong point of entry into evolutionary theory. What becomes obvious from the book's case studies is that the course of technological development has required evolutionary processes to become increasingly enveloped inside other evolutionary processes. What began as a simple trial-and-error production process in the original hand manufacturing of tools, for instance, has become, in contemporary circumstances, a multi-stage phenomenon. When designing new artefacts, purely psychological choice mechanisms have been coupled to the use of computers to come up with alternative artifact forms, prior to any testing. Then, on the production side, prototype artefacts get certified for specific traits by firms prior to full-fledged manufacture, or the building of factories for the purpose of making those artefacts. Selection processes are increasingly embedded as well. Firstly, individual R and D technologists must select a place they wish to occupy in the intellectual "space" within firms. Then, factions within firms that prefer to produce different products compete amongst themselves. Finally, firms must compete in the economic market for market share and the hearts of consumers as well. This last arena can also be manipulated by marketing, which seeks to spread memes affecting product choice through branding, the linking of artefacts to social elites, popular cultural elements and so on. So each side of the production-and-consumption equation has been vastly amplified during the course of time. This elaboration is necessary to carry forward, in a systematic fashion, the larger and larger quantities of information involved in constructing more and more complex artefacts. Just as this recursion has occurred during history, it must occur in our modelling of contemporary technological change. This may make our models more complex, but not prohibitively so, since they just consist of evolutionary algorithms nested inside one another. As Edward Constant concludes a somewhat different analysis, "It is precisely this feature of recursive practice that gives variation and selective retention in science and in technology its rational, reasoned, intentional character. These characteristics suggest that the evolution of technology is neither Lamarckian nor Darwinian, but rather comprises a unique, or at least different, instantiation of more general variation-selection-retention processes" (p. 232-3).
So what we really need to explain is the increasing complexity of technological practice over time. I would argue the main problem with the book is that its analogy between technological and biological change is based on the typical Darwinian model of intergenerational selection among variant genes in a population. But "Nature red in tooth and claw" played out again in the techno-economic field, is not what is required here. To explain the large-scale trend in recursive technological practices, in my view, we need an approach based on "Major Transition Theory" (Maynard Smith and Szathmary 1995). This theory, which is becoming more and more dominant in biology, asserts that there have been a number of points in history when new evolutionary agents have come into being, after which time a degree of restructuring of biological ontogeny (or, in technological terms, production systems) became necessary. In the course of such a transition, the agents that used to compete - such as cells in the primordial soup - learn to co-operate, forming a higher level of organisation, such as multi-cellular organisms. These organisms then become the target of selection, and the focus of newly evolved adaptations, as it is now organisms rather than cells that compete. It is during such evolutionary transitions that novel elaborations get incorporated into the life cycle of the new evolutionary agents. These agents require more complex production processes than before (e.g. simpler cell-based processes get nested inside the developmental processes specific to building a complex organism). Adding steps of computer-aided design to manufacturing would be an example of this process in the technology domain. New artifacts can then also be seen as an elaboration of earlier models. Thus, the combination (or recombination) of earlier artifacts into more complex wholes is often the process leading to the creation of new machines. This is obviously a much broader picture of evolution than the standard one, taking place over a much longer time scale and covering everything from the origin of life to our contemporary Western lifestyles.
Once you switch perspective to this "new" evolutionary theory, the problem of having to think of technological change as an analogy stretched to cover a domain of phenomena which it doesn't really fit goes away. Nor does one need to water down the theory until it does fit, but at the cost of making it so abstract that it becomes uninteresting and unproductive. One can apply MTT at full strength to both the biological and technological domains, so the application is a proper one. The problem of founding analysis on mere analogy - and a weak one at that - is avoided. Eva Jablonka and John Ziman discuss MTT in the context of the evolution of biological complexity, but none of the other contributors take up this approach and apply it to their technological subject matter. The book would have been much stronger in my opinion had they followed up on this cue. The contributors appear to acknowledge that this would have been a good idea in their collective afterword as well, but presumably many did not know of this work in evolutionary theory - which is new and developing - prior to writing their individual chapters. Others seem to sneak up on the idea from behind, like Edward Constant, who notes the recursive nature of technological practice in "...alternate phases of selection and of corroboration by use. The result is strongly corroborated foundational knowledge: knowledge that is implicated in an immense number and variety of designs embodied in an even larger population of devices, artifacts, and practices, that is used recursively to produce new knowledge" (p. 221). Adopting this new framework doesn't mean that every term in evolutionary biology will have to be carried across into the new territory, however. For example, the concept of species is unlikely to transfer to the technology domain because, as many have noted, there is simply too much reticulation - the borrowing of ideas or bits and pieces of machinery between lineages - in technological evolution. But then, the notion of a species doesn't actually apply very well for much of the organic kingdom either. Reticulation is also common in several corners of the biological world, where clear boundaries cannot be drawn between reproducing populations, due to the rather frequent exchange of genes between them. So perfect lines of descent are the exception rather than the rule, once you take the entire living world - and technology - into account. We have just been biased in our appreciation of this problem in biology by our almost exclusive focus on the large animals which dominate the landscape and which happen to exhibit good separation between genetic lineages. Reticulation therefore doesn't count as a disanalogy between technology and biology. I conclude that there is really no problem in pushing the evolutionary viewpoint more seriously than appears to be favoured by those in this book, once we begin with an appreciation of the hierarchical nature of evolutionary domains, and their long-term development, rather than focusing on short-term competition between current models.
In conclusion, this book is state-of-the-art in both theoretical and empirical terms. Unfortunately, that isn't saying much. A lot of work remains to be done to make evolution a viable research strategy and school of thought in the study of technology. The saving grace for the evolutionary approach is that Science and Technology Studies - which currently dominates the intellectual scene, and hence is its primary competitor - is just as (un)developed and (un)successful in predicting technological change. Technology is a field that remains wide open for analytical progress.
Meanwhile, the book under review can be enjoyed for its fascinating historical studies of technological change, which bring rich detail to the description of particular events, along with tentative links to the grander edifice that remains lacking in this area. Many with a professional or even passing interest in technology will therefore want to read this book to check where our understanding of this important force in contemporary life is poised. But since there is little in the way of formal theorising, and nothing in terms of model-building - a report by Paul David of a simulation of individual-level Bayesian learning doesn't count - this book can only provide hints to those readers of JASSS who seek to simulate artificial societies with their own "technological" capabilities.
BASALLA G. 1988. The Evolution of Technology. Cambridge University Press, Cambridge.
MAYNARD SMITH J. and E. Szathmary 1995. Major Transitions in Evolution. Oxford University Press, Oxford.
NELSON R. R. and S. Winter 1982. An Evolutionary Theory of Economic Change. Belknap/Harvard University Press, Cambridge, MA.
Return to Contents of this issue
© Copyright Journal of Artificial Societies and Social Simulation, 2001