The ideal of diffusion
The 1960s produced a series of economic commentaries on the way in which innovations spread throughout a market. Ranging from Rodgers in 19621 to Bass in 19692 these all developed around the concept of diffusion. Developed formally by Rodgers, and held by many people today, the diffusion theory is a specialised concept of communication. This communication, verbal, nonverbal, and observational, then produces a diffusion curve. What I will attempt to lay out here is a very short critique of the basic theory of diffusion. Unlike most economic commentaries however I will start before the innovation stage and use more recent studies into the sociology of science to examine the nature of scientific investigation and draw parallels to the nature of innovation. When diffusion models are examined in this way they are not only revealed as misleading descriptions of the interaction between firms and consumers, but lacking in any real explanatory power. I suggest instead that a translation model would be far more effective at both explaining the spread of innovations and offering advice towards more effective interactions between firms and markets in the future.
The sociology of scientific knowledge (SSK) seeks to describe the development of science from a social perspective. The history of science has recognised that the development of science is not linear. Theories are not isolated, they arise under competitive circumstances. The aim of SSK is to explain why one theory or object is accepted over another and to understand the pathway followed in bringing the object to realisation. This requires following the interactions of the different actors involved in the movement of the object through society. Objects are then transformed as they pass from hand to hand of the multitude of actors that impress themselves upon it. Actors here can be both human and non-human, scientists and non-scientists, and the object has a defined meaning within the network that these actors form.3
The picture drawn by SSK of the birth and growth of a scientific object is highly seductive. We begin with the acceptance that rarely does a new object or theory spring into a single, isolated, mind fully formed. Instead the object is created in parts within many minds and forms only through the interaction of these actors. The object has still not been brought to realisation however. For this to happen it must be brought to the centre of the larger network of actors involved in the field to which this object might belong and established there as an immutable mobile.4 To ‘win’ in this regard, one of the original actors must achieve this goal whilst also appearing to remain in control of the object.
It is at this point that we must start drawing a distinction between the ‘Diffusion model’ of the motion of the object throughout the network, and what we shall refer to as the ‘Translation model’. Diffusion gives the object itself an inherent inertia, granted by its own explanatory power or its general applicability, which propels it through the network. It moves at different speeds through particular areas due to the resistance or acceptance of those groups to the new object. Translation, on the other hand, focusses on the attempt of the scientist to enrol others in the realisation of the object. In doing this the actor often becomes the spokesperson for a variety of different groups within the network. Bruno Latour describes various ways in which this can occur ranging from the actor attempting to provide the object to others as a possible shortcut to their pre-established goals, to creating new goals and convincing others that they must use the object to achieve them, to becoming indispensable.5
The power of SSK then is only realised with the use of the Translation model as here the union between the social and the scientific allows for consistent explanations for the acceptance of new objects within the scientific community as well as the usefulness of these objects to social needs. The continual interaction and transformation of each participant and the object means that the simple model of diffusion is not capable of providing insight into the resolution of scientific or technological controversies.
Decisions by the scientific community are rarely the end of the line for an object however. Next comes the application of the process of innovation to bring the object to market. This involves a combination of the adaptation of the object in its physical form and the adaptation of its presentation to the consumer. Some products succeed primarily because of the former, the internal combustion engine for example, and some grow primarily because of the latter, the branding of cigarettes as symbols of woman’s liberation in the beginning of the feminist movement for example6. The majority of successful innovations however are products of a combination of the two.
Models for the diffusion of technology are often concerned with an explanation for the S-curve representation of the adoption of new technology by consumers. Economic models which attempt to describe this curve hold that the technology spreads through a combination direct consumer interaction and central source diffusion. They then mix in certain other limitations on the speed of diffusion by introducing pre-determined aims/abilities (the Probit model), the result of competition (density dependence model), or the band wagon effect (resulting from information cascades)7. Like most models, each of these has their own advantages and failings and is best applied in different situations. However, none of them account in any appreciable way for the effect of marketing strategies on the success of the innovation. Though some of them allow for the transformation of the product due to market pressures they do not differentiate actual transformation from the control of perception.
This failing is best explained in the same way as the diffusive model fails in SSK, as being due to the separation of the innovative process and societal acceptance of the innovation. Economic models of technological diffusion rest on the premise that decisions made in the innovative process are part of a separate step in the diffusion of the technology. This consequentially does not allow for the inclusion in the models of the effect of actions such as marketing. Firms do not simply produce an innovation and allow it to diffuse through the market due to its own innate usefulness. Instead firms participate in a process of attempted enrolment of other actors in the hopes that this will establish the innovation as a necessary reference point in the larger network. This is the Translation model just as it has been described in SSK.
Once it is acknowledged that the analysis of innovative success cannot be achieved by dividing the firm from its target audience then the diffusive models are no longer applicable. Yes, they produce the same result as that which is observed, however they lose the vast majority of their explanatory power.
The requirement for economic analysis to involve symmetrically considering the enrolment of human and non-human resources is only becoming more important with the rise of ICT industries. Further to this the effect of social media on the style of network interaction only serves to complicate the interaction between the firm and society. Customer data can be gathered at a much more prodigious rate and in vast quantities, the requirement for repeated intervention to control the perception of the product only increases, and the international nature of markets supported by ICT creates a competitive structure that means the intrusion on domestic markets by international players also brings with it the need for interaction with networks on a joint local and global scale.
None of this can be satisfactorily accounted for in diffusion models. Diffusion is an ideal that presents saturation of a market as an inevitability if the technology is sufficiently suited to the needs of the society. If, however, we wish to understand, predict or interact with the process of innovation it seems that more effort must be placed in the use of translation models. We exist in an increasingly connected and virtual world, one where innovation can be crowd sourced as easily as production can be outsourced. Diffusion’s greatest failing in this regard is that it counts on the transfer of initial product knowledge; initial product awareness in today’s digital culture however is increasingly difficult to locate temporally. The lack of appreciation for the efforts of the firm in moving the innovation through the network is corrected in the translation model, the result is vastly more complicated but at least it doesn’t attempt to shroud what is actually happening.
1 (Rodgers, 1962)
2 (Bass, 1969)
3 (Law, 1984)
4 (Schaffer, 2008)
5 (Latour, 1987)
6 (Brandt, 2007)
Bass, F. (1969). A new product growth model for consumer durables. Management Science, 215-227.
Brandt, A. M. (2007). The Cigarette Century. New York: Basic Books.
Geroski. (2000). Models of Technology Diffusion. Industrial and Corporate Change, 603-625.
Latour, B. (1987). Science in Action. Boston MA: Harvard University press.
Law, J. a. (1984). Science for social scientists. London: Macmillan Press LTD.
Rodgers, E. (1962). Diffusion of Innovations (1st edition). London: The Free Press.
Schaffer, S. (2008). The Information Order of Isaac Newton’s Principia Mathematica. Salvia småskrifter.