Chapter 1

Rat Race Dynamics and

Crazy Companies: The
Diffusion of Technologies
and Social Behavior

Jesse H. A usubel

1.1 Introduction
How and why do technologies spread when and where they do? What are the
implications and consequences for the structure, wealth, and management
of human organizations? These expansive queetions were the subject of the
presentations and discussions of the International Conference on Diffusion of
Technologies and Social Behavior, summarized in this chapter. The chapter
is organized under the following headings: empirical regularities; theoret-
ical issues; predictability; roles of time and space; definition of niche and
innovation; selection dynamics; role of marketing; social aspects of diffusion;
globalization of diffusion processes; and applications of diffusion. While the
chapter treats some questions for policy in both the public and private sec-
tors, it emphasizes research needs and opportunities in the diffusion field.
2 Diffusion of Technologies and Social Behavior

The conference represented a convergence and a maturation of studies

of diffusion. A great range of disciplines was represented from both the so-
cial sciences and the natural sciences. There were geographers, historians,
economists, sociologists, psychologists, and political scientists. There were
physicists and mathematicians. Along with researchers, there were also prac-
ticing engineers and managers. The conference was made more special by
the participation of several of the modern pioneers of the exploration of dif-
fusion, including Torsten Hagerstrand, Harold Linstone, Cesare Marchetti,
and Robert Pry, people who have facilitated diffusion research over the years
and provided many of the ideas on which the conference was built.
The first point to address is why the group came together. The an-
swer is the importance of diffusion as a key process in social and economic
change, made powerfully evident by the growing and widespread recognition
of regularities of diffusion processes.

1.2 Empirical Regularities

In a sense, the conference, like diffusion research itself, had an empirical ori-
gin and a phenomenological orientation. Each discipline, each group of re-
searchers, discovered, somewhat independently, diffusion phenomena. One of
the most satisfying aspects of the conference was the presentation of data on
newly charted diffusion processes. There were examples of resins and plastics
from Vladimir Falzman from the Soviet Union and examples in transport by
Veniamin Livshits, also from the Soviet Union. There were two examples on
AIDS. There were examples from Oskar Ullman from the Federal Republic of
Germany in the solar energy area. There were examples on automated bank-
ing from the Netherlands (Paul Diederen and Rene Kemp), electronic mail
from Sweden (Tomas Astebro), chain-saws from Sweden (Johnny Hjelm),
and Catholic saints from Italy (Marchetti). George Modelski presented the
spread of democracy as a global diffusion process. The multitude of exam-
ples is most important. One of the significant features of the conference was
the recognition that there is in fact now a large library of cases of diffusion,
perhaps 3,000 cases that are well-documented and quantified.
One of the major tasks for diffusion research for the next years is the
meta-analysis of ensembles of diffusion processes analyzed in the various
disciplines. Can one undertake some meaningful taxonomy or classification
of the many examples? There might be various criteria, for example, the
time constant of diffusion processes (the so-called "~t"). Other facets to
J.H. Ausubel 3

examine might include relationships of clusters of technologies, relationships

of levels in the system, and pervasiveness of phenomena, as Roberto Vacca
suggested at one point. The job is to look at the patterns of the patterns:
to compare countries, to compare industries, and so forth, according to their
characteristic forms and configurations of technological innovation and diffu-
sion. The objective would be to identify and eventually explain differences,
similarities, and congruences of diffusion processes and their causes.
Whatever discipline one comes from, we all now have a rich empirical
library on which to draw. There is, however, a need to make this library and
its raw data more accessible.

1.3 Theoretical Issues

While there is excitement and satisfaction with the construction of the empir-
ical base, there is considerable questioning of the adequacy of the theories
resting upon the data. In each field, theoretical models have been devel-
oped. For example, in economics new mathematical approaches to treat
the complex dynamics of diffusion and selection processes and the collective
behavior of economic agents were presented by Giovanni Dosi and Gerald
Silverberg. They also referred back to earlier contributions in economics
by Josef Schumpeter, Edwin Mansfield and others. In geography, theories
and models of Hagerstrand and others were mentioned. Vijay Mahajan pre-
sented an overview of diffusion models from the perspective of marketing and
management science. New models were shared at the conference as well, by
Heinz-Dieter Haustein and in a paper by Ove Granstrand. But, one senses
that there is considerable dissatisfaction about the theoretical base. We do
not feel able to explain the phenomena well. We do not feel we understand
mechanisms.
At the same time, there seems to be an acceptance of a vocabulary for
talking about the phenomena. The vocabulary is largely derived from the
field of biology. There is not too much debate about the usefulness of the
biological metaphor. Harvey Brooks presented the metaphor clearly and suc-
cinctly, describing principal features of the evolutionary process as generally
understood in life sciences and extending them to the seemingly inanimate
world of technological objects. Other biologically-derived ideas were pre-
sented as well, for example by Michael Sonis who suggested examining the
usefulness of the competitive exclusion principal and principals of collec-
tive behavior, not based on optimization. There were other suggestions of
4 Diffusion of Technologies and Social Behavior

this kind, but certainly no field represented seemed to be satisfied with its
theoretical base.
At a general theoretical level, I was reminded of the statement by the late
Elliot Montroll, a well-known American physicist, from 1978, that "Evolu-
tion is a sequence of replacements" (Montroll, 1978). That statement is, for
the most part, consistent with research presented in the conference. There
may be some examples to contest, but Montroll's assertion is the kind of
theoretical hypothesis that might be used generally in diffusion studies. It
also reminds researchers of the limits of focusing on only a single diffusion
process in which a new technology replaces an old one. As Arnulf Griibler
emphasized, evolution is composed of a series of interlaced and multiple
diffusion processes, characterized by various driving forces and adoption en-
vironments.
Among the most promising directions for the search for theory appears
to be the field of communications, as well as biology. In communication, the
aspect stressed was networks. Networks were raised in different forms by
several speakers, for example, Kirk-Jan Kamann and Peter Nijkamp from
the Netherlands. Gerhard Rosegger, talked about a particular type of net-
work, the curious mix of formal and informal alliances that give shape to the
global automobile industry. Hagerstrand mentioned the need to research the
architecture of social communication and relayed the wonderful quote, "First
I make friends, then I make business." There were also the AIDS examples
(Emilio Casetti and Cindy Fan), which stressed very much the importance
of networks. Here is an area in which the existence of networks, the revealing
of social networks over the last few years as the epidemic has spread, has
been readily apparent. lllumination may come from understanding commu-
nication in more detail, as manifested in spatial and other characteristics of
networks.
The question of networks is intimately linked to the question of people's
behavioral rules. A relevant insight comes from The Book of The Courtier,
by Baldesar Castiglione, one of the first advisors to policy makers (Cas-
tiglione, 1528). Castiglione did not emphasize analytic processes in explain-
ing diffusion. "Usage is more powerful than reason in introducing new things
among us and in blocking out old things and anyone who tries to judge of
perfection in such matters is often deceived." That was published in 1528.
Perhaps, as Marchetti would say, the rules of the game do not change very
much.
The need to understand the filtering and acceptance of messages evokes
a remark about causality. There was discussion at the conference, and I will
J.H. Ausubel 5

return to this later, about the meaning of space or location. Ake Ander-
son raised the issue of space. Students of philosophy may remember that
the 18th century Scottish philosopher and historian David Hume had three
principles of causation: first, temporal precedence of cause over effect; sec-
ond, spatial contiguity; and third, constant conjunction. In the context of
diffusion, it may be useful to revisit ideas about causation from philosophy
and other fields. What are the underlying principles and requirements for
the reproduction and spread of technology?
As mentioned already, another field which merits a careful diffused look
is biology. It is clear that there are exciting developments going on in the
life sciences that bear both directly and indirectly on diffusion. Within the
discipline, there is a highly developed vocabulary of passive and facilitated
diffusion, active transport, and membrane inhibitors. There is also poten-
tially relevant work with regard to the role of messenger chemicals (of course,
the messenger RNA). In neurosciences, there have been fascinating discov-
eries about so-called cell adhesion molecules that also transfer messages and
guide cell-cell interaction.
Vladimir Rudashevsky commented in the conference that diffusion is a
process of mutual exchange. That statement can certainly be fruitfully re-
flected upon by our biologist colleagues, who would probably agree and have
many examples at the cellular and genetic level. Back at the level of human
society, Marchetti made the remark that people diffuse in and out of scien-
tific and other sectors based on the rates of difficulty and success in finding
new things. As the effort to develop a theoretical basis for diffusion studies
continues, individual researchers must look beyond their own disciplinary
borders to areas of communication theory, biology, and other fields.
There was much discussion about the importance of diversity. Diversity
again recalls biological issues, this time those current in the vast undertaking
proposed to map the human genome. Most biologists think that much of the
genome may contain little information, or is junk as the mappers sometimes
say. There may be only particular segments, or sequences of genes, that
are vitally important. There may be some metaphors here with economic
and social processes that are worth exploring. Dosi and others noted the
importance in economics of the heterogeneity of preferences, expectations,
and competencies, and that the heterogeneity is required for evolution. What
seems to be true at the genetic level needs to be understood and appreciated
better at the economic and industrial levels.
Now this seemingly inefficient evolutionary model introduces some awk-
wardness for traditional economic theory. It runs counter to some of the
6 Diffusion of Technologies and Social Behavior

preferred assumptions in economics for the last three or four generations

about perfect information, rationality, and existence of equilibria. Another
aspect that came through quite clearly in several talks is that the role of
prices from a diffusion point of view is unclear. Prices convey certain kinds
of information and send certain signals, but their roles and causal function
from a diffusion perspective would seem to be less important, especially in
the early phases of diffusion, than mainstream economic theory would say.
What emerges is the importance of specific history, captured in the phrase
that Brian Arthur (1988) has popularized and that was mentioned by several
speakers, including Yuri Kaniovsky, "path dependence."
Castiglione was an early supporter of path-dependency. Let me share
another quote, which relates to some of the marketing questions discussed
in the conference. "Custom often makes the same things pleasing and dis-
pleasing to us. Whence it comes about that customs, dress, ceremonies,
and fashions that were once prized become despised and contrariwise the
despised become prized." History matters. Where you are will affect where
you will be the next day. Where you are now is determined by where you
were before. You do not and cannot reshuffle all the cards everyday.
Continuing on these general issues of methods and theory, certainly one
of the debates is about the literal use and exercise of curve-fitting. Many
examples of curve-fitting are seen in the diffusion literature. There are sim-
ple examples taken from biology, the growth-to-limits behavior of bacteria.
There are the Fisher-Pry competition models. There are the multiple sub-
stitution models. There are logistic functions, Gompertz functions, and
modified exponential functions. There are also questions raised about the
applicability of each kind of curve-fitting to each case. Moreover, does the
curve-fitting in practice usually apply to cases that are so simple as to be
trivial? There is no resolution of that debate, other than to say, use a model
that makes sense and enjoy the fact that some simple relationships are deep
and important.

1.4 Predictability
The sight of good fits of data and the scent of a theory raise hope of pre-
dictability. There was disagreement in the conference about the extent to
which understanding of diffusion enhances ability to predict. A majority was
on the positive side, saying there is quite a lot of predictability, but there
J.H. Ausubel 7

were several strong caveats. Haustein reported one example where predic-
tion based on a diffusion theory failed. Was it a failure of diffusion theory
in general or of a particular model or application? Robert Ayres presented
a possible counter-example in the area of motor fuels, where the product
under study did not follow a simple S-shaped path. John Tilton shared the
problems of using diffusion for prediction in multiple substitution, with a
case of six competing kinds of beverage containers. How strong are current
tools in the face of real world complexity?
At the same time, there were promising papers about methods in pre-
dictability. Vacca and Valerio Franchina presented a method which is a
virtually completely automatic procedure using triplets - three data points,
including early data points out of the chaotic, turbulent, initial phase of
technology innovation - to estimate growth processes, and showed several
striking examples, excellent fits. They say it appears to be a more objective
process than others that have been employed. Alain Debecker also presented
some interesting rules of thumb about the accuracy of prediction, certainly
handy for practitioners to use. The prerequisite for most analyses though
is that the process already has to be visible, signalling to the human eye.
There has to be a reasonably high level of signal. The crystal ball cannot
be completely empty or cloudy. There has to be a reasonable number of
data points in it. Underlying any look into the crystal ball are diverse views
about the extent to which natural and social systems are chaotic, tending
toward certain forms of (self-) organization, or more strongly deterministic.

1.5 Time and Space

The question of predictability naturally raises the question of time. Follow-
ing is a quote from another distinguished physicist, Robert Herman. Back
in the late 1940s and early 1950s, Herman and Ralph Alpher were responsi-
ble for predicting the background radiation of the universe, the black body
radiation, which led to the Big Bang Theory. This was certainly a discovery
that would reinforce the arrow of time. Herman is said to have commented
subsequently, "Time is the disgusting coordinate of the universe." The con-
ference participants heard a lot about time, mostly as a strict clock ringing
the hours of diffusion.
Scientists who build models of the global climate talk about processes,
like the changes in the ice caps, as having slow physics and other processes,
like formation of thunder clouds and storms, as having fast physics. Social
8 Diffusion of Technologies and Social Behavior

processes may also be said to have slow and fast physics. While transporta-
tion infrastructures spread over intervals of fifty years or more, as shown by
Nebojsa Nakicenovic, clothing fashions diffuse worldwide in a few months. In
fact, it was suggested by Gru.bler that diffusion processes have a hierarchical
structure, perhaps a fractal structure. Alternatively, there is a continuum of
parameters on short and long time scales. To what extent is there measur-
able structure within the temporal dimension of technological diffusion?
Perhaps the most widely discussed feature of temporal diffusion is the
bunching of innovations. Like a baby boom there appear to be periods of in-
creased birth activity, when clusters of technologies are initiated together for
certain reasons that we may not understand. Under what circumstances are
there multiple innovations or clusters? Similarly, is there a focusing of diffu-
sion phenomena at the end of certain time periods? On the one hand, there
were proposals that there are bouquets of innovations, and on the other
hand, there are forces that act as lenses or cones that appear to concen-
trate diffusion processes. So, there are most interesting phenomena in time.
Of course, the question was raised if technological life-cycles are becoming
shorter. Is there an acceleration taking place such that each successive in-
novation spreads more rapidly? Is there an acceleration taking place within
the hypothesized 50-year long-waves or pulses of economic growth such that
innovations taking off later within such a time frame diffuse more rapidly?
Time is an important issue not only for research. Time is important
because it has implications for equity. It is not only that everybody gets
things, but who gets something first, that matters a great deal. The time is-
sues are also important for education, as was raised by Maryellen Kelley and
some others. There may be mismatching of technology diffusion processes
with the educational system. To what extent does this mismatch stem from
ignorance that might be overcome?
Returning to the issue of whether the average life-cycle is becoming
shorter, I, for one, am unconvinced by the evidence shared at the confer-
ence. It is possible there is such a shortening. However, the situation may
be confused by an increase in the number of diffusion processes. To give
an illustration, the typical supermarket in America now has an average of
18,000 items for sale, whereas 40 years ago it had 2,000 items. It may well
be that the life-span of an average product has not changed much. However,
because there are so many products, there is the illusion of acceleration.
Perhaps changes in quantity and complexity of processes are being mistaken
for an acceleration. This is a question that should be researched.
J.H. Ausubel 9

One more question in the time area relates to the notion of appropri-
ability. This is the ability to prevent others from taking or making use of
technology without authority or right and thereby relates to the capability of
innovators to internalize some of the economic benefits of technical progress.
Appropriability was raised several times by economists. Is appropriability
just a diplomatic word used by economists for the control of diffusion over
time? It seems to be a word that landlords would like and renters might not.
To turn from time to space, there was lively discussion about spatial dif-
fusion. Helga Nowotny asked whether, when diffusion researchers talk about
space, is the meaning of space metaphorical? The meaning appears to go
beyond traditional geographical coordinates. Anderson talked about tech-
nology dissolving the role of contiguity. Several speakers, including Lawrence
Brown, Hagerstrand, Kamann, and Sonis pointed out the need to examine
spatial and temporal diffusion together. It was said that more complete
coordinate systems are needed. This view recalls the question of networks.
Clearly, spatial issues need to be revisited.
Yet, the traditional notion of space can still be important. There is the
question, for example, of diffusion within the single Europe that is foreseen
(Charles Edquist and William Peirce). So, even as new kinds of spatial or
space/time relationships are important, space by itself may still be meaning-
ful. Anderson offered a reminder about the parts of the world that are not
included in rapid diffusion. The relationship of diffusion and development is
obviously critical. Sonis also emphasized special spatial niches, wombs, one
might say, that are needed, areas where technological innovations take hold.
The conclusion is that it would be a loss to abandon completely the study
of space as simply metaphorical or illusory.

1.6 Innovation and Niche Definition

There was forceful debate about what is an innovation and what is a niche.
Anderson warned about perils in theories based on ill-classified phenom-
ena. There was also discussion about the differences between fundamental
and incremental innovation. For example, there was debate about, what is
Computer-Integrated Manufacturing (CIM), what is a Flexible Manufactur-
ing System (FMS), what is just-in-time inventory. Are these innovations?
If not, what are the boundaries of innovations? Certainly, such concepts are
not as clear-cut as are innovations in genetics or the invention of the auto-
mobile. Modelski, with his analysis of democracy as an innovation, raised
10 Diffusion of Technologies and Social Behavior

more generally the question of forms of government and forms of social and
institutional innovation.
The argument about niches had several facets. The work by Dominique
Foray on metal casting raised the issues of the appropriate definition of the
diffusing object and the relationship of the occupant to the niche. There
is also the important question of how to modify the niche. The fascinating
case of the car and the horse was presented. At first, the niche for the car
is replacement of the horse. Then all the horses are gone, and the niche for
the car becomes a new expanding niche which the car itself largely creates.
In short, innovations and niches are themselves interacting and dynamic.

1. 7 Selection Dynamics
Analysis of the population of innovations naturally leads to the issue of
selection. Selection was addressed in a variety of ways. There is the canonical
statement from the Bible, "Many are called, few are chosen." Mahajan said
that the rule of thumb in marketing is that 70 percent to 90 percent of 'all
new products fail. Marchetti said that from his studies looking earlier in
the process, one percent of innovations succeed. Then the question is, what
drives us to experiment under such bad odds? It seems crazy.
Silverberg pointed out that many innovations are inferior and more ex-
pensive than their competitors at the outset. Bruce Guile noted that perhaps
the way out is to recognize the triumph of action over analysis. Maybe some
combination of ignorance coupled with a general attitude of optimism be~
gins to explain the yearly parade of bankruptcies. Dosi and Silverberg argued
that diversity of expectations, including those that may lead to failure, are
required to explain diffusion processes. Also striking was Theodore Modis'
statement that studies done about the history of diffusion are usually the
history of winners. It is a bloody history, and Modis suggested more effort in
counting the casualties, the deaths, the lunatics, and losers. It is important
to understand more about what has happened at the end of life-cycles and
in aborted processes.
Dosi made the perceptive comment that product markets select tech-
nologies, but financial markets select firms. The two processes are not iden-
tical. There were several papers relevant in this area. A paper by Charlie
Karlsson that examined why enterprises adopt technologies emphasized in-
formation channels and frequency of exposure. In a way, his view is similar
J.H. Ausubel 11

to a marketing perspective. Edquist looked at empirical differences in diffu-

sion within the countries of the Organisation for Economic Cooperation and
Development (OECD) and emphasized social and cultural factors. Kelley
and Brooks offered a study in the manufacturing technology area trying to
understand a particular sequence of adopters among firms. Of course, there
is the question of the role of entrepreneurs. It would seem that there are
many open questions in the area of selection.

1.8 Role of Marketing

By now marketing has been mentioned several times. Marketing is an at-
tempt to change or modify selection. Marketing people may be considered
true aspiring bio-engineers of human society. But, one may also raise the
question whether marketing matters. We saw several comparisons between
the Western economies and the Soviet Union with some strikingly similar
diffusion data. One has to ask the question, given such different marketing
and distribution strategies and channels, how is it that some results are so
similar? Perhaps marketing is the pageantry, the flags flying in the proces-
sion of the king. To use a different metaphor, perhaps it is the navigation
system for a largely pre-set trajectory. Marketing may only rarely change
the niche to be filled.
It is also important to ask how marketing changes over technology life-
cycles. Thomas Lee raised this issue well. One idea is that it is necessary
to market in pulses in order to fill successive niches, with information that
needs to be distributed through time in certain ways. There were several
comments, some flippant and others serious, about the importance of word-
of-mouth. This appears to remain the dominant way for people to communi-
cate decisively. Technologies may make the shelves of our marketplaces look
modern, but diffusion processes may show how close today's humans are to
our chattering, oral ancestors.
Cross-cultural comparisons would help show what is deeply similar and
what is superficially different, given similar outcomes of diffusion processes.
It should be possible to do revealing comparative studies of the effects of
marketing strategies for the same products or technologies in different coun-
tries. In such studies, it would be useful to explicitly employ hypotheses
from the diffusion literature. For example, one could explore reasons that
early penetration may be associated with high ultimate levels of market sat-
uration, while cases of later penetration are associated with lower ultimate
12 Diffusion of Technologies and Social Behavior

levels of saturation, as found by Nakicenovic in the analysis of diffusion of

transportation technologies.

1.9 Social Aspects

Quite a few participants raised the question of social factors. Everyone rec-
ognizes that the diffusion of an innovation or several innovations can have
many social effects. Several people noted that there are many causal chains
leading to an overall effect. To give an example, automobiles, which were
mentioned a number of times, replaced horses, diminished the number of
stables, and reduced the number of flies and the amount of solid waste.
They thus may have lessened somewhat the spread of communicable dis-
eases. They helped the growth of suburbs. They reduced railroad traffic
and transformed shipping. They changed the nature of the hotel business.
They diminished the employment of domestic servants. They changed mar-
keting areas. They caused international difficulty over oil resources. They
affected rural life. They made drive-in movies and new kinds of vacations
possible.
There were suggestions that more of the social dimensions of diffusion
should be explored. At the same time, Silverberg and others made the point
that social change represents the combined contributions of many inven-
tions. While the automobile made possible the suburbs, the suburbs may
also have required the telephone. What is visible is almost always the re-
sult of an accumulation of influences, in many cases of smaller innovations.
These smaller innovations are a significant part of the process. The force
of any particular invention or innovation might be quite weak. The phrase
that Silverberg used was the collective nature of technological progress. In
fact, several participants referred to the notion of development trajectories
and the concept of techno-economic paradigms consisting of clusters of in-
terrelated technical, organizational, and social innovations that has been
associated with Christopher Freeman and Carlota Perez (1988).
One issue that might be evaluated more in the future is a social, even a
moral and an aesthetic, one. Diffusion itself (or competition) has countervail-
ing effects, and the balance changes through time. On the one hand, diffusion
is a force for homogenization. On the other hand, it multiplies difference.
We have both the increase in the number of inventions and innovations, and
at the same time the possibilities for greater or wider standardization. In
J.H. Ausubel 13

the discussion sometimes people were talking about innovation and diffu-
sion as something quite subversive that would overthrow and would change
the society. This was evident in discussions and papers of Soviet colleagues
about innovation in the context of perestroika. At other times, there was a
sense that technology and diffusion are a force of standardization, a conser-
vative force in a certain way. These offsetting tendencies might be explored
more. A starting proposition may be that fluctuation generates diversity,
but propagation leads to homogenization.

1.10 Globalization
Another social, and also political and economic, issue that was discussed
was that of globalization. This was raised by Hagerstrand, Lee, Marchetti,
and Modelski. Two aspects of globalization were talked about. There is
the global diffusion of technology, and there is the more specific process of
technology transfer between nations. Research presented indicated isolation
of technology, technological protectionism, fails in the long run. Technology
simply does diffuse globally. The question is then how much one can abet or
retard it. In the East-West context especially, are there more positive ways
to handle the diffusion of technology?
At an abstract level, the issue may be phrased, is there a widely ac-
ceptable way for nations to capture income globally that is attributable to
research and development? It appears that nations behave, or would like to
behave, much like the successful firms described in the Silverberg model. It
is therefore not surprising that there is growing international debate about
equitable and optimal national levels of investments in R&D. It is the appro-
priability question in another guise or at a higher level of the system. The
critical point is whether the diffusion of innovation and its benefits remain
ordered in a civilized way, without wars and other violent conflicts referred
to by Modelski and others.
In globalization, with regard to technology transfer, the issue is catch-up.
To what extent can the introduction and diffusion of innovations be acceler-
ated, especially to developing countries like China? Sergei Glaziev's paper
had a good, provocative phrase, "the advantages of the backward." There
is a need to understand better what the potential advantages of the back-
ward are. Some examples of acceleration of diffusion processes were shown;
these appeared to follow a kind of learning or experience curve in which late
adopters were spared some of the time-consuming experimentation of the
14 Diffusion of Technologies and Social Behavior

early adopters. To what extent does this hold for sophisticated, as well as
trivial, cases?
Ultimately, it must be asked, "After catch-up - What?" It is unrealistic
to think that everybody, at some moment in future history, whether firms or
countries, will be at a relatively even point in the adoption of technologies. It
is even less realistic to expect that such parity could be sustained. Diffusion
phenomena in a way seem to be elastic, with some leaders always pulling
away, but then a pack of followers periodically coming nearer, only for a
leader, sometimes a new one, to pull away. Although not frequently, the
ordering of diffusion processes does change among nations, and the desire
to lead, to take the lead, is the essence of competition in politics, as well as
business.
To return to the status of the less developed countries, one of the trou-
bling features of the conference was that, while there is a large library of
evidence on diffusion processes, quantitatively documented cases from de-
veloping countries are scarce. The past behavior of countries like the United
States, Sweden, the United Kingdom, and increasingly the Soviet Union has
been studied extensively. In contrast, there are few data and analyses from
India, China, Argentina, and other such countries. On the empirical side it is
urgent to do more work on less developed countries. It would help crucially
to answer questions about whether and how catch-up occurs.

1.11 Applications
The last area for comment is applications of diffusion for policy, especially
in firms and the national economy. The prospects for applications were ad-
dressed by several people, including Brian Sullivan and Guile. Judgements
here are closely related to views on the predictability question. Sullivan
points out that application of diffusion methods is itself diffusion and adop-
tion. It might be called the diffusion of diffusion. Presumably there should
be a competitive edge from employing the kind of ideas and analyses shared
at the conference, at least until they reach saturation. There was debate
about whether there is a vicious circle in use of diffusion research, in that
if the information is widely publicly available, it may no longer be valu-
able. This debate resembles other debates about the value of information in
markets.
At the same time, there are broader and probably useful guidelines about
behavior related to diffusion that appear to emerge from the presentations
J.H. Ausubel 15

and papers. For example, there is the evidence from Griibler and Nakicenovic
that late adoption is associated with fast diffusion, but a relatively low satu-
ration level. There is also evidence of seasons of saturation, when many
diffusion processes concurrently reach their culmination and create both
economic stress and windows of opportunity for initiation of new growth.
Rudashevsky explored how such macro patterns of d,iffusion link to issues
of restructuring or perestroika. Thus, at a conceptual level, there may be
useful notions for policy and applications apart from specific predictions.
At the micro-level, there are, however, potentially specific applications
of diffusion analysis. Among these is the remarkably precise application
from Vacca and Franchina in the AIDS area. The question is if it would be
socially robust to ac_cept or use such a prediction? How much should it be
relied upon? Should the health minister of Italy risk the entire government
health care system on this prediction? Would even supporters of diffusion
theory prefer to place, say, half of their resources on their own forecast and
then diversify a little? Even if one has a high level of confidence in the
diffusion-based prediction, strategically and tactically, how is it best to deal
with this faith?
Both at the macro and micro level, the system almost always seems
to keep away from homogeneity. Dosi made the playful comment that for
growth it is important to have many economic agents grossly uninformed
about the future. This appears to be a general prerequisite for evolution.
It may also assuage any fears that experts may have that everyone will
adopt or act upon their ideas. Linstone emphasized the benefits of multiple
perspectives, of exploration of several paths. Shunsuke Mori provided a
valuable caution about the extent to which one enterprise, or presumably
one nation, can effectively manage the whole process from invention through
innovation and diffusion. There may always be gaps between intention and
performance in application, whether due to inherent capability or external
surprise. The environment may be so turbulent that one is unable to pursue
a strategy over a long enough period for it to matter.
Guile stressed the delicacy of timing for successful applications of diffu-
sion analyses. He talked about possible mismatches, the need for matching
time scales, and the need for the fertility of the receptor. To illustrate, here is
a quotation about the fax machine, which has proliferated rapidly worldwide
since about 1987. "One possible extension of electrical invention ... is fac-
simile transmission. Newspapers have been thus sent from New York to San
Francisco ... Other uses of the same mechanism are for sending news pic-
tures, identifications of criminals, x-ray photographs, weather maps, signed
16 Diffusion of Technologies and Social Behavior

documents, chemical formulae, graphs and messages in other alphabets and

symbols." This was an exactly correct statement by an analyst, a technol-
ogy assessor, a distinguished sociologist ... in the year 1933 (Ogburn and
Gilfillan, 1933). Of course, the simple fax machine concept was 54 years too
early from the point of view of diffusion. A related cluster of innovations,
including high quality telecommunications, was required to potentiate the
fax. Certainly, timing is critical in commercial application.
Finally, in the applications area, one must raise again the question that
has been posed in the work of Henry Ergas (1987) and also raised by Edquist.
At the government level, what specifically constitute diffusion-oriented poli-
cies? Peirce wondered whether governments have at their disposal appropri-
ate policies that can influence diffusion processes at all. Diffusion-oriented
policies is a most tempting term and needs to be explored and clarified.
To conclude, let me recall two of the best phrases used in the confer-
ence. One participant (Kamann) referred to diffusion processes as "rat-race
dynamics." Another (Dosi) commented on the necessity for "crazy compa-
nies." Tilton asked the question what has changed in 20 years of diffusion
studies in economics and other fields. My answer is that we have sorted a lot
of order from the apparent chaos of social behavior, but have also recognized
better the necessity of disorder, and we are trapped as ever in the race. But,
as shown in Figure 1.1, our paradigm is gaining.

102

Keynesian Schumpeterian
10 1 Classical 0.90

......... 0.70 ..."T1

-
u.. I»
n
.... 10° 0.50 o·
r+

:::J
'-'
u..
0.30 "T1

10- 1 0.10

10- 2
1800 1850 1900 1950 2000 2050

Figure 1.1. Diffusion of techno-economic paradigms.

J.H. Ausubel 17

Acknowledgments
The author gratefully acknowledges suggestions from Arnulf Griibler, Robert Her-
man, Peter de Janosi, Nebojsa Nakicenovic, and Jerry Weisbach and the assistance
of Margret Holland.

References
Arthur, W.B., 1988, Urban systems and historical path dependence, in J.H. Ausubel
and R. Herman, eds., Cities and Their Vital Systems: Infrastructure Past,
Present, and Future, National Academy Press, Washington, DC, USA.
Castiglione, B., 1528, The Book of the Courtier, translated from the Italian by
Charles S. Singleton, Anchor Books edition: 1959, New York, NY, USA.
Ergas, H., 1987, Does technology policy matter? in B.R. Guile and H. Brooks, eds.,
Technology and Global Industry, National Academy Press, Washington, DC,
USA.
Freeman, C. and Perez, C., 1988, Structural crises of adjustment: Business cycles
and investment behaviour, in G. Dosi et al., eds., Technical Change and Eco-
nomic Theory, Pinter Publishers, London and New York.
Ogburn, W.F. with Gilfillan, S.C., 1933, The Influence ofInvention and Discovery,
Report of the President's Research Committee on Social Trends, in Recent Social
Trends in the United States, McGraw-Hill, London and New York.
Montroll, E.W., 1978, Social Dynamics and the Quantifying of Social Forces, Pro-
ceedings of the US National Academy of Sciences, Vol. 75, No. 10.