The Competitiveness of Nations
in a Global Knowledge-Based Economy
October 200
3Joel Moykr *
Punctuated Equilibria and
Technological Progress
American Economics Review
80 (2), May 1990
350-354
Charles Darwin and Alfred
Marshall were both extremely influential men, whose influence has waned
somewhat in recent decades without diminishing in any sense their stature in
their respective disciplines. Marshall
and Darwin shared a mutual interest in each other’s science. Marshall suggested that “the Mecca of the
economist lies in economic biology,” though he never explained what he exactly
meant by that. Recently, economists and
biologists have found some common ground, as in their shared interest in game
theory (John Maynard Smith, 1982), and in adaptive system analysis of
disequilibria (Richard Day, 1987). Analogies
between economics and biological sciences have become more popular among
economists in recent years, and the number of books and journals published with
the word “evolution” figuring in the title is rising rapidly. Physical and mechanical metaphors are coming
under fire (Philip Mirowski, 1990).
One isomorphism that thus
far seems to have escaped notice is that both economic historians and
evolutionary biologists have been debating for years whether past changes in
the objects of their studies have been smooth and gradual or whether history
has moved in leaps and bounds. Darwin
and Marshall both believed that nature does not make leaps. Both were influenced by a long and venerable
tradition that harked back to Leibniz rooted in the Aristotelian notion of the
continuity of space and time (Alexander Gerschenkron,
1968). Darwin and Marshall, however,
were concerned with information systems, and here the absence of leaps is less
obvious.
The term evolution, after
all, has two distinct meanings: one is a dynamic system of natural selection
imposed on blind variation; the other is evolution as distinct from revolution,
gradual as opposed to abrupt. But how
evolutionary was evolution? Economic historians,
and their equivalents in the biological world, the paleontologists and population
geneticists, have been debating for a long time whether the past demonstrates
that change is gradual or not, and whether at times novelties emerged without
clear-cut precedence. Recent thinking
among historians interested in the history of technology and long-term economic
change seems to favor once again the basic notions of gradualness (Eric Jones,
1988) and precedence (George Basalla, 1988). In what follows, I reexamine the “continuity
in history” question in the context of technological change, using the analogy
with evolutionary biology as a mirror.
What genetics and
technology have in common is, above all, that they are decentralized, self-organized
systems of information that are transmitted from generation to generation. The differences between them are quite
obvious: biological information is transmitted through parental genes, and acquired
characteristics are not passed on. In
all cultural evolution, including technology, information is transmitted
laterally and diagonally as well as vertically. Cross-lineage hybridization, common in all
cultural evolution systems, is rare in the biological world. The time scale of change is totally different.
There are complex and subtle
difficulties with the analogy, some of which I discuss elsewhere (1989, 1990b).
All the same, the systems are inherently
comparable and something can be learned from the analogy.
Information systems
resist change, or else they would quickly degenerate into anarchy. [1]
* Departments of Economics and History, Northwestern
University, Evanston, IL 60208. The
comments of Louis Cain, Paul David, and Steven Matthews are gratefully acknowledged.
1. In biological change, the vast majority of all mutations are weeded
out by nature as birth defects. Moreover,
structural constraints limit the extent to which new genotypes can differ from
the status quo. In techno-[logical systems, the equivalent of structural
genetic constraints are production externalities recently emphasized by Paul
David (1988) and Brian Arthur (1989).
Furthermore, as technological progress is never Pareto superior,
resistance to change often resulted in Luddism, and
political struggles between winners and losers (Mancur
Olson, 1982).]
HHC: [displayed]
on page 351 of original.
350
Yet total stability is not
a defining characteristic, or else there would be no history. Observed changes in genetic and technological
systems can be classified into four different classes:
a) Phenotypical changes
without genotypical causes. Much observed variation is not the result of
different information, but of a reversible response of the phenotype (visible
features) to changes in environment, such as shedding a fur. The equivalent in economics is a movement
along a production function, that is, a choice between known technologies
responding for instance to changes in factor prices or to seasonality.
b) Changes in gene
frequency or dispersion of existing information, a process governed by natural
selection. Given that there is some
genetic variation to start with, the genes of the species with the highest
fitness (i.e., reproducing the fastest) will come to dominate the population. This will be recognized readily as the
analogue of the diffusion process of new technologies. Diffusion will eventually peter out if no new
inventions are forthcoming. The
description of natural selection by a population geneticist as “a fire that
consumes its own fuel” (Richard Lewontin, 1982, p.
151) is thus apt for both processes.
c) Mutation, a change
in the genotype, is analogous to the emergence of new ideas. Here, too, the analogy should not be interpreted
literally because in nature mutations occur as copying errors in the DNA,
whereas new ideas, though highly stochastic, have an element of intentional
directionality in them. Most mutations
are rather small and can hardly be regarded as the beginning of new species,
just as most inventions are individually insignificant cumulative improvements
on existing technologies.
d) A very small
minority of mutations result in the generation of new species, a process known
as speciation. A felicitous metaphor
coined by the geneticist Richard Goldschmidt (1940) for macromutations
(that lead to the evolvement of a new species) is “hopeful monsters.” The metaphor seems as apt a description of an
early model of the Newcomen steam engine or the early
power loom as one of a mutant living being. Hopeful monsters are rare in technology as
they are in living species, and successful ones far
rarer. By analogy, I will define macroinventions as technological breakthroughs that
constitute discontinuous leaps in the information set and create new
techniques.
Changes
of types (a), (b), and (c) are consistent with the gradualist axioms. Phenotypical change tends to be short-run in nature, and
rarely involves very dramatic changes. Switching
to a different kind of grain in response to a pest, or to a more labor-intensive
technique as a response of population growth, are examples of such changes.
Similarly, the continuity of diffusion processes is preserved in our S-shaped diffusion
curves.
The emergence of new
techniques, however, has at times taken the form of dramatic changes. Such heroic interpretations of history have been
properly criticized as oversimplified, but they contain a kernel of historical
truth. The biological analogy of such
abrupt changes is speciation. Speciation
played a relatively small role in Darwin’s biology, but recent developments
have restored it to its rightful place in the center of evolution. Paleontologists such as Gould, Eldredge, and Stanley have rebelled against the ruling neo-Darwinian
orthodoxy that insisted that the evolutionary process works in small and
continuous increments. Instead, they maintain,
Goldschmidt was fundamentally right when he argued that evolution proceeded in
fits and starts (Stephen Gould, 1982). Long
periods of stagnation and very slow gradual change were punctuated by
feverishly rapid changes in which large numbers of new species emerged, a
process known as adaptive radiation.
My argument is simply
that much of the record of the economic history of technology displays -a
similar dynamic pattern of long
351
periods of stagnation or very slow, change, punctuated by
sudden outbursts like the Industrial Revolution. Evolutionary systems driven by the interaction
of radical innovations and smaller innovations, and diffused by natural selection
do not necessarily advance in a smooth and gradual way. Self-organized information systems tend to be
nonlinear, and cannot be characterized in any sense by linear stability.
A few remarks are in order
about these macroinventions. Some of them were genuine “eureka” type of
breakthroughs in which a single inventor revolutionized a whole technical mode
in one bold stroke. In other cases, a
second invention of major importance had to be made to perfect the first macroinvention. [2] The relation between macro- and microinventions
is one of complementarity. Such complementarities occur in biological
evolution as well, although they take a different form. It seems incomplete, however, to argue only
that “the major innovations are made possible by numerous minor innovations” (Devendra Sahal, 1981, p. 37). Instead, the reverse seems more important. Radical advances in the manipulation or
understanding of physical processes are usually the beginning, not the end, of
a prolonged process of improvements and modifications.
While these microinventions may well have been the principal source of
measured productivity growth, without the original macroinvention
there would be nothing to improve. The
chief importance of radical inventions is that they raise the marginal product
of effort in development, and thus lead to a sequence of further improvements. An implication of this theory is that, in
periods of radical inventions, we observe an intensification of smaller
inventions as well.
On the other hand, a
radical insight is not enough. Just as a
mutant who survives but cannot reproduce is not a successful mutation, the many
inventions made by Leonardo, and seventeenth-century curiosa such as Stevin’s submarine, Branca’s
steam turbine, and Pascal’s calculator cannot be regarded as successful macroinventions because they could not be produced at
acceptable costs. Lewis and Paul’s 1740
invention of the roller replacing human fingers as a yarn-twisting device had
to wait until it was complemented a quarter-century later by Arkwright’s relatively marginal but crucial insight to use
two (instead of one) sets of rollers. Bessemer’s
invention of the converter would have been useless had it not been for Robert Mushet’s addition of spiegeleisen
(an alloy of carbon, manganese, and iron) into the molten iron as a recarburizer. Such
information complementarities are parallel to the complementarities implicit in
biparental reproduction and the phenotypic expression
of mutations in recessive genes.
How does one
distinguish between macro-and microinventions? Devising a metric that maps
a single value of “radicalness” of the novel
information from the set of complex characteristics of each invention is, of
course, impossible. Although no
such exact metric can be devised, some examples should help.
The later middle ages
in the West witnessed a number of macroinventions:
the weight-driven clock, the windmill, and the blast furnace. These three inventions created techniques that
differed enormously from previous usages. The weight-driven clock used a verge and foliot escapement to convert time, a continuous and
unidirectional variable, into oscillating and bidirectional motion. The windmill was the first use of wind power
for anything except sailing and represented a revolution in power transmission.
The blast furnace completely revolutionized
the iron smelting process and made its casting possible.
Most macroinventions between 1500 and 1750 were stillborn, in
part because of structural constraints on workmanship and materials. By 1750, the environment changed and the
Industrial Revolution was, in its barest essence, a clustering of macroinventions (see my 1990a paper). By the criterion of the novelty of information,
the quintessential inventions of the Industrial Revolution were
2. Thus the addition of a separate condenser to an atmospheric engine or
the four-stroke principle to the internal combustion engine must be counted as
part of the macroinvention. The weight-driven clock was vastly improved by
the pendulum.
352
not the mule or the rolling-and-puddling
process, but hot air ballooning and gaslighting. The clustering phenomenon of radical
innovations is widely observed in all cultural processes, and represents a
combination of a conducive environment and
interactions between the agents themselves. In biology, the agents do not respond to
mutations in other species directly, yet each species takes changes in others
as changes in its environment. These
interactions, too, may result in critical mass situations leading to occasionally
intensive and sudden outbreaks of speciation.
The best examples for macroinventions come, understandably, from the second Industrial
Revolution of the later nineteenth-century. One example that neatly illustrates the
differences between micro- and macroinventions was
telegraphy. Before the 1830s, information
could not travel faster than people with the exception of homing pigeons and
semaphore, both clumsy and unreliable systems. Telegraphy, like many other macroinventions, cannot be attributed to a single inventor,
even though it evolved in a short period. The idea that information could travel at the
amazing speed of an electric current occurred in the first decades of the
nineteenth century. [3] Decades of gradual improvements followed.
The first commercial telegraph line was established in London by
William Cooke and Charles Wheatstone in 1837, and the first successful
submarine cable between Dover and Calais was laid in 1851. Yet, long-distance cables were found to be a
difficult technology to master. Signals
were often weak and slow, and the messages distorted. Distortion increased with distance, a problem
known as capacitance. Moreover,
submarine cables were subject at first to intolerable wear and tear. Physicists, above all William Thomson (later
Lord Kelvin), made fundamental contributions to the technology. [4] By 1910, cable technology was still imperfect. In the 1910s, automatic relays were installed,
and later complemented by repeaters that sampled the incoming signal, greatly
increasing accuracy and speed. At about
the same time, it was recognized that by wrapping the core of cables in an
alloy of soft iron and nickel (a process known as “loading”), cables could
overcome the problem of capacitance and carry signals further and three to five
times faster than before. Compared to the
breakthroughs in the 1830s, however, these inventions were marginal.
The last stages of the continuous improvement in cable technology was
more or less contemporaneous with the development of the wireless by Guglielmo Marconi. Marconi’s
success in using Hertzian waves for the transmission
of information and applying Oliver Lodge’s idea of syntony
to fine-tune transmissions has been well-described (Hugh Aitken,
1976). In his work in the 1890s, Marconi
worked with long-wave radio, as it was found that long waves followed the
contours of land. For long distance communications,
long-wave radio needed enormously large and expensive antennas, and could not
compete with cable.
Short waves, because
their behavior was more like light waves, were long believed to be useless in
long-distance communications. Amateurs
kept, however, picking up signals transmitted from inexplicable long distances. In February 1924, Marconi announced a
startling finding: shortwaves were bounced off the
ionosphere, and thus could be used for long-distance telegraphy. Marconi thus “changed the world twice” as Headrick has remarked (1990). Shortwave was to radio what the separate
condenser was to steam, the second stage of a radical breakthrough. The telegraph as a new specie was born in the
1830s; the birth of long-wave radio in the 1900s and of shortwave radio in the
1920s were equally revolutionary.
3. In 1820, André Ampere suggested using the deviation of a magnetic
needle in a telegraph, and in the same year François Arago
established the principle of the electromagnet. Together, these two insights formed the
theoretical basis of telegraphy.
4. In the 1850s, Thomson invented a special galvanometer,
and a technique of sending short reverse pulses immediately following the main
pulse, to sharpen the signal (Damel Headrick, 1990, pp. 215-18).
353
Long-distance
communications thus illustrate the abruptness of technological change. There was no natural transition from semaphore
to the electrical telegraph, nor a gradual movement
from the telegraph to the first radio transmission by Marconi in 1894, nor a
smooth natural development from the long-wave radio used in the first thirty
years to the shortwave systems of the later 1920s. Each of the three systems was subsequently
perfected by a long sequence of microinventions, but
these would not have occurred without the initial breakthroughs. Their concept was novel, they made things
possible that were previously impossible, and they were pregnant of more to
come. Therein lies
the essence of a macroinvention.
We know relatively little
about the causes of macroinventions. Here, too, the similarity with evolutionary
biology holds up. Whereas microinventions largely result from the responses of
research and development to market forces and opportunities, the original
events creating those opportunities are less readily explained. Many macroinventions,
just like the emergence of species, were the result of chance discoveries,
luck, and inspiration. Biologists agree
that certain environments are more conducive to speciation than others. F or example,
speciation tends to occur in relatively small and isolated populations. Similarly, we can identify (though with a much
greater margin of uncertainty) elements that contribute to an environment
conducive to macroinventions. Such a search for the true causes of change
remains speculative, in both the economic history of technology and
evolutionary biology. We cannot hope,
however, to understand the historical changes that really mattered without
realizing that nature makes leaps, from time to time.
References
Aitken,
Hugh G. J., Syntony and Spark: The Origins
of Radio, New York: Wiley & Sons, 1976.
Arthur, Brian, “Competing Technologies, Increasing Returns, and Lock-in
by Historical Events,” Economic Journal, March 1989, 99, 116-1.
Basalla,
George, The Evolution of Technology, Cambridge:
Cambridge University Press, 1988.
David, Paul A., “Path Dependence: Putting the Past into
the Future of Economics,” unpublished paper, Stanford University, 1988.
Day, Richard H., “The General Theory of Disequilibrium
Economics and of Economic Evolution,” in D. Batten et al., eds., Economic
Evolution and Structural Adjustment, Berlin: Springer, 1987.
Gerschenkron, Alexander, “On the Concept of Continuity in History,”
in Continuity in History and Other Essays, Cambridge: Belknap,
1968.
Goldsclunidt, Richard, The
Material Basis of Evolution, New Haven: Yale University Press, 1940.
Gould, Stephen Jay, “The Uses of Heresy,” An
Introduction to the Reprint of R. Goldschmidt’s The Material Basis of Evolution,
New Haven: Yale University Press, 1982.
Headrick, Daniel R, The Invisible Weapon: Telecommunications and
International Politics, 1851-1945, New York: Oxford University Press, 1990.
Jones, Eric L.,
Growth Recurring: Economic Change in
World History, Oxford: Clarendon Press, 1988.
Lewontin, Richard C., Human
Diversity, New York: Scientific American Books, 1982.
Maynard Smith, John, Evolution and the Theory of
Games, Cambridge: Cambridge University Press, 1982.
Mirowski, Philip, More Heat than Light, New York: Cambridge
University Press, 1990.
Mokyr, Joel, “Evolutionary Biology, Technological Change,
and Economic History,” unpublished manuscript, Northwestern University, 1989.
_____ (1990a) “Was There a British Industrial Evolution?,” in his The Vital One: Essays Presented to Jonathan R.
T. Hughes, Greenwich: JAI Press, 1990.
_____ (1990b) The
Lever of Riches: Technological Creativity and Economic Progress, New York:
Oxford University Press, 1990.
Olson, Mancur, The Rise and Decline of Nations, New Haven:
Yale University Press, 1982.
Sahal, Devendra, Patterns of
Technological Innovation, Reading: Addison Wesley, 1981.
354
The Competitiveness of Nations
in a Global Knowledge-Based Economy
October 200
3