The dynamic of technological modernity

The essence of technological modernity is non-stationarity: many scholars have ob­served that technological change has become self-propelled and autocatalytic, in which change feeds on change.

To oversimplify, the Industrial Revolution could be reinterpreted in light of the changes in the characteristics and structure of propositional knowledge in the eighteenth century and the techniques that rested on it. Before 1750 the human race, as a collec­tive, did not know enough to generate the kind of sustained technological progress that could account for the growth rates we observe. In the absence of such knowledge, no set of institutions, no matter how benevolent, could have substituted for useful knowledge. Pre-modern society had always been limited by its epistemic base and suppressed by economic and social factors.

How do we explain this change in technological dynamic? In economics, phase tran­sitions can be said to occur when a dynamic system has multiple steady states such as an economy that has a “poverty trap” (low-income equilibrium) and a high income (or rapid growth steady state). A phase transition occurs when the system switches from one equilibrium or regime to another. A simple model in which this can be illustrated is one in which capital and skills are highly complementary. In such models one equilibrium is characterized by rapid investment, which raises the demand for skills; the positive feedback occurs because the increase in the rate of return to human capital induces par­ents to invest more in their children and have fewer children (since they become more expensive), which raises the rate of return on physical capital even more and encourages investment. A second equilibrium is one of low investment, low skills, and high birth rates. A regime change may occur when an exogenous shock is violent enough to bump the system off one basin of attraction and move it to another one. The difficulty with this model for explaining the emergence of modern growth is to identify a historical shock that was sufficiently powerful to “bump” the system to a rapid growth trajectory.

Recent work in growth theory have produced a class of models that reproduce this feature in one form or another. Cervellati and Sunde (2005) for example assume that human capital comes in two forms, a “theoretical” form and a “practical” form, corre­sponding roughly to “scientific” and “artisanal” knowledge or the categories of useful knowledge proposed above. They assume that human abilities are heterogeneous but that there is a threshold at which people start to invest in “theoretical” knowledge as opposed to “crafts”, determined endogenously by life expectancy.

54

The alternative is to presume that historical processes cause the underlying parame­ters to change slowly but cumulatively, until one day what was a slow-growth steady state is no longer an equilibrium at all and the system, without a discernible shock, moves rather suddenly into a very different steady state. These models, pioneered by Galor and Weil (2000), move from comparative statics with respect to a parameter de­termining the dynamic structure, to a dynamical system in which this parameter is a latent state variable that evolves and can ultimately generate a phase transition.^[82] In the Galor-Weil model, the economic ancien regime is not really a steady state but a “pseudo steady state” despite its long history: within a seeming stability the seeds for the phase transition are germinating invisibly.

A similar model, in which technology plays a “behind the scenes” role, is the highly original and provocative model by Galor and Moav (2002). In that model, the phase transition is generated by evolutionary forces and natural selection. The idea is that there are two classes of people, those who have many children (r-strategists) and oth­ers (K-strategists) who have relatively few but “high-quality” offspring and who invest more in education. When “quality types” are selected for, more smart and creative peo­ple are added and technology advances. Technological progress increases the rate of return to human capital, induces more people to have more “high quality” (educated) children which provides the positive feedback loop. Moreover, as income advances, households have more resources to spend on education, which add to further expansion. Again, technology in this model is wholly endogenous to education and investment in human capital, and an autonomous development in the social factors governing human knowledge and the interplay between propositional and prescriptive knowledge is not really modeled.

In one sense Galor and Moav’s reliance on evolutionary logic to explain technological progress is ironic. In recent years it has been realized increasingly that knowledge it­self is subject to evolutionary dynamics, in that new ideas and knowledge emerge much like evolutionary innovations (through mutations or recombinations) and are selected for (or not). Knowledge systems follow a highly path-dependent trajectory governed by Darwinian forces [Ziman (2000) and Mokyr (2005b)]. Yet this important insight still awaits to be incorporated in the “take-off” models of growth theorists. Evolution­ary models predict that sudden accelerations or “explosions” of evolutionary change (known oddly as “adaptive radiation”) occur when conditions are ripe, such as the so-called Cambrian explosion which has been compared to the Industrial Revolution [Kauffman (1995, p. 205)]. Another example of rapid evolutionary innovation is the spectacular proliferation of mammals at the beginning of the Cenozoic following the disappearance of the giant reptiles. The idea that evolution proceeds in the highly non­linear rhythm known as “punctuated equilibrium” has been suggested as a possible insight that economic historians can adapt from evolutionary biology [Mokyr (1990)].

Some of these (and other, similar) models may be more realistic than others, and economic historians may have to help to sort them out. A phase transition model with­out reliance on the quality of children and human capital is proposed by Charles Jones (2001) relying on earlier work by Michael Kremer (1993). In Jones’ model, what mat­ters is the size rather than the quality of the labor force. In very small populations, the few new technological ideas lead in straightforward Malthusian fashion to higher pop­ulations and not to higher income per capita.

Yet the exact connection between demographic changes and the economic changes in the post 1750 period are far from understood, and much of the new growth literature pays scant attention to many variables that surely must have affected the demand for children and fertility behavior. These include technological changes in contraceptive technology, a decline in infant and child mortality, and changing demand for children in the household economy due to technological changes in agriculture and manufacturing. It is also open to question whether and to what extent “numbers matter”, that is, whether the more people are around, the more likely - all other things equal - new technological ideas are to emerge.^[83] The real question is whether the ideas that count are really a monotonic function of population size (Jones assumes a positive elasticity of 0.75 to generate his results), or whether they are generated by a negligible minority and that small changes in the fraction of creative people matters more than a rise in the raw size of population.^[84] The historical record on that is subject to serious debate. It might be added that population growth in Britain was almost nil in the first half of the eighteenth century, and while it took off during the post-1750 era, the same was true for Ireland, where no comparable Industrial Revolution can be detected.

Most endogenous growth historical models, however, depend on the notion that the variable critical to the process of “take-off” or phase transition is investment in human capital.^[85] Historically, however, such a view is not unproblematic either. The idea that the fertility reduction was a consequence of changing rates of return on human capi­tal, especially advanced by Lucas (2002), runs into what may be called the European Fertility Paradox: the first nation to clearly reduce its fertility rate through a decline in marital fertility (that is, intentional and conscious behavior) was not the country in which advanced technological techniques were adopted in manufacturing, but France. In Britain fertility rates came down eventually, but the decline did not start until the mid- 1870s, a century after the beginning of the Industrial Revolution [e.g., Tranter (1985, chapter 4)]. Imperial Germany, which became the technological leader in many of the cutting-edge industries of the second Industrial Revolution, maintained a fertility rate far above France’s and Britain’s.^[86] To argue, therefore, that technological progress was rooted in demographic behavior (through smaller families) seems at variance with the facts. It may well be that this nexus held in the twentieth century, but given the decline in wage premia it is hard to see the rate of return on human capital to be the driving fac­tor. BeyondEurope, of course, population-driven theories of the “the-more-the-merrier” variety must confront the difficult fact that China not only had a population vastly larger than any European economy but that its population grew at a rapid rate in the very cen­tury that Europe experienced its Enlightenment: from a low point of about 100 million in 1685, it exceeded 300 million in 1790, thus experiencing a per annum population growth of 1.05 percent, though admittedly from an unusually low base.

To understand the “phase transition” within the dynamic of useful knowledge, we need to look again at the relationship between propositional and prescriptive knowledge. As the two forms of knowledge co-evolved, they enriched one another increasingly, eventually tipping the balance of the feedback mechanism from negative to positive and creating the phase transition. During the early stages of the Industrial Revolution, propositional knowledge mapped into new techniques, creating what we call “inven­tions”. This mapping should not be confused with the linear models of science and technology that were popular in the mid-twentieth century, which depicted a neat flow from theory to applied science to engineering and from there to technology. Much of the propositional knowledge that led to invention in the eighteenth century was artisanal and mechanical, pragmatic, informal, intuitive, and empirical. Only very gradually did the kind of formal and consensual knowledge we think of today as “science” become a large component of it. It was, in all cases, a small fraction of what is known today. What matters is that it was subject to endogenous expansion: prescriptive knowledge in its turn enhanced propositional knowledge, and thus provided positive feedback between the two types of knowledge, leading to continuous mutual reinforcement. When power­ful enough, this mechanism can account for the loss of stability of the entire system and for continuous unpredictable change.

The positive feedback from prescriptive to propositional knowledge took a variety of forms. One of those forms is what Rosenberg has called “focusing devices”: technology posed certain riddles that science was unable to solve, such as “why (and how) does this technique work”. It has been suggested, for instance that the sophisticated waterworks that supplied power to the famous Derby silk mills established by the Lombe brothers in the 1710s stimulated local scientists interested in hydraulics and mechanics [Elliott (2000, p. 98)]. The most celebrated example of such a loop is the connection between steam power and thermodynamics, exemplified in the well-known tale of Sadi Carnot’s early formulation, in 1824, of the Second Law of Thermodynamics by watching the difference in fuel economy between a high pressure (Woolf) steam engine and a low pressure one of the Watt type.^[87] The next big step was made by an Englishman, James P Joule, who showed the conversion rates from work to heat and back.^[88] Joule’s work and that of Carnot were then reconciled by a German, R.J.E. Clausius (the discoverer of entropy), and by 1850 a new branch of science dubbed “thermodynamics” by William Thomson (later Lord Kelvin) had emerged [Cardwell (1971, 1994)].⁶² Power technol­ogy and classical energy physics subsequently developed cheek by jowl, culminating in the career of the Scottish physicist and engineer William Rankine, whose Manual of the Steam Engine (1859) made thermodynamics accessible to engineers and led to a host of improvements in actual engines. In steam power, then, the positive feedback can be clearly traced: the first engines had emerged in the practical world of skilled blacksmiths, millwrights, and instrument makers with only a minimum of theoretical understanding. These machines then inspired theorists to come to grips with the natural regularities at work and to widen the epistemic base. The insights generated were in turn fed back to engineers to construct more efficient engines. This kind of mutually reinforcing process can be identified, in a growing number of activities, throughout the nineteenth century. They required the kind of intellectual environment that the Indus­trial Enlightenment had created: a world in which technical knowledge was accessible and communicable in an international elite community, a technological invisible college that encompassed much of the Western world.

A less well-known example of this feedback mechanism, but equally important to economic welfare, is the interaction between the techniques of food-canning and the evolution of bacteriology. As noted earlier, the canning of food was invented in 1795 by Nicolas Appert.⁶³ He discovered that when he placed food in champagne bottles, corked them loosely, immersed them in boiling water, and then hammered the corks tight, the food was preserved for extended periods. Neither Appert nor his English emulators who perfected the preservation of food in tin-plated canisters in 1810 really understood why and how this technique worked, because the definitive demonstration of the notion that microorganisms were responsible for putrefaction of food was still in the future. It is therefore a typical example of a working technique with a narrow epistemic base. The canning of food led to a prolonged scientific debate about what caused food to spoil. The debate was not put to rest until Pasteur’s work in the early 1860s. Pasteur claimed ignorance of Appert’s experimental work, but eventually admitted that his own work on the preservation of wine was only a new application of Appert’s method. Be that ously concerned with the economic efficiency of electromagnetic engines...he quite explicitly adopted the language and concerns of the economist and the engineer” [Morus (1998, p. 187), emphasis in original]. As Ziman (1976, p. 26) remarks, the first law of thermodynamics could easily have been derived from Newton’s dynamics by mathematicians such as Laplace or Lagrange, but it took the cost accountancy of engineers to bring it to light.

⁶² Research combining experiment and theory in thermodynamics continued for many decades after that, especially in Scotland and in Mulhouse, France, where Gustave Adolphe Hirn, a textile manufacturer, led a group of scientists in tests on the steam engines in his factory and was able to demonstrate the law of conservation of energy.

⁶³ Experimental work by, among others, the Italian naturalist Lazaro Spallanzani, had earlier indicated that heating organic materials and subsequent airtight flashing would prevent putrefaction. It is unclear whether Appert and his British imitators knew of this work. See Clow and Clow (1952, p. 571). as it may, his work on the impossibility of spontaneous generation clearly settled the question of why the technique worked and provided the epistemic base for the technique in use. When the epistemic base of food-canning became wider, techniques improved: the optimal temperatures for the preservation of various foods with minimal damage to flavor and texture were worked out by two MIT scientists, Samuel Prescott and William Underwood.^[89]

A different feedback mechanism from prescriptive to propositional knowledge was described by Derek Price as “Artificial Revelation”. The idea is fairly simple: our senses limit us to a fairly narrow slice of the universe that has been called a “mesocosm”: we cannot see things that are too far away, too small, or not in the visible light spectrum [Wuketits (1990, pp. 92, 105)]. The same is true for our other senses, for the ability to make very accurate measurements, for overcoming optical and other sensory illu­sions, and - perhaps most important in our own time - the computational ability of our brains. Technology consists in part in helping us overcome these limitations that evolution has placed on us and learn of natural phenomena we were not meant to see or hear.^[90] The period of the Industrial Revolution witnessed a great deal of improvement in techniques whose purpose it was to enhance propositional knowledge. The great potter Josiah Wedgwood maintained a close relationship with the chemist James Keir: while Keir supplied Wedgwood with counsel, Wedgwood’s factory provided Keir with the tubes and retorts he used in his laboratory near Birmingham [Stewart (2004, p. 18)]. The accuracy of instruments that measured time, distance, weight, pressure, temperature and so on increased by orders of magnitude in the eighteenth century.^[91] Pumps and electri­cal machines allowed the study of vacuums and electrical phenomena. Lavoisier and his circle were especially good in designing and utilizing better laboratory equipment that allowed them to carry out more sophisticated experiments.^[92] Alessandro Volta invented a pile of alternating silver and zinc disks that could generate an electric current in 1800. Volta’s battery was soon produced in industrial quantities by William Cruickshank. Through the new tool of electrolysis, pioneered by William Nicholson and Humphry Davy, chemists were able to isolate element after element and fill in much of the detail in the maps whose rough contours had been sketched by Lavoisier and Dalton. Volta’s pile, as Davy put it, acted as an “alarm bell to experimenters in every part of Europe” [cited by Brock (1992, p. 147)]. The development of the technique of in vitro culture of micro-organisms had similar effects (the Petri dish was invented in 1887 by R.J. Petri, an assistant of Koch’s). Price feels that many such advances in knowledge are “adventi­tious” (1984a, p. 112). Travis (1989) has documented in detail the connection between the tools developed in the organic chemical industry and advances in cell biology. These connections between prescriptive and propositional knowledge are just a few examples of advances in scientific techniques that can be seen as adaptations of ideas originally meant to serve an entirely different purpose, and they reinforce the contingent and acci­dental nature of much technological progress [Rosenberg (1994, pp. 251-252)].

The invention of the modern compound microscope in 1830 attributed to Joseph J. Lister (father of the famous surgeon) serves as another good example. Lister was an am­ateur optician, whose revolutionary method of grinding lenses greatly improved image resolution by eliminating spherical aberrations.^[93] His invention and the work of oth­ers changed microscopy from an amusing diversion to a serious scientific endeavor and eventually allowed Pasteur, Koch, and their disciples to refute spontaneous generation and to establish the germ theory, a topic I return to below. The germ theory was one of the most revolutionary changes in useful knowledge in human history and mapped into a large number of new techniques in medicine, both preventive and clinical. Indeed, the widespread use of glass in lenses and instruments in the West was itself something coincidental, a “giant accident”, possibly a by-product of demand for wine and different construction technology [Macfarlane and Martin (2002)]. It seems plausible that with­out access to this rather unique material, the development of propositional knowledge in the West would have taken a different course.^[94]

A third mechanism of technology feeding back into prescriptive knowledge is through what might be called the “rhetoric of knowledge”. This harks back to the idea of “tightness” introduced earlier. Techniques are not “true” or “false”. Either they work ac­cording to certain predetermined criteria or they do not, and thus they can be interpreted to confirm or refute the propositional knowledge that serves as their epistemic base. Propositional knowledge has varying degrees of tightness, depending on the degree to which the available evidence squares with the rhetorical conventions for accep­tance. Laboratory technology transforms conjecture and hypothesis into an accepted fact, ready to go into textbooks and to be utilized by engineers, physicians, or farmers. But in the past a piece of propositional knowledge was often tested simply by verifying that the techniques based on it actually worked. The earthenware manufacturer Josiah Wedgwood felt that his experiments in pottery actually tested the theories of his friend Joseph Priestley, and professional chemists, including Lavoisier, asked him for advice. Similarly, once biologists discovered that insects could be the vectors of pathogenic mi­croparasites, insect-fighting techniques gained wide acceptance. The success of these techniques in eradicating yellow fever and malaria was the best confirmation of the hy­potheses about the transmission mechanisms of the disease and helped earn them wide support.

Or consider the matter of heavier-than-air flight. Much of the knowledge in aero­nautics in the early days was experimental rather than theoretical, such as attempts to tabulate coefficients of lift and drag for each wing shape at each angle. It might be added that the epistemic base supporting the first experiments of the Wright brothers was quite untight: in 1901 the eminent astronomer and mathematician SimonNewcomb (the first American since Benjamin Franklin to be elected to the Institute of France) opined that flight carrying anything more than “an insect” would be impossible.^[95] The success at Kitty Hawk persuaded all but the most stubborn doubting Thomases that human flight in heavier-than-air fixed wing machines was possible. Clearly their success subsequently inspired a great deal of subsequent research on aerodynamics. In 1918 Ludwig Prandtl published his magisterial work on how wings could be scientifically rather than empir­ically designed and the lift and drag precisely calculated [Constant (1980, p. 105) and Vincenti (1990, pp. 120-125)]. Even after Prandtl, not all advances in airplane design were neatly derived from first principles in an epistemic base in aerodynamic theory, and the ancient method of trial and error was still widely used in the search for the best use of flush riveting in holding together the body of the plane or the best way to design landing gear [Vincenti (1990, pp. 170-199; 2000)].

It is important not to exaggerate the speed and abruptness of the transition. Thomas Edison, a paradigmatic inventor of the 2nd Industrial Revolution, barely knew any sci­ence, and in many ways should be regarded an old-fashioned inventor who relied mostly on trial-and-error through intuition, dexterity and luck. Yet he knew enough to know what he did not know, and that there were others who knew what he needed. Among those who supplied him with the propositional knowledge necessary for his research were the mathematical physicist Francis Upton, the trained electrical engineer Hermann Claudius, the inventor and engineer Nikola Tesla, the physicist Arthur E. Kennelly (later professor of electrical engineering at Harvard), and the chemist Jonas W. Aylsworth. Yet by that time access costs had declined enough so that he could learn for instance of the work of the great German physicist Hermann von Helmholtz through a translated copy of the latter’s work on acoustics.

The positive feedback from technology to prescriptive knowledge entered a new era with development of the computer. In the past, the practical difficulty of solving differ­ential equations limited the application of theoretical models to engineering. A clever physicist, it has been said, is somebody who can rearrange the parameters of an insolu­ble equation so that it does not have to be solved. Computer simulation can evade that difficulty and help us see relations in the absence of exact closed-form solutions and may represent the ultimate example of Bacon’s “vexing” of nature. In recent years simula­tion models have been extended to include the effects of chemical compounds on human bodies. Combinatorial chemistry and molecular biology are both equally unimaginable without fast computers. It is easy to see how the mutual reinforcement of computers and their epistemic base can produce a virtuous circle that spirals uncontrollably away from its basin of attraction. Such instability is the hallmark of Kuznets’ vision of the role of “useful knowledge” in economic growth.

In addition to the positive feedback within the two types of knowledge, one might add the obvious observation that access costs were themselves a function of improving techniques, through better communications, storage, and travel techniques. In this fash­ion, expansions in prescriptive knowledge not only expanded the underlying supporting knowledge but made it more accessible and thus more likely to be used. As already noted, this is particularly important because so much technological progress consists of combinations and applications of existing techniques in novel ways, or parallels from other techniques in use. Precisely for this reason, cheap and reliable access to the mon­ster catalog of all feasible techniques is an important element in technological progress. As the total body of useful knowledge is expanding dramatically in our own time, it is only with the help of increasingly sophisticated search engines that needles of useful knowledge can be retrieved from a haystack of cosmic magnitude.

Technological modernity is created when the positive feedback from the two types of knowledge becomes self-reinforcing and autocatalytic. We could think of this as a phase transition in economic history, in which the old parameters no longer hold, and in which the system’s dynamics have been unalterably changed. There is no necessity for this to be true even in the presence of positive feedback; but for certain levels of the parameters, the system as a whole becomes unstable. It may well that this instability in the knowledge-producing system are what is behind what we think of as “technological modernity”. Kuznets, of course, felt that the essence of modern growth was the in­creasing reliance of technology on modern science. This view, as I have argued above, needs clarification and amplification. Inside the black box of technology is a smaller black box called “research and development” which translates inputs into the output of knowledge. This black box itself contains an even smaller black box which models the available knowledge in society, and it is this last box I have tried to pry open. Yet all this is only part of the story: knowledge creates opportunities, but it does not guarantee action. Knowledge is an abstract concept, it glosses over the human agents who possess it and decide to act upon it. What motivates them, and why did some societies seem to be so much more inclined to generate new knowledge and to exploit the knowledge it had? To understand why during the past two centuries the “West” has been able to take advantage of these opportunities we need to examine the institutional context of innovation.

7.