The Plan of the Book
In the age of AI, as we shall see, technological progress has become increasingly labor replacing. Thus, to understand the future of progress, we must understand its political economy. The notion that technology can leave groups in the labor market worse off for the rest of their working lives is quite sufficient to justify their resistance to automation. And for governments seeking to avoid social unrest, it is also quite sufficient to justify restricting some technologies. For these reasons, the long run cannot be disconnected from the short run. Our long-run trajectories can be interrupted and changed by short run events, with dismal consequences for our long-term prosperity.
We all know that human history has proceeded very differently in different parts of the world. Economists and economic historians have devoted considerable attention to the question of why some places have grown rich while others have remained poor. This book is not quite as ambitious. It examines why people have fared differently in the places of the world where the frontiers of technology have been allowed to progress throughout the centuries. The relationship between new technology and the wealth of humans has never been tidy and linear. History never quite repeats itself. But as Mark Twain noted, sometimes it surely rhymes. As I write, middle-income jobs are disappearing, and real wages are stagnating, just like in the classic period of industrialization. Of course, the computer technologies of the twenty-first century could not be more different from the machines that made modern industry. But many of their economic and social effects now look exceedingly similar. The Industrial Revolution has made us infinitely wealthier and better off over the long run. AI, similarly, has the potential to make us much wealthier, but just as in the Industrial Revolution, there is concern that it is leaving large swaths of citizens behind, possibly causing a backlash against technology itself. Many observers of current affairs have pointed out that the recent populist renaissance cannot be explained without reference to the losers of globalization. But technology has been just as important in driving down the wages of the middle class. And we have seen nothing yet. As AI becomes more pervasive, so will automation and its effects.
Economic historians have long debated why the technology boom of the 1760s in Britain took so long to produce higher standards of living, and economists are now engaged in a strikingly similar debate about why staggering advances in automation so far have failed to show results in the pockets of average people. This book is an attempt to connect two large bodies of scholarly research to put the Gates paradox in historical perspective. It tracks the expanding frontiers of technology from the invention of agriculture to the rise of AI, tracing the fates of humans as technology has progressed. I should warn the reader that this is not a balanced account. A book of this scope must be selective and carefully prioritize what it discusses. The history of technology is the subject of an extensive literature that I cannot do justice to here. Rather, by reviewing some of the most important technological advances, I shall try to convince the reader that the price of progress paid by the workforce has varied greatly in history, depending on the nature of technological change, and has increased in the twenty-first century—which explains many of the discontents people now face.
The reader should also be aware that because the Industrial Revolution happened in Britain and technological leadership has remained firmly in Western hands since then (though it remains to be seen for how long that will be the case), this book is a Western-biased account. The West caught up with more advanced Islamic and Oriental civilizations only in the fifteenth century. But to paint the contrast of the West before and after the Industrial Revolution, I shall primarily focus on the Western experience. I should also say that most of the history in this book concerns Britain and later America. The simple reason is that the Industrial Revolution first happened in Britain. America took over world technological leadership during the so-called Second Industrial Revolution, and I shall primarily focus on the U.S. experience thereafter. As the economic historian Alexander Gershenkron noted, catch-up growth, which rests on adopting existing technologies invented elsewhere, is fundamentally different from growth that rests on expanding the frontiers of technology into the unknown, and this book focuses on the latter. Some readers may also find it disappointing that many major technological breakthroughs are not even mentioned. To take just one example, the rise of modern medicine has arguably been the greatest boon to humanity but is shamelessly left out here. Technological developments in recent years, including advances in AI, mobile robotics, machine vision, 3-D printing, and the Internet of things, are all labor saving. The purpose of this book is to shed light on present times and challenges facing the workforce today, and for this reason labor-saving technology will receive the bulk of the attention.
It must also be emphasized that though the focus in the later chapters is much on the American experience, technology is not a soloist but part of an ensemble. It interacts with institutions and other forces in society and the economy, which explains why the rise of economic inequality has been less dramatic in other industrial nations over the past three decades. Yet stagnant wages, disappearing middle-income jobs, and a falling labor share of income are common features of Western countries, and they are all related to trends in technology. While there can be no doubt that numerous forces have shaped the income distribution, my focus here is on the long run rather than cyclical matters, and it is about the 99 percent rather than the top 1 percent. And over the grand sweep of history, average people’s incomes have come to depend on technology more than any other factor.
The main challenge this book faces, however, is probably to convince the reader that we can learn from the past. Economists and economic historians alike tend to treat this idea with skepticism. As one anonymous reviewer of this manuscript put it:
Economists are the obvious “History deniers.” They are reluctant to accept that economists could learn anything from the past even as analysed by economic historians. The humbling experience of failing to predict (indeed perhaps unwittingly helping to create) the 2008 financial crisis produced an uncharacteristic expression of interest in economic history as economists sought insight into events that otherwise seemed unpredictable and disturbing. But the interest (and the humility) was temporary and superficial. However, economic historians too are reluctant to claim present-day insight from their studies of the past; it claims too much for their humble discipline. So, both of the disciplines that Frey addresses will be uneasy with his central proposition. Behind this issue is the bigger one of the communication difficulties between these two disciplines. Both have similar technical toolboxes but economics has honed its contents to a fine edge and is hostile to other approaches whereas history’s contents are sometimes not on the technological frontier and have to be used in the context of a narrative. Any author wanting to make the point that we can learn from history to these two different audiences faces serious challenges.
In the remainder of this book I shall nonetheless try to convince the reader that history is more than one damn fact after another. There are broad patterns that we can learn from. When technological progress is labor replacing, history tells us, hostility and social upheaval is more likely to follow. When progress is of the enabling sort, in contrast, and the gains from growth are more widely shared, there tends to be greater acceptance of new technologies. The chapters that follow divide economic history into four episodes. Part 1, titled “The Great Stagnation,” consists of three chapters that concern preindustrial technologies and their effects on people’s standard of living. Chapter 1 gives a succinct summary of advances in technology from the invention of agriculture some 10,000 years ago up until the dawn of the Industrial Revolution. It shows that many significant technologies emerged before the eighteenth century, but they failed to improve material conditions for ordinary people. Chapter 2 demonstrates that though living standards had improved before the Industrial Revolution, growth was predominantly based on trade. The Schumpeterian growth of our modern age,
based on labor-saving technology, creative destruction in employment, and the acquisition of new skills, was not the engine of economic progress. Chapter 3 seeks to explain why this was the case. As we shall see, innovation also flourished at times before the Industrial Revolution, but it rarely served to replace workers—and when it did, it was vehemently opposed or even blocked. A powerful explanation for why the technologies of the Industrial Revolution did not arrive earlier is the widespread opposition to machines that threatened citizens’ livelihoods. The landed classes, whose members controlled the levers of political power, had little to gain and much to lose from replacing technologies, as workers might rebel against the government in fear of losing their jobs.
The second part, called “The Great Divergence,” provides a whirlwind tour of the Industrial Revolution in Britain. It shows that preindustrial monarchs were right to fear the disruptive force of machinery. As the mechanized factory displaced the domestic system, working people raged against the machine. Chapter 4 zooms in on the technologies that made the Industrial Revolution, showing that nearly all of them served to replace workers. Chapter 5 shows that the result was the hollowing out of middle-income artisan jobs, causing a great divergence within Britain—which explains why industrialization brought so much conflict. But the ruling classes now had more to gain from allowing mechanization to progress, and effectively enforced the first machine age on the populace. Workers’ resistance ended only when people began to see their wages rise in the closing decades of the Industrial Revolution.
Part 3, titled “The Great Leveling,” shifts the focus to the American experience. With the Second Industrial Revolution, America took over technological leadership from Britain and the world. The purpose of this part is to examine why the twentieth century did not see the same hostility to mechanization, even as the frontiers of technology were advanced at an accelerating pace. Chapter 6 sketches the technological changes that accompanied the Second Industrial Revolution. It examines the enormous shifts that took place in the labor market as factories electrified, households mechanized, and people left the countryside for mass production industries in the city. We all know that these transitions weren’t painless. Chapter 7 shows how machinery anxiety returned temporarily, as parts of the workforce struggled to adjust as some occupations vanished. But even though concerns about new technologies taking people’s jobs were widespread at times, few people seriously believed that restricting the use of machines was a good idea. Why? America perhaps had the most violent labor history of the industrial world, but after the 1870s, workers rarely, if ever, targeted machines when violence erupted. Chapter 8 is devoted to the question of why labor didn’t oppose machines in the way it did in the nineteenth century. I harbor no illusions that I have succeeded in providing a full answer to that question, but technology is certainly part of the story. A flow of enabling technologies pulled people into new and better-paying jobs in the smokestack cities of the Second Industrial Revolution. As labor began to see technology as working in its self-interest, the rational response became to seek to minimize the adjustment costs imposed on the workforce rather than retarding technological progress. Labor de facto accepted a laissez-faire regime with regard to mechanization but insisted on establishing a welfare and educational system to help people adjust while making individual costs for those who lost their jobs more narrowly constrained. This became the social contract of the twentieth century.
Part 4, called “The Great Reversal,” concerns the era of computers. Chapter 9 shows that the age of automation was not a continuation of twentieth-century mechanization. On the contrary, it was a complete reversal of it. The first three-quarters of the twentieth century has rightly been regarded as producing “the greatest levelling of all time.”48 It was a period of egalitarian capitalism when workers’ wages rose at all ranks, to the point where Karl Marx’s proletariat could join the middle class. In the 1970s, the American middle class had become a diverse blend of blue- and white-collar citizens. Many of these workers tended machines of some kind, in offices and factories. As we shall see, robots and other computer-controlled machines cut out precisely the middle-income factory and office jobs that mechanization created. Chapter 10 shifts the focus from the aggregate to the communities that have seen jobs disappear. Despite the promise of digital technology to flatten the world, it has done the opposite. Since the dawn of the computer revolution, new jobs have overwhelmingly been clustered in cities with skilled populations, while automation has replaced jobs in old manufacturing powerhouses, amplifying the polarization of the American social fabric along geographic lines. And as America has become increasingly polarized along economic lines, it has also become more politically polarized.
Chapter 11 turns to the question of why citizens who have seen their wages fall have not demanded more compensation, as the median voter theorem would predict. If the middle class declines and inequality rises, we would expect workers to vote for more redistributive policies. One reason that they haven’t, I shall argue, is that they have lost political influence. Growing socioeconomic segregation has made people who have suffered hardships increasingly detached from the rest of American society. Meanwhile, the would-be working class, whose members would have flocked into the factories during the postwar boom years, has become increasingly detached from both labor unions and mainstream political parties. The growing populist appeal, it seems, in large part reflects diminishing opportunities for the losers to globalization and automation and the lack of a political response to address their concerns. Globalization has already become a populist target. Looking forward, however, more and more workers are becoming shielded from the force of globalization, as a growing percentage of the workforce is employed in nontradable sectors of the economy. But they are not shielded from automation. If current economic trends persist for several years more or even decades, as they did during the Industrial Revolution, there is nothing that shields automation from becoming a target, as globalization already has.
FIGURE 2: Number of Laundresses in Private Households in the U.S., 1850–1990
Source: M. Sobek, 2006, “Detailed Occupations—All Persons: 1850–1990 [Part 2]. Table Ba1396-1439,” in Historical Statistics of the United States, Earliest Times to the Present: Millennial Edition, ed. S. B. Carter et al. (New York: Cambridge University Press).
Part 5 is titled “The Future,” although it does not attempt to predict what will happen. As discussed above, much depends on the race between replacing and enabling technologies, but obviously the next three decades must not mirror the past three. The idea here is not that we can simply extrapolate from current trends, which is what economists usually do. Nor is my ambition to predict future technological breakthroughs. The best I can do is examine the prototypical technologies coming out of the labs today that have not yet found widespread use. Take, for example, the employment prospects of laundresses, which peaked around 1910—the year when Alva J. Fisher took out a patent for the first electric washing machine, called Thor (figure 2). If economists had extrapolated from the recent past in 1910, they would have inferred that there would be jobs in abundance for people doing laundry in the coming decades. By looking to trends in technology, in contrast (which is what chapter 12 will do), one might have concluded that the electric washing machine would replace laundresses in this task.
After reviewing many recent technological developments, including in machine learning, machine vision, sensors, various subfields of AI, and mobile robotics, my conclusion is that while these technologies will spawn new tasks for labor, they are predominantly replacing technologies and will continue to worsen the employment prospects for the already shattered middle class. Thus, assuming that the positive attitudes toward technological progress of the twentieth century will continue to hold, regardless of how working people fare from automation, is an exceedingly strong assumption. As we shall see, people are already turning more pessimistic about the future and even about automation. A majority of Americans would v
ote for policies to restrict it, and populists may well tap in to growing automation anxiety. How events play out will likely depend on policy choices. To that end, chapter 13 concludes by sketching some strategies and pathways to help people adjust.
PART I
THE GREAT STAGNATION
No craftsman shall think up or devise any new invention, or make use of such a thing, but rather each man shall, out of citizenly and brotherly love, follow his nearest and his neighbour, and practise his craft without harming another’s.
—KING SIGISMUND I OF POLAND
Inequality was pervasive in the agrarian economies that dominated the world in 1800. The riches of a few dwarfed the pinched allocations of the masses. Jane Austen may have written about refined conversations over tea served in china cups. But for the majority of the English as late as 1813 conditions were no better than for their naked ancestors of the African savannah. The Darcys were few, the poor plentiful.
—GREGORY CLARK, A FAREWELL TO ALMS
The wealth of humans is best understood as the cumulative effect of technologies that allow us to produce more with fewer people. Yet before the Industrial Revolution, living standards were less dependent on the spread of technologies that substituted mechanical power for human muscle. This is not to suggest that technological progress commenced in the eighteenth century. Disparities in the adoption of technology across societies reveal substantial technological progress during the preindustrial era. No single event illustrates this better than when the Dutch explorer Abel Tasman discovered Tasmania in 1642, ending the longest isolation of a population recorded in history. Elsewhere in the world, the diffusion of technologies had significantly altered human development. In Tasmania, by contrast, people still lacked agriculture, metal, pottery, fire-making equipment, and even mounted stone tools.1
The Technology Trap Page 4