Book Summary: “Learning by Doing” by James Bessen

Title: Learning by Doing: The Real Connection between Innovation, Wages and Wealth.
Author: James Bessen
Scope: 4 stars
Readability: 4 stars
My personal rating: 5 stars
See more on my book rating system.

If you enjoy this summary, please support the author by buying the book.

Topic of Book

Learning from his career in the Technology field, Bessen argues that implementing new technology is very difficult and it takes decades for people to learn how to do it.

If you would like to learn more about how humans have implemented new technologies, read my book From Poverty to Progress: How Humans Invented Progress, and How We Can Keep It Going.

Key Take-aways

  • Invention is only the first step of technological innovation. Far more difficult and time-consuming is implementing a new technology. This process takes decades.
  • Implementation largely comes from trial-and-error experimentation, because new technology is complex.
  • New technologies require technical knowledge which is rare in the marketplace. Therefore, those who are willing to learn those skills earn high incomes.
  • Over time, the technical knowledge get codified and training courses emerge, so lower skilled people can make and use the technology.
  • New technologies also require new organizational strategies, which take a great deal of time to evolve.

Important Quotes from Book

“The past was not so different. There have been long periods in which advancing technology failed to provide much economic benefit to workers and when machines took over job tasks. For decades at the beginning of the Industrial Revolution, factory wages remained stagnant despite major new technologies that brought dramatic gains in output per worker. Wealth was created, but it went into the pockets of the investors, managers, and a few key skilled employees. Profits grew and inequality rose, without bringing much gain to the workers. These trends eventually reversed. After decades, the pay of even uneducated factory workers rose substantially, and workers gained a substantial share of the benefit of the new technologies.

I argue that developing the knowledge and skills needed to implement new technologies on a large scale is a difficult social problem that takes a long time to resolve. It was a difficult problem in the past and remains so today, yet most workers will only benefit once it is resolved. Resolution will take time and the right policies.

Many people fail to appreciate the complexity and slow pace of the current transition because they confuse technology with inventions and they confuse skill with education… Throughout history, workers have acquired their technical knowledge through a combination of formal training and experience. They gained much of their important technical knowledge on the job, through “learning by doing.”

The distinction between invention and implementation is critical and too often ignored.

It takes a long time for technical knowledge to be developed, longer for it to spread, and even longer for institutions to emerge, such as new labor markets, that allow ordinary workers to benefit from their new knowledge. Such learning on a mass scale was and is a difficult problem for society.

Perhaps because it is difficult to acquire, technical knowledge is central to the economic well-being of large numbers of peoples.

That is, markets can also fail at developing new implementation knowledge. Frequently, during their early stages, major new technologies lack the institutions to train workers. Also, labor markets often fail to provide sufficient rewards initially. One common problem is a “coordination failure”: early-stage technologies typically have many different versions.

The insight here is that easily replicable technical knowledge is frequently the product of a long evolutionary process. Replicable knowledge doesn’t just happen; it is made. Institutions, policies, and economic incentives combine in a dynamic process that changes the nature of technical knowledge, and in the process, whole industries may be overturned.

Technology implementation is challenging because large numbers of workers need to acquire new skills and technical knowledge.

Economists’ common practice of defining “skilled workers” as those with four years of college is particularly misleading. In our meritocratic society it is perhaps too easy to associate skill with educational credentials.

The myth of de-skilling is no more accurate in other industries. While skill requirements fell in some jobs in some industries when new technologies were introduced, other jobs required greater skill. Indeed, it is hard to find a clear example where a new technology was uniformly de-skilling during the nineteenth century. Some technologies generated whole new skilled occupations.

Although a number of historians have described the introduction of interchangeable parts as a means of replacing skilled machinists with unskilled operatives, the technique depended very much on the skills of “artificers” who adjusted the parts to make them fit, and these skills took considerable time and effort to develop.

All of these different kinds of learning meant that our company and our customers both made substantial human capital investments—often much greater than the cost of the technology itself—in order to use the software productively.

Moreover, the new organizational structures made necessary by new technologies often require substantial learning on the job. Economists have found that computer use is associated with a variety of changes in workplace organization. Organizations decentralize; they move to team-based work, with new incentives, occupations, and task assignments; workforce training is increased; and computers often come with new products, services, and customer-supplier relationships. Many of these changes require learning by employees or sharing of knowledge about production or about the needs of customers and suppliers.

More generally, economists have found that human capital is responsible for a great deal of the variation in wages from one person to the next. As much as 77 percent of the variation in wages is “explained” by differences in workers’ experience and individual characteristics, including innate abilities and schooling. But only a small part of the variation in workers’ human capital comes from differences in formal education…. much human capital is acquired on the job.

Major new technologies typically go through long periods of sequential innovation, where a string of improvements, new knowledge, and new skills are developed one after the other. Central to this process—and for why it takes so long—is learning by doing.

Decade-long delays are not at all unusual. Technology after technology, many decades passed before an inventive idea was first commercialized, then it took even more decades before the product was commercially successful, and even more before major benefits flowed to workers and consumers.

Why is so much trial-and-error learning needed? Because technologies are complex. Typically, new technologies demand that a large number of variables be properly controlled.

What really distinguished the early humans from other species was not conceiving technology ideas, but perfecting those ideas to deal with the complex natural environment.

An invention that would not have been commercially feasible in 1803 or 1833 became feasible by 1883 and critical by 1903. This illustrates a broader rule I call the “remainder principle”: as technology reduces costs or increases performance on one task in a process or one component in a product, the value of performance on the remaining tasks or components increases.

“These interactions happen because technologies are modularized in complementary parts: complex technological processes are broken into steps and complex products are broken into components in order to manage the development and sharing of knowledge more efficiently. Each module is complementary to the rest, so that improvements in one module increase the payoff to an invention that improves the performance of another module”

Standardization of technical knowledge occurs when knowledge is simplified by limiting the range of technical parameters used to describe and implement the technology. Standardization makes knowledge easier to acquire and communicate by reducing the essential amount of information that must be conveyed.

This, then, is the paradox of technical knowledge: technology creates aggregate wealth for a nation because new ideas can be replicated at low cost, but technology creates wealth for the people of a nation by requiring new technical knowledge that cannot be easily replicated.

Many manufacturing workers during the Industrial Revolution had important new skills and knowledge, but at first their wages did not reflect these skills.

We are accustomed to thinking of new markets emerging rapidly, more or less as soon as a new product emerges or a new need is identified. But that view is true only for standardized products…. Initially, these skills are always limited to specific employers.

The emergence of a robust market for skilled weavers in New England not only led to a dramatic increase in wages of textile workers—refuting Engels—but it also led to a much more effective way for workers to acquire skills and knowledge on a large scale. Standardization meant both that workers shared more in the benefits of new technology, and that new technology could be used more widely.

If job experience is a major requirement for vacancies, then employers are not looking to fill those jobs by hiring entry-level applicants right out of school. Employers want new hires to be able to start contributing, with no further training or start-up time. That’s certainly understandable, but the only people who can do that are those who have done virtually the same job before, and that often requires a skill set that, in a rapidly changing world, may be dying out even as it is perfected.

This catch-22 is a symptom of poorly developed training institutions and labor markets for new skills. Employers limit their investment in training because employees leave; workers limit their investment in training because without standardized skills they lack a clear career path that will provide them a secure return on that investment. And the lack of institutions that can train and certify new skills makes these investments cost more.

While demand for college graduates has grown in relative terms, it appears to be mainly because college-educated workers are better at learning new, unstandardized skills on the job, not because their college educations conferred specific technical skills.

That changed as technologies matured. Many technologies seem to have followed a common pattern: a regime with knowledge sharing and little patenting is followed by a regime with aggressive patent acquisition and enforcement

Competition between users of the new technology was “soft” for two decades because the new technology coexisted with more costly alternatives. Additional power loom manufacturers did not significantly affect prices as long as most of the market was still being supplied by handloom weavers or by expensive imports. In this situation, it did not hurt to share technology with prospective rivals. Sharing only expanded the new technology’s market share. It took two decades for competition between power loom manufacturers to grow intense.

But if textile manufacturers were making large profits, why did not more firms quickly enter the market? An important reason was learning by doing. Acquiring the requisite knowledge took time and was costly. Few people initially knew how to build, install, operate, and maintain the new machinery.

While noncompete enforcement may be beneficial to established firms, it inhibits the development of early-stage technologies.

Not all markets in all developing nations end up being dominated by a few large firms. The United States itself was the first major developing nation to play catch-up, in this case with British technology. Yet for most of the twentieth century, the United States was able to develop and maintain a balanced, flexible policy that supported the needs of both mature industries and start-ups. Recent developments seem to be undermining that balance.

Old technologies tend to support vested interests, such as Big Steel. Mature technologies, established on a large scale, provide jobs and profits for many people… But new technologies rarely have such established interests.

  1. “The Nature of Technology” by W. Brian Arthur
  2. “Nonzero: The Logic of Human Destiny” by Robert Wright
  3. “The Origin of Wealth” by Eric D. Beinhocker
  4. “Creating the Twentieth Century” by Vaclav Smil
  5. “Transforming the Twentieth Century” by Vaclav Smil
  6. “Unbound: How Eight Technologies Made Us Human” by Richard L. Currier
  7. “Technology: A World History” by Daniel Headrick
  8. “The Box: How the Shipping Container…” by Marc Levinson

If you would like to learn more about how humans have implemented new technologies, read my book From Poverty to Progress: How Humans Invented Progress, and How We Can Keep It Going.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s