Free exchange | Innovation

Sharing the knowledge burden

Discovery and innovation in the modern world

By R.A. | WASHINGTON

ALBERT EINSTEIN was a singular genius. The Albert Einstein model of discovery, however—the solitary mind producing remarkable insight—was not particularly unusual for his time. For much of the period from the beginning of the industrial revolution, scientific and technical advances—including the occasional stroke of brilliance—were within the reach of the diligent amateur and the garage tinkerer. That is no longer the case. As the stock of human knowledge increases, the time needed to move oneself to the knowledge frontier grows. In a 2009 paper, Benjamin Jones noted:

If knowledge accumulates as technology advances, then successive generations of innovators may face an increasing educational burden. Innovators can compensate through lengthening educational phases and narrowing expertise, but these responses come at the cost of reducing individual innovative capacities, with implications for the organization of innovative activity-a greater reliance on teamwork-and negative implications for growth. Building on this "burden of knowledge" mechanism, this paper first presents six facts about innovator behaviour. I show that age at first invention, specialization, and teamwork increase over time in a large micro-data set of inventors. Furthermore, in cross-section, specialization and teamwork appear greater in deeper areas of knowledge, while, surprisingly, age at first invention shows little variation across fields. A model then demonstrates how these facts can emerge in tandem. The theory further develops explicit implications for economic growth, providing an explanation for why productivity growth rates did not accelerate through the 20th century despite an enormous expansion in collective research effort.

The growing "burden of knowledge" needed to move science forward has increased the age at which researchers begin making contributions, while also increasing instances of collaboration on research. This helps explain why a huge increase in resources dedicated to research over the 20th century didn't lead to an acceleration in productivity growth. The productivity of research itself has been falling. Or to use Tyler Cowen's phrase, the low-hanging fruit has been picked.

Mr Jones implies that this may have negative implications for long-run economic growth. It could, but I'm not certain that it must. I'll give you a couple of reasons why.

The first is that it has become dramatically easier in just the past two decades to leverage research resources thanks to stunning improvements in computing power and communication technology. Powerful, easy-to-use statistical software is only about 20 years old. Widespread broadband access is far younger. The potential of the massive processing power of a system like Watson is only just being explored. For most of the industrial revolution, technological change was labour augmenting, enabling men and women to produce far more textile or agricultural output than had previously been the case. The past two decades have given the world enormous gains in human-capital-augmenting technology. It's now remarkably easy to share data around the world, mine massive data sets for interesting relationships, test those relationships with powerful statistical software, and publish and share results with audiences the world over, all in a matter of hours. As we improve these technologies and become better at using them, research productivity may enjoy surprisingly rapid growth.

Second, the relationship between the discovery of new knowledge and economic growth is not straightforward. Scientific knowledge accumulated in Europe and Asia for centuries before finally touching off sustained economic growth. Seemingly small leaps in knowledge may also contribute to decades of growth. It takes a long time for institutions and economies to rearrange themselves to take best advantage of new technologies. The industrial revolution once again supplies a blueprint for these processes. Scientific and technical advances were clearly critical to the revolution, yet growth itself was largely driven by the slow transformation of society into a geography capable of applying old technologies at massive scale. Populations that had been mostly rural for millenia suddenly concentrated in enormous urban centres supported by huge infrastructure investment. On a smaller (but still massive) scale, consider the growth benefits of reorganisations built around assembly lines, highways, and container shipping.

Information and communication technology—as it exists right now, not even assuming future progress—offer the potential for similar bursts of growth. Growth will take similarly transformative reorganisations within society and across geographies. That doesn't happen overnight. It also doesn't happen on its own. Governments need to be willing to accommodate change, and conditions need to be in place to allow private individuals and businesses to experiment with new and innovative deployments of existing technologies.

That's another reason the burden of knowledge issue is less of a concern to me than it may be to others. Advancement of the scientific frontier is growing more difficult. Yet deployment of existing technologies to more productive ends may well be growing easier. Consider the tacocopter. The tacocopter is a not-quite-real-not-quite-a-joke business idea that became a brief internet sensation back in March. The concept is stunningly simple: order tacos on your iPhone and a quadracopter drone will deliver them to your doorstep. As you can read here, the plan would face technical and (especially) regulatory hurdles if implemented today. Yet the potential, for this or similar experiments, is obvious. Cheap, agile drone technology is available now. Building apps is trivially easy. Mapping and location technology and data are getting better all the time. If not drone copters, perhaps 3D printers or autonomous vehicles. It's a short leap from the ridiculous to the transformative. And the ideas needed to transfer these technologies to everyday life are increasingly the domain of entrepreneurs rather than academics. One doesn't need 20 years of study to spot profit opportunities.

The takeaway should be that it doesn't appear to be necessary to generate a steady or accelerating flow of revolutionary scientific and technical innovation to maintain sustainable growth. It seems to take long enough to wring growth out of existing technologies that a society committed to research can accumulate insights in bursts sufficient to power growth during droughts. Put more simply, a lack of big discoveries doesn't strike me as the binding constraint on growth. Conceivably it is the pace of entrepreneurial adaption that matters. I'm most inclined to think that its the pace of societal evolution that is most binding: growth proceeds at the fastest pace that legal and social institutions can tolerate.

Think of the challenges that would face the would-be tacocopter entrepreneurs. Consider that issues surrounding liability and law, rather than technology, now appear to be the biggest obstacle to autonomous vehicles. Look at the legal struggles faced by innovative services like Uber and Airbnb. Disruptive innovations are bumping against a broad array of regulatory hurdles that built up during a very different era of economic growth.

Or take an area of great personal interest to me: land-use policy. The information technology revolution seems to have increased the importance of and returns to clustering in high-skill sectors. I recently attended a discussion on the economic study of innovation, at which one of the participants pointed to this fascinating paper:

Are scientific knowledge flows embodied in individuals, or "in the air"? To answer this question, we measure the effect of labor mobility in a sample of 9,483 elite academic life scientists on the citation trajectories associated with individual articles (resp. patents) published (resp. granted) before the scientist moved to a new institution. We find that article-to-article citations from the scientific community at the superstar's origin location are barely affected by their departure. In contrast, article-to-patent citations, and especially patent-to-patent citations, decline at the origin location following a star's departure, suggesting that spillovers from academia to industry are not completely disembodied. We also find that article-to-article citations at the superstar's destination location markedly increase after they move. Our results suggest that, to be realized, knowledge flows to industry may require more face-to-face interaction than those to academics.

Emphasis mine. Proximity matters in turning ideas into applications. It also matters in turning applications into employment. The Bay Area is booming precisely because of the importance of knowledge clusters. But the Bay Area is the least productive large metropolitan area in the country in terms of adding new housing, thanks largely to very restrictive rules on new construction. And so you end up with stories like this, on concerns that high real estate costs will check the Bay Area's boom and prevent it from benefitting low- and middle-income households.

The burden of knowledge is an important dynamic, but it's not the aspect of innovation that concerns me most. My greatest concern is that Americans may have lost their ability to tolerate the messiness of disruptive growth. The problem is not a shortage of ideas. It's that ideas gather dust on shelves, the better to preserve the views of rich, satisfied homeowners.

Discover more

Religious competition was to blame for Europe’s witch hunts

Many children are still persecuted as alleged witches in Africa for similar reasons

Has BRICS lived up to expectations?

The bloc of big emerging economies is surprisingly good at keeping its promises


How to interpret a market plunge

Whether a sudden sharp decline in asset prices amounts to a meaningless blip or something more depends on mass psychology