Maybe This Is as Good as Innovation Gets1 August, 2018 / Articles
One of the scariest ideas in the world is the possibility that technological progress is petering out. Without it, there can be no long-run economic expansion or ongoing improvement in the human condition. Modern societies and economies are built on the assumption of growth, but even if they were to be reorganized to adapt to a static world, it could be very bad. The return to a zero-sum world could bring back the constant warfare and strife that prevailed in the distant past, if the only way for people to prosper was at other people’s expense.
But what does it mean for innovation to slow down? Technology isn’t just one thing — progress can be rapid in some areas but slow in others. Economists typically measure a society’s overall level of technology by looking at total factor productivity, a measure of how much an economy produces relative to inputs of capital and labor.
Productivity isn’t just technology — it also takes into account institutions, like taxes and infrastructure. But the fact that the slowdown is happening in so many countries, with very different institutions, implies that something deeper than the quirks of local policy is at work.
One possibility is that the rise of free digital services is causing productivity to be undermeasured, since official numbers don’t include the value of things that people get for free. But even if this is true, research by economist Chad Syverson shows that adding back the missing productivity probably wouldn’t change the story much.
The most ominous possibility is that new scientific and technological ideas are simply getting harder to find. This possibility was raised by my Bloomberg View colleague Tyler Cowen in his 2011 book “The Great Stagnation.” Economist Robert Gordon elaborated on the thesis five years later in his magisterial work, “The Rise and Fall of American Growth.”
The scary idea is that easy-to-discover technology is a finite, exhaustible resource. You can derive some of the basic laws of physics by rolling balls down ramps in your house. You can deduce the idea of biological inheritance by playing with pea plants in your garden. Because these things are so easy to research, they get discovered first. But to develop lithium-ion batteries, invasive neurotechnology, or gene therapy requires hundreds or thousands of researchers, strung out in a massive network of academic, government and corporate labs all over the world, using billions of dollars of equipment. Once the universe’s low-hanging fruit has been picked, it takes ever more money and manpower to pick the next bushel.
That daunting prospect has received some support from a new paper by economists Nicholas Bloom, Charles I. Jones, John Van Reenen and Michael Webb, entitled “Are Ideas Getting Harder to Find?” They look at several areas in fields like semiconductors, agriculture and medicine, and find that while steady progress continues, keeping it going has required an ever-increasing army of researchers. That’s true whether productivity is measured economically, or in more concrete terms like crop yields, new drug discoveries, transistor density on computer chips, or the number of academic publications.
Even the fabled Moore’s Law — the tendency of the number of transistors in an integrated circuit to double every 1.5 to 2 years — is requiring a rising number of semiconductor researchers to sustain.
I think Bloom et al.’s paper paints too dismal a picture. In reality, technology doesn’t only grow by improving old products and ideas, but by inventing totally new ones — progress is qualitative as well as quantitative. Innovation is sort of like mining — researchers are always hitting new veins of ideas that they rapidly exploit, but old veins get exhausted. Forty years ago, fields such as gene therapy and machine learning were niche areas, while solar power was stuck in slow mode due to high costs. Now, those are some of the hottest areas in all of science. By focusing more attention on older, mature areas of incremental innovation, Bloom et al. underemphasize the new.
But even so, the economy-wide numbers are discouraging. That the number of researchers keeps increasing while productivity stagnates suggests that the old veins of technological progress are being mined out faster than the rich new veins are being discovered. Spectacular progress in a few high-profile technologies is worth cheering, but it doesn’t necessarily make up for slowdowns in a much larger number of less-glamorous fields.
If this is the case, what can humanity do about it? There probably is room for marginal improvements in research programs. The grant-approval process can be tweaked to favor potentially transformative, risky research over safe bets, or to reduce the time scientists have to allocate to writing grant proposals. Rich countries can admit more skilled immigrants from poor countries, so that they have access to better-funded labs. The patent system can be reformed so that patents don’t block new innovation.
But these moves, although good, will have a limited effect. If the productivity stagnation really is due to most of the easy ideas having been discovered, rich countries have no choice but to stay on the research treadmill — keep hiring more and more researchers, in order to keep progress going.
Meanwhile, there’s always the hope that a transformative new technology — perhaps machine learning, or even true artificial intelligence — will power a dramatic acceleration in research productivity and a flood of cheap new discoveries. It wouldn’t be the first time humanity has been saved by a lucky break — the Scientific Revolution of the 1600s and the Industrial Revolution of the 1800s and 1900s led to big, sustained increases in the rate of innovation. There’s always the chance that something equally amazing and unexpected lies in our future.