Are all the significant inventions already achieved? Economist Robert Gordon identified five Great Inventions, whose discovery in the late nineteenth century powered what he deems an unrepeatable burst of economic growth between 1920 and 1970. These inventions—electrification, the internal combustion engine, chemistry, telecommunications, and indoor plumbing—were indeed far more significant than what often passes for innovation today. While some recent IT breakthroughs are important, no number of Snapchat filters can hold a candle to—well, not needing to use candles to see at night.
The phenomenon that Gordon—a careful, data-driven economist—attempts to explain is real. Economists use the concept of total factor productivity (TFP) to track the degree to which output is not attributable to observable inputs like labor-hours, capital, or education. When TFP increases, it is due to intangible factors such as innovation or better institutions. From 1920 to 1970, TFP grew at about 2 percent yearly. Since then, it has grown at less than half that rate—and in the last 15 years, it has grown at less than 0.3 percent per year, according to the San Francisco Fed’s utilization-adjusted series.
Is this slowdown due to a small number of crucial past innovations running their course? Do no Great Inventions remain to be discovered? Are we now doomed to eternal stagnation? Short answer: no. All it takes to see this is a visit to the technology frontier and a little imagination. But if there is no shortage of technological possibilities, why, then, is economic growth stagnating?
Rapidly developed Moderna and BioNTech/Pfizer Covid-19 vaccines are not only saving countless lives; they have also powerfully demonstrated the utility of mRNA technology. Messenger RNA is a molecule containing instructions that the cell’s ribosome uses to produce proteins. The RNA directs the ribosome to start with one of 20 amino acids and link it to another, and then another, to form a chain that is hundreds or thousands of amino acids long. The ribosome assembles the requisite amino acids in the prescribed order and extrudes the resulting chain, which collapses on itself, in a manner prescribed by the laws of physics, to form a protein. The Covid vaccines deliver mRNA with instructions to build a coronavirus spike protein. Inside the cell, the ribosome dutifully assembles the protein, which our immune system can learn to recognize—and defeat. The vaccines, that is, program a human cell to assemble a protein from a coronavirus, with slight, deliberate modifications.
More generally, mRNA technology lets us program our cells to produce proteins of our choosing. The Covid vaccines represent the first mRNA treatments approved in humans, but the same concept is being studied to prevent HIV infection and malaria, and even to treat cancer. If we addressed these problems with the same urgency as we have the pandemic, AIDS and a number of cancers could soon yield to human control.
Another protein-related breakthrough happened last year. The team at DeepMind shocked the world by announcing that it had essentially solved the protein-folding enigma. Proteins are linear sequences of amino acids; but once created, atomic forces cause them to self-assemble into messy 3D structures that determine their function. In 1972, Christian Anfinsen postulated in his Nobel lecture that it should be possible to determine the 3D structure of a protein from its linear amino-acid sequence. The problem was so computationally complex, however, that it remained beyond our reach until DeepMind attacked it with machine learning. AlphaFold, DeepMind’s protein-folding algorithm, demonstrates the power of machine learning to solve otherwise intractable real-world challenges. We have already seen some of AlphaFold’s methods seep into other groups’ work. As biology continues to leverage computational methods, more and more secrets will be uncovered.
Particularly promising in the long term is the prospect of protein design. Proteins, after all, carry out the most fundamental functions of life. While evolution endowed us with genes that enable us to achieve reproductive success with high probability, it neglected to supply us with genes (which code for proteins) needed to, say, live past 120 years. If we want the functionality of these missing proteins, we will have to engineer them ourselves, something that scientists have already done to a limited degree. Using a next generation of the technology that builds off of AlphaFold, we could, in principle, design proteins to remove arterial or amyloid plaques, curing atherosclerosis and Alzheimer’s disease. We could program our protein nanobots to pare down protein crosslinks in the extracellular matrix to give our tissues back their youthful function and appearance. We could direct them to methylate and demethylate our DNA, optimizing gene expression. From an evolutionary perspective, these life-extending proteins have no use—they don’t increase reproductive fitness; protein engineering could allow us to break free from the limitations of evolutionary neglect.
The changes that such capabilities deliver would not be limited to health. As we begin to limit and reverse aging, medical spending—currently 17.7 percent of GDP—will decline as we reduce illnesses associated with old age, leaving more resources for other pursuits. People will have longer productive lives. At today’s retirement age, you could embark on a whole new career. People also directly value not getting sick and dying—in the U.S., a quality-adjusted life-year is worth $50,000 to $150,000. (Extended life spans would open up new difficulties, of course, in everything from inheritance expectations to retirement planning.)
With the ability to design new, useful proteins, to manufacture them in vivo at scale, and even to stitch their recipes into our genomes using CRISPR, we would, in principle, be able to exercise almost total control over biology. But the breakthrough—amazingly—needn’t end there. Proteins are molecular machinery. They include components that act as proton pumps, rotating motors, conveyor belts, cabling, driveshafts, and more. Nanotechnology visionary Eric Drexler believes that we can use such protein-based tooling to bootstrap nonbiological nanotools. Through perhaps several generations of iteration, nanomachinery could mature to the point of enabling atomically precise manufacturing. This would be such an advance—it would upend the entire economy—that it’s hardly possible to speculate what lies beyond that horizon. It would certainly eviscerate the idea that no Great Inventions remain.
The amount of energy trapped inside Earth is staggering. The temperature at the center of Earth (about 4,000 miles below the surface) is about the same as at the surface of the sun—about 6,000ºC. The Union of Concerned Scientists observes that “the amount of heat within 10,000 meters (about 33,000 feet) of Earth’s surface contains 50,000 times more energy than all the oil and natural gas resources in the world.” This heat is continually replenished by decaying radioactive elements within Earth’s interior at a rate of 44.2 TW, itself about twice humanity’s rate of primary energy consumption. Subsurface heat is a virtually inexhaustible resource that will last for billions of years.
Today, geothermal energy is harvested only near surface features like hot springs and volcanoes, where subsurface heat has made itself evident. But with the improvements in exploration, drilling, and subsurface engineering emerging from the shale boom of the last decade, geothermal energy in the United States can scale to terawatts of electricity production within 20 years.
Next-generation geothermal could work in various ways, spanning a spectrum from evolutionary to revolutionary. At the evolutionary end, modest subsurface engineering techniques could be used simply to extend the reach of conventional geothermal practice. Conventional wells could extract energy from heat resources previously just a bit too deep or improperly situated to be viable. At the revolutionary end sits closed-loop geothermal energy. In a closed-loop system, engineers construct a set of pipes that circulate fluid from the surface to the heat source and back. Heat gets absorbed by the fluid when it is near the heat source and is extracted from the fluid and converted to electricity at the surface.
A killer advantage of advanced, closed-loop geothermal technology is that it can be applied anywhere. Everywhere on the planet, if you dig deep enough, you find heat; a rough maximum-depth requirement is six miles—though, in many locations, the required depth is less. Recent and on-the-horizon improvements in drilling technology make these depths economical, whatever the terrain. The ability to locate geothermal wells anywhere makes extensive transmission of electricity unnecessary. Advanced geothermal wells could be built under major cities, with even the electricity-generating equipment located out of sight, underground.
A second major advantage of geothermal technology is the quality of the electricity that it can produce. Like wind and solar, geothermal produces no carbon-dioxide emissions. Unlike wind and solar, it is available 24 hours a day, regardless of the weather. This feature is critical because electricity grids need to operate in supply-and-demand balance every second of every day. A grid too dependent on wind and solar with inadequate storage will experience instability. Geothermal energy can make the electricity grid rock-solid, while reserving battery storage for electric vehicles, where we need it most.
Perhaps geothermal energy’s biggest plus is that it would unleash energy abundance. As J. Storrs Hall notes in his lament of stagnation, Where Is My Flying Car?, from the early 1800s, American energy use per capita increased by about 2 percent yearly. In the 1970s (ironically, about the time the Department of Energy was created, Hall notes), this trend reversed. We have been doing more with less—15 percent less per-capita energy consumption than the late 1970s peak, to be exact. While energy efficiency is a wonderful thing, we have forgotten the virtues of doing more with more. With clean, dirt-cheap energy, we can stop economizing and start thriving. We can use cheap power economically to pull CO2 from the atmosphere, desalinate water, deploy formerly exotic materials, and travel faster around the globe.
When the history of our species is finally written, the most pivotal moment might not be the development of any of Gordon’s five Great Inventions but instead when we leave our nest to explore the vastness of space. I don’t mean Yuri Gagarin’s first orbital flight in 1961. The cosmonaut slipped the surly bonds of Earth, true, but he returned, and remained dependent on our home planet to survive. At some stage in human history, we will venture out into the cosmos—and stay. We will build habitats, terraform planets, mine space resources, and learn to live off the land, in both senses of that phrase.
The fullest realization of this vision will take decades, but a critical step is happening right now with the development of SpaceX’s Starship. (See “Liftoff in Brownsville.”) We will never truly go to space with today’s launch costs. On SpaceX’s Falcon 9, it costs $2,600/kg to get to low Earth orbit (LEO)—about three times cheaper than on an Atlas V, arguably Falcon 9’s closest competitor, and about 25 times cheaper than the space shuttle. But that’s still far too expensive to launch enough people and material to create a sustainable human civilization in space. Starship, in contrast with every rocket that came before it, aims to enable exactly that.
Everything about Starship is designed to lower the cost of getting large payloads to Mars. The entire system—both the booster and the space vehicle—is reusable, unlike Falcon 9, in which only the first stage can be reused. Starship runs on dirt-cheap liquid methane instead of expensive rocket fuel. It is made from stainless steel instead of more expensive traditional aerospace materials. SpaceX talks about churning out Starships at a rate of one every 72 hours, for a cost of $5 million each. Operating costs drop with a high flight rate, so founder and CEO Elon Musk is figuring a $1.5 million fully burdened launch cost for 150 tons to LEO. That is $10/kg—more than 100 times cheaper than a Falcon 9 launch today.
To get to the moon, Mars, minable asteroids, and beyond, we will need to go past LEO, and that requires more fuel. On today’s rockets, we send cargo past LEO by trading available payload for more fuel. The same Falcon 9 that can send 15 tons of payload to LEO (reusably) can send only four tons to Mars (and that, only when the booster is expended). Starship, by contrast, is being designed to refuel in orbit. Once it reaches LEO, it will dock with specialized tankers pre-positioned in orbit and receive a fuel transfer. Refueled, it can then rocket its full 150-ton payload to Mars, or virtually anywhere else in the solar system.
A 200-fold reduction in the cost of space access will have second-order effects. Satellites and other space payloads are currently overengineered because component failure after launch is a catastrophe. When launching costs millions of dollars, it makes sense to ensure that you don’t have to pay for it twice. When the cost falls to tens of thousands, companies will be more willing to risk a redo. This higher risk tolerance will result in cheaper and more capable space gear—benefiting from the huge performance increase of consumer-grade information technology and from the ability to use less reliable mechanical components rather than solid-state ones. By the end of the decade, Internet access will blanket the planet, there will be live satellite maps of the entire globe, and new large-scale structures will be under construction in orbit—initially to host sensor payloads and in-space manufacturing but eventually human workers, too. If we’re lucky, we will also have a permanent base on the moon and humans setting foot on Mars.
Over the next several decades, then, humanity may conquer biology, develop clean energy too cheap to meter, and start expanding into the cosmos. The real GDP of our species could be orders of magnitude higher than today. At this point, a diehard Gordonian may argue the following: it is all well and good that there are future Great Inventions to be reached. But it will take time to reach them, and it will take time for them to transform society. After all, the first five Great Inventions were developed in the 1800s, but they did not generate significant economic growth until the mid-twentieth century. Until these new Great Inventions mature, it is pointless to wish for faster growth, which is impossible until the appointed time arrives.
While it is true that technology takes time to mature, this school of thought misses the many obstacles that we have raised to new inventions, both Great and lesser. We could have much higher productivity growth right now, if only we had the political will to make it happen.
A perfect example of productivity stagnation due to political dawdling is housing. We know how to lower the cost of housing—build with higher density so that each house uses less land, which is in fixed supply. This can mean putting houses closer together or stacking them as apartments. It doesn’t take a Nobel Prize in economics to understand that building more housing would lower the price of housing. Doing so would allow more people to live in high-productivity areas. Yet local zoning ordinances, rooted in neighborhood opposition to construction, limit builders’ ability to exploit high demand through density. The political will to take the productivity-maximizing path is missing.
Biotech is probably the area with the greatest divergence between the rate of scientific progress and the degree to which that progress is felt by consumers. We are too cautious. Consider the FDA’s handling of emergency test and vaccine approval in the ongoing pandemic. The agency delayed approval of lab-developed tests for Covid at a time when the CDC test was known to be faulty and, in any case, unavailable. And it waited for weeks, while thousands died, to approve vaccines known to be safe and effective. If this is how the agency behaves under national and global scrutiny, what should we infer about business as usual? It appears that the FDA is thoroughly indifferent to deaths caused by type II errors (the non-rejection of a false hypothesis). Drug approval times are too slow. Medical treatments that could save lives languish. Investment capital looks elsewhere for timely returns.
It’s not just the FDA. Our research funding does not aim at the most promising targets. Consider longevity research, which aims to find treatments to slow or reverse biological aging. Last year, a group at Berkeley published two papers showing that blood-plasma dilution rejuvenated mouse tissues and brain function. In 2019, scientists were able to rejuvenate the human thymus, a key organ underpinning the immune system. Another 2020 paper showed that stiffening of extracellular tissues is a significant driver of aging. Despite these and other scientific advances, less than 1 percent of the National Institutes of Health budget goes toward understanding biological aging. As aging is responsible for a huge fraction of our medical spending, this low level of support represents mismanagement of our national research funds. Perhaps the politicians think that life extension is too outré.
Our energy policy is similarly hampered. Nuclear power in the U.S. is six times more expensive than in South Korea. To approve an oil and gas well on federal land takes two weeks; to approve the exact same kind of well for geothermal energy takes two years. These are all the results of policy decisions.
Civil supersonic flight over the United States remains banned, despite the existence of technology to muffle the boom and of startups eager to reboot the supersonic era. Musk’s Boring Company wants to build a tunnel connecting DC and Baltimore in 15 minutes, but the project has sat in environmental review for two years (spoiler alert: no significant environmental impact will be found). NASA has developed an air-traffic control system that would allow delivery drones and other autonomous aircraft to operate in low-altitude airspace, but the Federal Aviation Administration is still years away from implementing it. Amid an unprecedented boom in commercial space, Congress is spending billions of dollars developing a uniquely wasteful, already-obsolete rocket under the old noncommercial model.
Most generally, our society does not seem to care about reversing the stagnation that began in the early 1970s. If the pre-1973 trend in productivity growth had continued, it would have added about 1.25 percentage points to the annual growth rate for the last 48 years. Living standards would be around 80 percent higher today. Shouldn’t there be an outcry? Unfortunately, politics has become dominated by what Tyler Cowen calls the Complacent Class, which would rather preserve neighborhood character than unlock such an increase in living standards.
If we have come to care less about absolute progress, it may in part be because we now spend more of our mental and emotional energy on relative status. Mass media shifted our social frame in the second half of the twentieth century, and the Internet intensified this trend. Prior to mass media, social competition was highly localized and thus accounted for only a limited portion of human motivations. People competed with their neighbors for status, but the scope of the neighborhood culture war was small, and absolute progress remained highly valued. In today’s supercharged national, and even global, culture war, partisan politics has become almost entirely noncognitive: yay teachers, boo police. There is little room left for discussing how we can drive economic growth, for every possible change is first evaluated in terms of who gains and loses in a zero-sum status competition, not on a policy’s likely material effects. To get back to sustained growth, we will have to transcend the need for every policy to be chosen based on the worthiness of various groups.
Stagnation, in sum, is a choice. We can be optimistic about the technological obstacles to economic growth: there are none. We are not, in the sense of the number of technical steps required, that far off from slowing biological aging or mining asteroids. There are short-run gains ready to be had—more drugs coming to market, increases in energy abundance, faster forms of transportation, even tacos delivered by aerial drone. We are not doomed to decades of stagnation. But we need to think differently.
City Journal