CHAPTER SIX

THE DEATH OF THE GOOD JOB

The Nature of Work in the Autonomous Revolution

ONE OF THE MOST FRIGHTENING EFFECTS of the Autonomous Revolution is that the new technologies that are replacing the human mind are rapidly creating a widespread shortage of traditional work—from service and blue-collar jobs through positions traditionally held by professionals. For now, though, most of the impact is being felt by blue-collar and entry-level white-collar workers. The “good job,” that gateway to and mainstay of the middle class for nearly three quarters of a century, is in steep decline.

The immediate consequence of this has been large numbers of people living lives of despair. Job loss may already be a major contributor to the increasing rates of opioid addiction, fatal overdoses, and suicides—and, for the first time in American history, a decline in life expectancy among white males. In the long run, it could spark a wholesale breakdown of civil society. Democracy is struggling. Hate crimes are on the rise. Our politics are as polarized as they’ve been since the Civil War.¹

We believe that the free market should be given a shot at solving the problem of creating good jobs, but it is highly unlikely, for reasons that we will detail below, that it will be able to deal with this problem on its own. We had better prepare ourselves to consider the alternatives, not excluding direct government subsidies to the growing ranks of the permanently un- and under-employed.

During its heyday, the good job was one of history’s most powerful weapons for reversing income inequality. Powered by the good job, between 1950 and 1980 family income grew faster for the bottom 60 percent of the population than incomes for the top 20 percent. In particular, income for the poorest quintile increased by 144.1 percent, while the top quintile grew at less than 100 percent and the top 5 percent by just 73.3 percent.²

But that situation has been reversed, owing in large part to job capture by autonomous workers.

The autonomous economy employs an endless supply of virtual workers—robots and computers—whose capabilities are increasing at astounding rates and whose costs are declining almost as quickly. When middle-class workers in more and more job categories demand higher pay, virtual workers are happy to step in and take their place for next to nothing.

Our economy is now bipolar, with a traditional economy at one end and a virtual one at the other. The traditional economy is biased toward inflation. The economy at the other end, the Autonomous Economy, is biased toward deflation.

The traditional economy provides us with most of the basic necessities of life—food, shelter, health care, clothing, transportation, and energy. Money spent on these necessities accounts for almost 80 percent of middle-class expenditures.³ And the costs of all of them are rising.

Between 2012 and 2014, the median home price rose by 17.3 percent, while weekly wages rose by only 1.3 percent.⁴ According to the U.S. Department of Agriculture, food prices rose 31 percent between 2005 and 2014, or about 8 percent more than inflation.⁵ The cost of health insurance has risen by more than 54 percent in the past five years.

The middle class is mired in this bipolar trough. Real median household income is below the level it was in 2000—$56,671 versus $57,372.⁶ Since 2000, the number of lower-income families has increased from 31 percent to 34 percent of the population, while the number of middle-income families has declined from 45 percent to 43 percent. The average weekly hours worked by production employees has declined from 38.8 in the 1960s to 33.7 today. Inflation-adjusted annual earnings for production employees peaked in the 1970s and is down by 14.6 percent.⁷ The bottom 50 percent of U.S. taxpayers, approximately 68 million people, had an average adjusted gross income of about $14,800.⁸ Those incomes are supplemented by transfer payments on the order of $13,000 per household.⁹

Nobody knows how many autonomous workers are now on the job; all we have is guesses and estimates. But the estimates of the job losses that are to come are staggering. A recent study by Frey and Osborne looked at 702 occupations and concluded that 47 percent of American jobs might be automated in the future.¹⁰

McKinsey estimates that 85 percent of the simpler business processes can be automated. Many of those processes are in companies that provide services. Using automation, one European bank was able to originate mortgages in fifteen minutes—instead of two to ten days—cutting origination costs by 70 percent.¹¹ A more recent study by McKinsey estimates that 400 to 800 million jobs around the world will be lost to automation by 2030.¹²

In 2011, W. Brian Arthur was probably the first person to describe the Autonomous Economy. He called it the “Second Economy” and pointed to air travel as a quintessential example of it. As he explained it, thirty years ago, when you arrived at an airport for a flight, you would present a paper ticket to a human being at a counter. That person would register you on a computer, notify the flight that you’d arrived, and check in your luggage.

Today, you walk into an airport and look for a machine. You insert your frequent-flier card or credit card, and, in just three or four seconds, the machine spits out a boarding pass, receipt, and luggage tag.

What interested Arthur about the latter scenario “is what happens in those three or four seconds. The moment the card goes in, you are starting a huge conversation conducted entirely among machines—talk that used to take place between people. Once your name is recognized, computers are checking your flight status with the airlines, your past travel history, your name with the TSA, and no-fly lists. They are checking your seat choice, your frequent-flier status, and your access to lounges.”¹³

Now to the sobering part: a sidebar in Arthur’s article speculates that by 2025, the output of this Second Economy will be equivalent to the output of the traditional First Economy in 1995—$7.6 trillion, according to a McKinsey estimate.¹⁴

If the Second Economy achieves that rate of growth, it will be doing the work of approximately 50 million workers.¹⁵ To put that number in perspective, the current total civilian labor force in the United States is about 160 million workers.¹⁶

Of course, the McKinsey numbers are a guess. Nobody knows precisely how many jobs will be superannuated. But to argue about the correct number or the precise timing is to miss the point. Arthur’s estimate points out that the effects of the Autonomous Economy will be very large.

Three factors affect the decline of the good job and work in general.

1. People are no longer required to perform a wide range of functions. This is the structural transformation Arthur described. In the case of the airline jobs he discussed, the form of the business stays the same, but autonomous workers replace employees and their good jobs.

2. The second change, driven primarily by substitutional equivalence, results in new forms of businesses replacing older forms. Netflix’s replacement of Blockbuster is such an example. Instead of distributing DVD discs in physical stores, Netflix streams video over the Internet. Most of Netflix’s business now resides in virtual space, and much of its work is done by computers. Tens of thousands of Blockbuster jobs were destroyed and not replaced, and thousands more spillover jobs (in construction, to build new offices; in retail and food service, to serve new workers; in logistics, to move those physical DVDs around) will never be created.

3. The third change is that Information Age tools make it possible to forecast workloads and schedule workers instantly. As a result, companies now use part-time workers in place of full-time employees. They are cheaper and receive fewer, if any, benefits.

The path to the Level 5 Autonomous Vehicle, or aCar, the robot-controlled vehicle that is being developed by Google and many others, provides a concrete way to speculate about the future.¹⁷ Numerous companies aspire to have these vehicles on the road before 2030.¹⁸

The aCar will make travel safer, cheaper, more enjoyable, and efficient. It will make the nation more capital-efficient and increase its productivity. It will free up as many as 100 million parking places—nearly 800 square miles of space.¹⁹ Unfortunately, it will probably also eliminate a huge number of jobs, cause markets to shrink, and exert downward pressure on wages. These deflationary effects are a result of numerous non-monetizable productivity gains—and they are a microcosm of our virtual economic future. Similar scenarios are being played out in numerous other economic sectors—finance, entertainment, education, media, health care, and so on.

The path to the aCar began in a boring way; all of us have watched it progress over the last quarter century. Cars were gradually equipped with cruise control, engine computers, Bluetooth connections, satellite radios, GPS and Internet mapping services, and emergency services. Consumers love the convenience these technologies provide. Suddenly, we can avoid traffic jams, stay in constant touch with our jobs and families while driving, and never get lost.

More and more technologies are being added: back-up cameras, distance sensors, lane sensors, and so forth. Cars can squeeze into tight parking spaces by themselves and are running into fewer things. The most advanced cars can drive themselves onto freeways, requiring help from their drivers only when they get into difficult situations.

Meanwhile, broadband communications and social networking have made ride and car sharing and other new transportation paradigms more efficient. Suddenly the full-time jobs of 250,000 U.S. taxi and chauffeur drivers are at risk of being taken away by 400,000 mostly part-time drivers for Uber, Lyft, and other services.²⁰

Cab companies are already having a difficult time competing. That’s not surprising: cab fares in Los Angeles are $2.70 per mile, while Uber charges about $1.00.²¹ The oversupply of Uber drivers drives down the price of the service and the value of the work done by drivers. There are other economic impacts as well. Some consumers are discovering that using Uber and occasionally renting a Zipcar or Car2Go is so convenient and cost-efficient that they can get rid of their own cars altogether and just ride and rent.

From the protests of cab drivers, you would think the sky had fallen. Yet in comparison to what is about to come, the economic impacts of all of this have been relatively small. Not for long: the Level 5 aCar will be the game-changer.

Strictly speaking, not much would change—at least initially—if consumers simply purchased self-driving vehicles to replace their existing automobiles, essentially hiring a robot to chauffeur them around. But people who have studied the problem believe another scenario will emerge.

In the autonomous transportation vision of the future, customers will be able to summon a vehicle on demand using their cell phones—a driverless Uber, so to speak. Car ownership will be expensive by comparison.

According to the American Automobile Association study, in 2013 the average annual cost of owning a car driven 15,000 miles per year was $9,122, or 60.8 cents per mile.²² Of this, ownership costs—depreciation, insurance, registration fees, financing fees—come out to approximately $6,000 per year or 40 cents per mile. Operating costs—primarily gas, maintenance, and repair—are roughly $3,000 per year, or 20 cents per mile.

A study has estimated the cost per mile for using a shared vehicle to be just 41 cents, thus saving the average user about 20 cents per mile. These savings result primarily from the shared costs of ownership. The user who drives 15,000 miles per year would save about $3,000 for every vehicle displaced. Even more remarkable, the same study estimates that a special-purpose vehicle designed for short trips of a few miles could deliver a driverless service at the cost of just 15 cents per mile—just a quarter of the current rate.²³

Moreover, an MIT study concluded that a shared fleet using one-third the number of vehicles currently owned by individuals would provide three-minute wait times during non-peak traffic environments and maximum wait times of fifteen minutes during peak periods.²⁴ For many people, fifteen minutes is about equal to the time it takes to get to the parking lot and get in their car.

OUT OF THE DRIVER’S SEAT

What are some of the impacts of such a structural transformation?

Americans own approximately 260 million vehicles.²⁵ If shared vehicles replace 30 percent of them, and each of those 78 million retired vehicles saves consumers the estimated $3,000, then $234 billion less would be spent on transportation. That’s the good news. The bad news is that this would cause about 1.5 million jobs to disappear in automotive related industries—manufacturing, service, insurance, and so forth.

The effects of autonomous vehicles will also be felt in wider swaths of the economy. Level 5 Autonomous Trucks—aTrucks—will move goods faster, more efficiently, and more safely than trucks driven by people. There are 3.5 million professional truck drivers in our country, and about 8.7 million people employed in the trucking business.²⁶ Many of them will be displaced. If just one-quarter of them are, that’s more than 2 million jobs.²⁷

No doubt autonomous vehicles will also drive down the costs of delivery services. The business model for groceries, retail stores, and many commodity products will consist of large automated warehouses that deliver products ordered over the Internet to the customer’s home or to a convenient location for pick-up within a few hours. Some stores will disappear; others will be transformed into showrooms.

Approximately 15.7 million Americans currently work in the retail sector.²⁸ By making delivery more convenient and cost-effective, the autonomous vehicle will spur the growth of virtual retailers, which need only about a third as many employees as brick-and-mortar stores to generate the same volume in sales.²⁹ So, every 10 percent increase in market share by virtual retailers will eliminate on the order of 1 million traditional retail jobs.

Earlier we discussed the structural transformations that are coming in the financial industry. Today, 8 million people are employed in what the government calls “financial activities.”³⁰ As the retail bank changes form and becomes an app on our iPhones, and as robo-advisers take over more and more highly paid financial consultants’ jobs, we may see as many as half of those jobs melt into cyberspace. The total number of estimated job losses due to the aCar and the virtualization of finance comes to 8.5 million.

Lost jobs have a multiplier effect. When manufacturing jobs decline in the auto industry, jobs vanish up and down the supply chain as well. People who lose jobs purchase fewer retail goods and eat out less—so a host of service jobs are lost as well.

Estimates vary dramatically about the size of this multiplier effect. One frequently cited Silicon Valley study estimated that every new high-tech job created 4.3 additional jobs in local goods and services.³¹ Economists refer to such job additions as the spillover effect.³² So when a high-tech company hires one new employee, other companies in supporting businesses hire 4.3 more workers, for a total of 5.3 new jobs. Another study found multipliers of 4.6 in automobiles; 2.5 in communications; 1.9 in construction; 1.6 in wholesale trade, transportation, and business services; 1.1 in hospitality; and 0.88 in retail trade.³³

If we assume that the multiplier works both ways and and pick a fairly conservative multiplier of 2.5, then the total jobs lost would be 2.5 times 8.5 million—or 21.25 million jobs. Thus, the total job loss comes out to about 15 percent of the existing workforce.³⁴

Admittedly these numbers are guesses. The aCar effect and the impact on financial services jobs could be larger or smaller. But in looking at just these two areas, we already have identified more than 40 percent of the 50 million jobs that W. Brian Arthur estimated would be absorbed in the Second Economy.

These new business forms substitute for the old ways of doing things. A truck still arrives but it has no driver. We still get our financial advice, but it comes in an email or over the Internet instead of from someone who meets with us in a well-appointed office. It is hard to conceive of these substitutions having the kinds of job creation effects that occurred when the Model T Ford replaced the horse and the semiconductor replaced the vacuum tube.

ERASING EMPLOYMENT

The aCar effect will be repeated across hundreds and very likely thousands of similar scenarios as more and more jobs move from the physical to the virtual economy. People have already been squeezed out of some job categories. For example, in the past a travel agent would use a computer to contact an airline to reserve a place for a passenger and then send the passenger a paper ticket. In 1996, airline customers gained access to reservation systems through Expedia and other online services—and soon after, travel agent employment began a precipitous decline. Employment in the industry peaked at 172,000 in 1999 and declined by 30 percent over the next five years.³⁵ In 2016, industry employment stood at approximately 65,000, a decline of 62 percent.³⁶ No one expects those jobs to come back.

This is a good example of the rapid rates of change that occur during transitions to the virtual economy. The decline in travel agent employment of approximately 6 percent per year is about three times the rate of decline in farm employment experienced over the course of 140 years.

Today there are 1.2 million professionals engaged in tax preparation.³⁷ As more and more tax data flows in electronic formats, many of those jobs will vanish. The system will become increasingly people-less as electronic systems suck in data and automatically prepare the returns, file them electronically, and directly deduct tax payments from bank accounts—while at the same time, the Internal Revenue Service replaces more and more of its auditors with robots. Tens of thousands of tax preparation (and tax compliance) jobs will almost certainly vanish into cyberspace.

Utilities are automating as well. Today, there are 15,600 meter readers. As Internet-enabled meters replace existing stand-alone versions, remotely accessed people-less utility billing systems will emerge.³⁸ So most of those jobs will also disappear.

The relentless reality is that machines are only going to get smarter.

The legal profession is now a target for attack. Venture capitalists are pouring money into legal tech. Ninety-six million dollars was recently invested in Zapproved. JPMorgan uses COIN to review documents that in the past required legal aides to expend 360,000 hours of work per year. CaseMine, an Indian company, is building systems to aid in document discovery.

A large number of things that government bureaucrats do are done by the book. If a machine can beat a Go champion, it can easily learn how to play many bureaucratic games.

According to the most recent data, labor productivity is growing at a 1.3 percent annual rate, which means that about 2 million workers are displaced every year by productivity improvements.³⁹ In a traditional economy, where every $150,000 in growth creates a new job, the GDP would have to grow by about 1.7 percent per year to absorb those displaced workers. The good news is that our economy is currently growing at more than 3 percent, so jobs are being created and unemployment has dropped below 4 percent.

Now imagine that our entire economy looked like Amazon, Google, Facebook, and Netflix, companies that are highly productive thanks to virtualization and intelligent machines. Those companies require $600,000 to $1 million increases in income to create a new job, not $150,000. GDP would have to grow at a rate of between 6.8 percent and 11.3 percent to absorb all those displaced workers. No modern developed economy has ever maintained that level of growth. While those companies may create dozens of billionaires and hundreds or even thousands of millionaires, they cannot save the middle class.

Consider what we might call the “Amazon Effect.” The Internet has made shopping more efficient and created more competition—and that has driven consumer prices down. But this transformation has had little or no effect on per capita sales. Monthly retail sales adjusted for both inflation and population growth are below where they were prior to the 2008 recession—$165 billion versus $168 billion—and have increased by just 0.6 percent per year in the last fifteen years, or about 10 percent.⁴⁰ Meanwhile, employment in the retail and wholesale trades dropped from about 21.2 million jobs in 2000 to 19.9 million in 2010.⁴¹ New technology is not creating new jobs in retail.

What is particularly disconcerting about this jobs scenario is what is missing compared to the past. Two hundred years ago, when jobs were vanishing in agriculture, they were on the rise in manufacturing. Then, as the latter area matured, new jobs were created in the service industries.

Eighty percent of the workforce, 104 million all told, now work in services. But as more and more of those jobs are automated, we need a new area of economic growth to absorb those excess workers. Unfortunately, that area appears to be the burgeoning workerless segment.⁴²

Many of the proposals for bringing back the good job involve investing in infrastructure and creating more manufacturing jobs. But here is the challenge: there are only 6.9 million jobs in construction and 12.5 million jobs in manufacturing, a total of about 19.4 million. Because the ratio of manufacturing and construction jobs to service sector jobs is one to five, every 1 percent productivity increase in the service sector would require a 5 percent increase in manufacturing and construction employment to offset its effects.

As pointed out earlier, thanks to non-monetizable productivity, we suspect that productivity increases in the services area have been understated. If productivity growth in services, when measured in terms of output rather than dollars, is more like 3 percent per year, then manufacturing and construction jobs would have to increase at a 15 percent rate to offset its effects, which is significantly less likely to happen than a 5 percent growth rate.

Some might say that our pessimism strikes a discordant note at a time when unemployment numbers are as low as 4 percent and companies are struggling to find skilled workers. Perhaps it does; we would like to believe that we are giving too much weight to our anecdotal observations. But those low unemployment rates have come after unprecedentedly high levels of deficit spending since the 2008 financial crisis. Currently deficit spending is running at a rate of 5 percent. During the fifty years before the economic crisis of 2008, that number averaged around 3 percent and only exceeded 5 percent one time, in 1983.⁴³ We believe that the current level of deficit spending is unsustainable and that it is masking the job displacement effects of the Autonomous Revolution.

BUILD FOR THE TWENTY-FIRST CENTURY

What then should we do? Unfortunately, there is no silver bullet. The problem demands a number of different actions.

Certainly reducing burdensome regulations, creating a more rational tax structure, better educating our citizens, and increasing investment in science and technology are crucial. But even if we could muster the political will to do those things, all of them taken together would still not be sufficient.

Fortunately, we will have a lot of construction to do. We argued above that it will be difficult to create enough new manufacturing and construction jobs to completely offset the effects of the Autonomous Revolution, but aggressively building infrastructure will certainly help. We believe that we have greatly underestimated the amount of infrastructure that will have to be built to support society in the future.

The reason for our underestimation is that every time humans have made significant advances in the ways we interact, connect, and transfer information to one another, we have transformed the physical infrastructure of society. The transition from tribal cultures to agricultural communities depended on the development of better modes of transportation—wagons drawn by horses and oxen and ships that could transport goods. Cities would never have been viable without the means to get food and energy (firewood at first, and later coal and other petroleum products) in from the countryside and the goods that they manufactured out to distant markets.

The railroad created densely populated industrial cities and the hub-and-spoke pattern of suburban infrastructure, in which homes were built along railroad lines. The automobile created a more sprawling, two-dimensional suburban infrastructure, in which homes, businesses, and shopping centers could be located anywhere a car could reach. A lot of the economic and employment growth of the post–World War II period was created by building those highways, shopping centers, and detached houses.

Just as in the past, our new ways of connecting will necessitate a new kind of physical infrastructure. We will need to do much more than repair and update our bridges, roads, pipelines, and electrical networks.

To understand why, consider that the physical infrastructure of today’s society evolved in response to basic information transfer problems. In order to efficiently exchange the information necessary to buy and sell goods, produce things of value, learn, or be entertained, people had to gather in physical places.

Our existing infrastructure assets, and the business processes supporting them, can be seen as information transfer proxies. As noted earlier, consumers go to retail stores in part to find out what is available at what prices—in other words, to get information. Workers go to office buildings to gain access to files and communicate with co-workers. Walmart stores and office buildings are essentially giant file cabinets.

But modern information and communications technologies increasingly remove the need for such proxies. To name some obvious examples, eBay, Google, Wikipedia, Amazon, and Orbitz have each in their own way reversed the traditional requirement that the customer or user must travel—to a garage sale, library, bookstore, or travel agency—to obtain a product or service. Information transfer processes are already efficient enough to allow stores to come to shoppers, files to laptops, and work to workers. And we are only at the beginning of that evolution.

At one time, Internet visionaries believed that people would choose to work remotely and live anywhere they chose. But a different scenario now seems just as plausible, which is that more and more people will choose to live in more densely populated areas that are served by intelligent public transportation. This could create an apartment construction boom. The cities of the future may come to look and feel more like the old cities that Jane Jacobs celebrated and mourned in her book The Death and Life of Great American Cities half a century ago—lively, diverse federations of mixed-use neighborhoods, but powered by twenty-first-century infrastructure.

The amount of construction required to create this autonomous country of the future is massive. The American Society of Civil Engineers (ASCE) has estimated that the country has a $3.6 trillion backlog and it is planning to spend only $2 trillion by 2020.⁴⁴ Completing the projects identified by the society in a five-year period would require spending an additional $300 billion a year, or about 2 percent of GDP.

The projects identified by the ASCE represent only a fraction of what has to be done to create a robust twenty-first-century infrastructure. Providing 100 million households with FTTH (fiber to the home) would require an investment of between a quarter and a half-trillion dollars—and that’s just the tip of the iceberg.⁴⁵ Massive investments in public transportation are needed—things like high-speed rail, and possibly even hyperloops that would allow trains to travel at the speed of sound.⁴⁶ Substantial portions of our cities and suburbs will need to be retrofitted or rebuilt from scratch.

In the 1960s and 1970s, when the 41,000-mile Interstate Highway System was being completed and the suburban population was burgeoning (it almost doubled between 1950 and 1970), construction spending ran 10 to 11 percent of GDP.⁴⁷ It falls in the 5 to 6 percent range today,⁴⁸ so it could easily double.

Creating the construction boom of the future would undoubtedly involve government subsidies and programs similar to Eisenhower’s Federal-Aid Highway Act of 1956.⁴⁹ It would also involve tax incentives to encourage entrepreneurs to build the right kinds of infrastructure.

These arguments about construction obviously go against the notion that white-collar work bolstered by a college education will be the best defense against the depredations of job loss due to robots and artificial intelligence. In fact, building and trades, traditional blue-collar work, could well turn out to be the Last Stand of the Good Job.

Another opportunity lies in entrepreneurship. New companies are the major source of new jobs in the U.S. economy. Unfortunately, new business formation has dropped to a forty-year low that is about 25 percent behind its best year.⁵⁰ Our lawmakers hardly seem to have noticed that fact—after all, small start-ups are too busy to do much lobbying on their own account, even if they could afford to. Because of that, legislators typically ignore entrepreneurs or freight them with company-killing regulations. But an economy that actively cultivated new company creation might be able to generate hundreds of thousands of those start-ups per year—and generate 1 or 2 million new jobs.⁵¹

BEYOND A FREE MARKET

As far as we can tell, the ideas above exhaust the free-market solutions that will have a major impact on the problem. Finding ways to better compensate large numbers of content providers will also help, but they are unlikely to create millions of high-paying jobs. If the goal is to create more meaningful work and raise the income levels of the less fortunate 60 percent, it will have to involve government programs, direct subsidies, and some form of income redistribution.

One of the most creative and well-thought-out proposals for addressing this issue that we have seen comes from the arch-conservative writer Charles Murray, who has proposed basically terminating all welfare programs and replacing them with a basic income of $13,000 per year for all citizens over the age of twenty-one. Everyone would be required to spend $3,000 on health insurance. Payments would start to taper off after someone was earning more than $30,000 per year.⁵² Murray argues that the savings resulting from the elimination of entitlements would be enough to finance much of the plan.

One of the most common objections to universal basic income (UBI) proposals is that they undermine the self-esteem and personal identity that comes with having a job. Opponents of UBI paint pictures of twenty-one-year-old white males living in cheap apartments, hooked on opioids, playing video games ten hours a day, and contemplating suicide. But as Murray points out, the plan would not incentivize unemployment, because everyone would still have to work to live at a comfortable level. And since everyone would receive payments, there wouldn’t be a stigma attached to them.

Another approach to consider is a version of the French solution—cutting the workweek. Let’s talk a little history and Information Age reality. The workweek has been declining for many decades. In the early 1800s, it was seventy hours. It fell to fifty hours in the early 1900s and then to forty hours.⁵³ For a large number of Americans, the less than thirty-five-hour workweek is already a reality. Part-time employees now make up 18.5 percent of the workforce, up from about 16.6 percent in the year 2000.⁵⁴

Of course, creating jobs with shorter workweeks and less pay would not put more wealth in the hands of the less fortunate 60 percent. Some kind of income subsidy would still be needed. Possibly the government could consider subsidizing the first twenty hours of work that employees perform with direct cash payments. Those subsidies could be reduced if a worker worked more than, say, thirty hours per week.

Perhaps the greatest opportunity for creating meaningful work will be to find ways to pay people who do socially useful work that today has no monetary value. Consider people who pay others for childcare during working hours, or families who pay thousands of dollars per month to a service provider for an infirm adult who lives in an adult care facility. If those children or elderly persons are cared for by family members at home, that same work is done for free.

Is there a workable alternative that monetizes the last scenario? It would be difficult to figure out how such a system might work—much less administer it. But it is easy to argue that doing socially useful work benefits society and, by freeing talented workers, also benefits the economy. Here is a perfect opportunity for creative minds to come up with a workable solution.

There is so much that is unknown. And there is always the possibility that the free-market optimists are right. So let’s by all means keep pushing the free-market solutions and pursue an aggressive construction agenda. If we do the right things to prepare for the Autonomous Revolution, construction could easily hit 10 percent of GDP. But let’s also start experimenting with some of the subsidy approaches as well. The experience we gain will help us avoid some mistakes if we have to roll them out in a bigger way. Let’s also search for ways to pay people for performing socially useful work.

The one thing that is certain is that phase change will alter the rules. The sooner we confront the problem—and learn the new rules—the sooner we will uncover workable solutions.