Classical approaches can work. That was the message delivered by discussants at our roundtable on the interaction of emerging technologies with the domestic economy. Education, migration, and responsive regulatory policy were all offered as examples of policies that have worked before to help the United States economy take advantage of rapid changes while mitigating their disruptions. It's tempting to frame rapid technological change as an unprecedented challenge for this country, and one requiring unprecedented forms of governance. Similar arguments were, for example, to try to deal with the unexpected inflation of the early 1970s through "new methods" such as draconian economy-wide wage and price controls. Those failed spectacularly and sent the U.S. economy on a decade-long spiral. Our discussants therefore warned against throwing out orthodox policies for untried alternatives, as the result of doing so would be to replace one set of uncertainties—the complexity of the coming change itself—with two.
Productive Work
Contributor Erik Brynjolfsson described machine learning as potentially the most important technology of the generation. Three things drive its rapid ascent: 1) the mass digitization of data throughout the economy, more closely linking the data sets of our computers with the environment of people and our daily activities; 2) significantly better computing power that reduces the time needed to run computationally-expensive machine learning decisions by orders of magnitude (and more closely matching human decision timeframes); and 3) better algorithms that give better or faster results given some set of data. These parallel changes allow machines to effectively share work with humans, with each assuming the tasks it does best.
Brynjolfsson quoted Stanford AI pioneer Andrew Ng in expecting, "Anything today which can be done by a human in less than one second is well-suited to be done better by a machine." Since 2015, for example, machines surpassed (a general population of) humans in image recognition. Since 2017, voice-recognizing machines have approximately equaled human abilities on call switchboards. Applications are proliferating. When eBay in 2014 introduced fully automated, machine learning-based language translations to its Latin American marketplace listing titles, U.S. exports to the region’s buyers increased by 11%, essentially overnight. So machine learning can improve the functioning of existing markets.
While some applications have burst on the scene very quickly, raising governance questions in the process, Brynjolfsson explained why potentially more fundamental changes will actually take some time. And the brake is not so much the technology as the need to redesign jobs, as some tasks in most occupations are suitable for machine learning, while others will continue to require human labor.
To that end, Brynjolfsson's paper aimed to tease apart one of the key governance questions around emerging technologies: the impact on employment. He described machine learning as being different from earlier types of automation, and it is not possible simply to extrapolate earlier experience to understand this new field. He described how machine learning would cut across wage and skill levels, with lower wage jobs disproportionately affected. But most jobs will be reinvented rather than eliminated.
First, Brynjolfsson argued the importance of seeing today's jobs as bundles of activities and tasks. A radiologist, for example, can be said to perform 27 distinct tasks as part of her work, some of which can be done better by machine learning, some of which cannot. A minority of radiologists of the future may spend more time teaching machines to do those particular tasks well. Most radiologists will spend less time on those tasks compared to other important parts of their jobs, such as patient interaction. Redesigning jobs will be key to machine learning productivity gains.
Secondly, employment is not a matter of slicing up a pie. The supply and demand of jobs and their tasks are dynamic. While machine learning and artificial intelligence can substitute for some human activities, it can also augment. Returning to the radiologist example, consider that a doctor using machine learning to augment her diagnoses could potentially diagnose more patients in a day, in fact making her a more valuable (that is, productive) employee than she was previously. Supply and demand elasticities could play a role as well, with hospitals able to offer more radiological services at lower prices, and consumers then deciding to take more CAT scans. Finally, there is the potential for new, utility-enhancing tasks that might emerge through invention and reengineering—using radiological machine learning to perform wholly new types of diagnosis and monitoring for example, or automated remote care that simply wasn't available before, and so on.
Finally, Brynjolfsson argued that machine learning job impacts can be predicted, or at least, understood, by using a skills-based framework. This breaks down the problem into manageable pieces: we can enumerate the skills required in each job, and we can separately evaluate the actual progress of machine learning or other potentially disruptive technologies on each of those skills. This does not dictate a single policy response, but it does create a useful map to guide good governance efforts for policymakers, who are understandably concerned about impacts to their constituents and want to prioritize their efforts where most needed. And it should give the confidence to freely encourage the productivity-enhancing aspects of these technologies, rather than taking a defensive crouch in an attempt to prevent the effects from arriving on their own.
***
It's important to remember that future upside, which does not otherwise get to advocate for itself in today's policymaking.
A country's wealth is directly linked to the growth of its workforce multiplied by the growth in its worker productivity. Through this project we have seen how U.S. demographic trends are relatively healthy compared to other world powers. But troublingly, growth in worker productivity has stagnated since the 2008 recession. Emerging technologies such as machine learning or additive manufacturing are promising antidotes to this, but they must first be applied in the existing economy. To that end, our discussants described how U.S. business reorganizations, which are an indicator of firms becoming more efficient to make use of new technologies or market conditions, are actually happening slower today than they did 20-30 years ago. And machine learning may be no different, with an estimated ten-to-one ratio for in-firm worker reskilling and reorganization costs versus actual investments in machine learning IT. Upfront costs to such investments are high, with productivity gains following later. Such trends are not new: electricity, for example, was an obviously excellent technology, but it took 30 years for factories to see productivity gains from it given the need to strand existing (often steam-powered) assets while developing new skills, methods, and business models to fully take advantage of the benefits of the new technology.
This points to the need to encourage not just entrepreneurship, which is always welcome, but also an efficient labor market more broadly.
How to do this? Updating worker skills will be important, and applied education is a topic to which we will return. But part of an efficient labor market is the matching of skills and capacities to the needs of employers. As architect and urban planner Alain Bertaud has observed, in prison everyone has a job, and a short commute at that, but you could not argue that convicts are living their most productive lives. Flexible markets create the structure for such matching to occur naturally, through an emergent order that makes use of individuals' knowledge about themselves, their capabilities, and their preferences and the workforce needs of industry in ways a government planner never could. Discussants offered a few concrete recommendations to drive improvements in U.S. labor markets given coming disruptions.
One was for a reversal of the explosion in occupational licensing, which has grown to cover 25% of all American workers, up from just 5% in the 1950s. Extensive licensing requirements for trades, many of them in the service sector—everything from yoga instructors to physicians to fruit pickers—began with the justification of protecting public health and safety. Amid the decline of U.S. union membership, however, licensing is now used by trade lobbies of existing workers in a good profession as a means to explicitly block qualified new workers (whom they fear might over time put downward pressure on wages) from entering their field. Licensing makes it harder for someone to start a new line of work without unnecessary time and expense. Until recently, for example, Maryland required 1,500 hours of training to work in a blow dry salon. Arizona required 1,000 hours. Since this licensing is generally state based, it can also restrict worker mobility, effectively segregating the great diversity and expanse of the United States, one of our unique attributes, into 50 disjointed labor markets.
Improving that mobility across jobs, especially across geographies, will be another key challenge. Americans have long been a mobile people, and the idea of setting out for new fortunes in a new town or state is part of our psyche. And Brynjolfsson's work suggests that machine learning will have varying geographic affects across the country. But Americans have become less mobile. Since the middle of last century, the share of Americans who had moved in the previous year to a different location within the same county fell by half, in a steady, secular decline. The share that had moved from one county to another fell by more than one-third. And the percent of young adults—generally the most mobile segment of the population given the need to establish households and establish careers—who have moved at all in the past year also fell by one-third since the 1960s. Too many Americans are stuck in areas or jobs that are not using their full potential, or in no job at all.
Our discussants described some of the theories of why this reduction in mobility may be happening, and potential governance strategies to address it.
Issues included an over-subscription to government welfare programs, such as disability-based social security, which create perverse incentives to stay on the programs instead of seeking new employment. Rolls for disability in the United States grew from 2.7 million in the mid-1980s to nine million in 2014 alongside an ageing population and a widening definition of eligibility, despite workers on average reporting feeling healthier and jobs becoming less physically-demanding. That gives a receiving individual less incentive to find new work given the "effective marginal tax rate" they would incur in doing so—and the potential loss of existing benefits, including monthly payments and Medicaid. Reforms to more strictly define beneficiaries to those who are unable to do work of any kind—as opposed to being unable to perform their existing profession—will be important as the entire U.S. population shifts towards more dynamic employment environments. Private employer benefits can also be a barrier to those who may wish to switch jobs but fear losing health insurance, particular physicians, or other built-up entitlements in doing so, pointing to the need for better benefit portability.
Other impediments to increased mobility and more efficient U.S. labor markets are less obvious. High student loan debt, for example, may be keeping even well-educated young people at home living with parents. Census results show that the share of young adults living with their parents stayed steady from 1990 until the mid-2000s but exploded thereafter, from 11.6% in 2005 to 22.0% in 2017. This is a bid to reduce cost of living, but it may also reduce their potential to gain income from finding the best jobs available to them, with career-long earning impacts. Another factor may be the prohibitive cost of real estate in the nation's most productive urban regions, such as the San Francisco Bay Area, Los Angeles, Seattle, Boston, Washington, D.C, or New York City, which now disproportionately produce jobs in excess of the suburban or rural areas, which once dominated. Many vibrant U.S. urban areas (with the notable exception of Houston) suffer from decades of housing development restrictions and high construction costs that have suppressed the supply of new housing to well-below job growth. This drives up rents and drives down home-ownership (down by one-sixth among young adults since the year 2000). Some have argued that it also drives down marital rates (down to 40% from 55% over the same period) and, eventually, fertility, with long-term demographic implications. One recent study by economists Chang-Tai Hsieh and Enrico Moretti estimated that such restrictions since 1964 have reduced the size of the U.S. economy by one-half. Now, one-half is probably an extreme number, but it points to the foundational importance of a well-functioning labor market. This is also why it is no bad thing that the salient threat of disruptions from artificial intelligence and other emerging technologies now prods policy makers to publicly revisit what may otherwise be considered an esoteric topic.
And of course all of this process of technology implementation will take human creativity. Amazon.com reinvented the bookstore (and other stores) not by substituting machines for humans in shelving and checkout, but by changing everything from the supply chain process to the end consumer, down to the location and operation of warehouses that could offer millions versus thousands of products. Going forward, the process of creative destruction will require rethinking entire businesses around taking advantage of machine learning technology. Brynjolfsson offered a startling prediction to technology bears: even if machine learning technology's progress froze today—no more announcements from Google's research teams or from Stanford's computer science labs—the U.S. economy would still see decades of innovation on business practices that would improve aggregate productivity. Put another way, the much-publicized spread of artificial intelligence is in fact firms applying existing machine learning technology to an ever-widening expanse of industries and problems. This innovation will require lots of intangible investments. Consider that Americans across a number of firms have already spent $100 billion on the development of vehicle self-driving technology, even though no driver has yet been replaced. Machine learning is not a loaded gun. In fact, the bullet has already been shot.
***
A word on social bias in emerging technologies, the subject of substantial roundtable discussion and disagreement. Discussants agreed that machine learning and other algorithmic decision-making systems of the sort now regularly used by both internet companies and other institutions, including the government, are biased, reflecting bias in the data used to train the systems, in the design of algorithms, and in the interpretation of results. Internet companies, for example, collect personal information, create and refine behavioral profiles of individuals, and develop algorithms that curate content, all with the objective of maximizing attention and commercial profits. So it is often in the firms’ economic interest to discriminate. And discussants readily accepted that such decision-making systems can perpetuate or even introduce new bias. Some forms of bias are illegal, such as when ad buyers target housing options in certain locations to one race and in other locations to another race. Legal discrimination might include showing quality investment opportunities to one class of users and shady schemes to another. And bias can also be an issue in many other contexts as well outside of internet advertising or content curation, such as machine learning systems that support decisions on hiring, mortgages, insurance availability and rates, and sentencing guidelines.
The disagreement was on the scale and importance of such bias within an overall system, what might be done to effectively counter it, and the collateral costs of doing so. While some argued that any amount of unintentional or illegal bias is problematic, and saw potential for machine learning decision-making systems without the biases common to human decision-making, others pointed to the known biases of humans in our own decision-making, which suggests comparing machine learning systems to a status quo (flawed) human baseline rather than an idealistic one of perfection.
And whereas some participants suggested (federal) government agency regulation or Congressional lawmaking as ways to provide strong enough enforcement that powerful technology companies would actually comply in removing bias, others warned that involving government agencies, who have less information and technical expertise available to them than private firms, could actually end up making bias problems worse in a fast-changing technological environment. These experts instead offered "the regulation of the marketplace" as a preferred alternative, whereby if one firm were to provide poor or otherwise undesirable services, they could naturally lose users to rival firms that did better. The technological landscape is littered with such corpses. Hewlett Packard, once a byword for dominance in scientific and engineering hardware, as well as high-risk research, rapidly fell amid the explosion of the low cost "PC clone" business in the 1990s. IBM, Kodak, Sun Microsystems, Aol, Yahoo, Blackberry, MySpace: each enjoyed seemingly unassailable product positions, often up until the point that those positions became obviously doomed as they were outflanked in unpredictable ways.
Finally, and more generally, is concern about unintended consequences. Those brought up in Silicon Valley have an innate sense that while the incredibly innovative technologies that were developed and popularized there were not necessarily done without government help, they certainly benefitted from a being largely "left alone" to do business, and repeatedly change how they do business, amid an ever-shifting technological landscape. The result of that has been a mix of good and bad, as with every industry, but the business and consumer products used around the world every day to great utility and at low cost argue that the good has been overwhelming. There is a fear then that a jump to government intervention could break that system, either by actually serving to better entrench the positions of those tech incumbents through regulatory capture, or by weakening the overall ecosystem’s attractiveness for investment. Our discussants furthermore observed how the reach of attractive internet technologies to willing users has proved longer than the reach of any one nation's domestic regulatory arm. The American parents of smartphone wielding middle schoolers may be concerned about the content of their Facebook newsfeeds, or Snapchat advertising, and press regulators to do something about it—but what of the China-based, adolescent-targeting short video app TikTok (抖音), which has grown to 500 million users worldwide, including 100 million U.S. downloads, since its release two and half years ago? This was probably not the sort of trade that those with real concerns on privacy or freedom of expression have in mind. If we are banking on new technologies to enable broad productivity gains in an emerging economy, then we should at every step consider the social costs of limiting that against any expected benefit. Monitoring and careful deliberation is in order.
Technical Education
The choices that American students make about acquiring skills through their educations, and that workers make about learning new skills while on the job or between them, underpins what bundles of "tasks" they will be able to productively perform in a changing economy. 65% of current U.S. job openings require some level of post-secondary skills, and our discussions of state and local community colleges as increasingly important institutions for providing applied education to a diverse spectrum of Americans supported that. Former vice-chancellor of the California Community Colleges Van Ton-Quinlivan reported that five years after completing a 2-year "career technical education" program at a California community college, a worker makes an average salary of $66,000. Five years after completing a 2-year general education associates degree the average salary is just $38,500. At the same time, employers report seeking a broad array of general skills in addition to occupation-specific or technical skills—critical thinking, problem solving, language and effective communication, teamwork.
Education in an emerging new world will not be a matter of funneling students into today's "recession-proof" jobs (which may see novel challenges from emerging technologies), or of focusing on STEM education. Rather, the goal should be to produce graduates who have specific skills that meet the needs employers are looking for today, and a broad enough framework for overall learning that they can successfully return to the education system again and again throughout their careers to quickly acquire new skills as the task bundles change.
Why our interest in two-year community colleges? First, they already exist and go relatively unnoticed in policy dialogues that jump between the deep dysfunction of the American K-12 system, and this country's relatively high performing—but very expensive—four-year university system. California's 115 school community college system, our discussants noted, is likely the largest higher education system in the county. Of the state's two million unfilled jobs, half require a four-year college degree, but half need less than that. Second is that in federal, state, and local governance environments, where budgets are likely to be increasingly crowded out by compulsory spending items such as health care entitlements and pensions, community colleges remain focused and cost-efficient. They are often located close to home with yearly price tags of $3,000-$6,000 versus many multiples of that for a longer (and often residential) four-year option. This is good for government budgets, and it is good for students who can better avoid loan debt traps. Third is their track record at educating a diverse range of students. In California’s two-year institutions, for example, 60% of students are women, 37% are 25 or older, 24% have children or other dependents, 31% are from families in poverty, and 64% work—40% of those full time. Their customer base more closely matches the profile and needs of mid-career students of the future.
Today’s students seek a return on their educational investment—good jobs to support themselves, provide for their families, pay their loans, and save for retirement. And community colleges with career technical educational offerings are attractive to them for their ability to make available occupations that are known to be productive and locally in-demand: nurses, emergency medical technicians, welders, utility line workers, plant operators, or maintenance and repair technicians, for example. 80% of California graduates stay within their own region to find work, and local community colleges are able to partner with nearby employers, individually or perhaps more effectively in regional coalitions, and in doing so keep pace with their evolving needs by mixing and matching task-specific modules with general skills such as English, math, and social reasoning.
Our discussants noted the increasing importance of two-year community colleges remaining nimble as emerging technologies drive acceleration across the economy and society itself. California’s system provides examples of agile programming, such as pooling resources across smaller colleges to build effective collaborations with regional employers such that they see it in their direct interest to interact with students through teaching, internships, and curriculum development, or minimizing the bureaucracy that can slow the roll-out of new curricula.
We also identified the opportunity for community colleges—with their relatively short "business cycles"—to more directly engage with both the employers and high schools, who will be providing their next crop of students, so as to reduce friction in the handoff. Employers can increase graduation rates, for example, by front-loading tuition reimbursement instead of paying employees back after they have incurred the costs. Or community colleges can expose their curricula and major options to high schoolers and their teachers to telegraph future career options. The latter idea was described as being particularly important for minority students who, once in the community college system, tend to select familiar but generally less-productive (and lower-earning) areas of study: in California, the top major for Latinos is early childhood education, and for African Americans it is social work. Ultimately, the country's emerging workforce needs will be met through self-responsibility as students and workers are exposed to incentives to learn, and community colleges will be a key infrastructure in enabling them to execute on those choices.
Of course reforming America's K-12 education itself is of paramount importance. Discussants were proud of the fact that America does have great primary and secondary schools. But they lamented that it also has terrible schools that hold back their graduates' achievements and earnings for life. And their quality is based largely upon their zip codes. For students who are driven, community and four-year colleges find themselves completing the remedial teaching that high schools failed to deliver. Again and again our project has identified the importance of a strong basic education in reaping the benefits of an emerging new world while avoiding the worst of its pitfalls. We will revisit this topic at another session in more detail.
New Arrivals
Since the 2008 recession, there has been much popular and academic discussion of economic inequality in the United States, with the spotlight on "the 1%" and billionaires like Jeff Bezos or Mark Zuckerberg. But focusing on the top misses the point. Jeff Bezos helped to create new growth that benefitted others, too. Instead, if you are really concerned about people's well-being, you should focus on the bottom (How are the least well-off Americans doing over time? Are they improving?), and on the middle (Are wages of those in the middle of the income distribution going up or down or sideways?).
And to that question, history would suggest that rapid economic growth rates are among the best ways to benefit Americans at the middle and lower rungs of society. When unemployment rates fall to their lowest levels and labor markets tighten, social groups with generally higher unemployment rates—women, minorities, the elderly, the less-educated, those with criminal records—are disproportionately pulled into the workforce or are able to upgrade from existing jobs. The converse is also true during a downturn: "last hired, first fired." During the depths of the 2008 recession, black employment rates were falling at 5-6% per year, nearly double the 3-4% rates for whites. But since 2012, black employment rates have recovered at 3-4% per year, versus approximately 1% for whites. Similarly, unemployment rates for workers with a college degree were 10% lower than those for high school dropouts in 2009, but after nearly a decade of continuous economic expansion that spread had fallen to just over 3%. So as a general rule this points to the importance of economic performance as a key enabler of social equality.
In an emerging world, however, our concept of labor markets is expanding. Globalization set off the freeing of capital flows around the world, allowing for huge cross-border investments and speculation. And it ushered in the global movement of goods as complex multi-country supply chains were established to take advantage of beneficial regional attributes and trade soared. Outside of elites, however, globalization did not fully unlock the third leg of the economic stool: labor. Cross-border services can be considered a form of labor mobility, and the value of trade in services is gradually rising versus that of trade in merchandise, particularly in advanced countries like the United States. But what of the movement of people themselves? We know a strong and dynamic economy pulls in marginalized workers in our own country’s labor market—and increasingly we see this now pulling in workers from other countries as well, both within or outside of the legal frameworks in place for that.
Our roundtable discussants considered the history of and continued role for immigration in the United States from this economic perspective.
In 2017, about 258 million people around the world (3.4% of the population) lived outside their country of birth. And countries consider multiple factors as they consider how many immigrants to admit, with what skills, with what status. These include security, cultural and ideological concerns, economic interests, and rights. Migration expert Jim Hollifield offered three such frameworks across the unique U.S. migrant history. The "Massachusetts model," which welcomed immigrants on the basis that they assimilate to the host culture. The "Pennsylvania model," which treated newcomers essentially equally given a baseline respect for local law and basic values. And the "Virginia model," which focused on bringing immigrants for labor.
The United States has applied those models to varying degrees across four major waves of immigration, and the sources of new arrivals has changed as well through those waves . Before 1820, English and Scots dominated—and Africans were forcibly immigrated through slavery. Between 1840 and 1870, economic motivations drove Irish, German, and Scandinavian migrants, including many Catholics. From 1880 to 1914, Chinese migrants went to the western United States for work and to escape upheaval at home, while southern and eastern Europeans went to the East Coast, Midwest, and Southwest. Finally, from the 1970’s to the present, migration has included both low-skill and high-skill workers from Mexico, Central America, and Asia. At each step, immigrants reshaped American society, and they played an increasingly important role in the economy. And each wave drew a reaction from other Americans, some of whom were directly economically impacted by their arrivals, and others who may have only considered themselves to be.
The foreign-born population in the United States today has grown by almost five times since 1970, and it has regained its peak share of nearly 14%, last seen at the turn of the 20th century. Discussants described how immigrants are increasingly important for economic growth, providing both labor and human capital. Today's immigrants, for example, are going to the states where the highest economic growth is. High-skilled immigrants such as engineers and scientists, nurses and doctors are a boon to the economy. And importantly, immigration now provides 30% of U.S. population growth—without immigration the U.S. population would have already stagnated and started to decline, like so many other countries an emerging world.
The most pressing immigration issue today is of course the 12 million people living in breach of U.S. law, half of whom have entered illegally and half of whom have overstayed visas in this country. An inability to enforce immigration laws poses challenges to state sovereignty and security, the legal system, and civil society broadly. At the same time, the country cannot deport millions of people who wish to be here, some of whom have already undertaken deep struggle or sacrifice to try to become Americans. So this creates a paradox. To maintain economic competitiveness, the United States should keep its economy open to trade, foreign investment, and immigration, but it also needs to control its borders so as to not undermine the social contract and rule of law. Clearly there is no black and white answer to this, and it calls for balance.
Our discussants sketched out what an updated immigration strategy might look like, arguing that immigration policy can be both compassionate and "greedy." Currently, for example, 80% of legal immigrants in this country come through the “family channel”—so-called "chain-migration"—where one legal immigrant is able to easily sponsor other family members abroad, some of them distant, to later immigrate to the United States as well. American public opinion is relatively forgiving of this practice, and it has been credited with helping ensure a social safety net for new immigrants who may otherwise struggle in a foreign society. But it does not necessarily do a good job of meeting the host country's goals in terms of desired skills or attributes, and it also is not fair to other desiring immigrants, who may actually have a stronger benefit in coming to this country but no family member to help them.
Canadians deal with this issue through a multifaceted point system to balance this with other interests and approximately 60% of Canadian immigrants are now considered to be "economic" versus 30% family. The United States could do something similar—perhaps focus the family channel on the nuclear family only—while also considering members with certain demographics or capabilities. This would be similar to President George W. Bush's idea of "matching willing workers with willing employers." While experiences then and today demonstrate the political difficulty of getting a sweeping deal through, our discussants were optimistic that with the right leadership, and perhaps by staging reforms into smaller pieces to get the ball rolling, the underlying fundamentals support good prospects for a deal. As one of us has said about Washington, "Sometimes when you know that you are right about something, but everyone tells you it can't be done, you just keep on with it and eventually the fundamentals will prevail."
And the upshot is that this is not new territory. The history of U.S. immigration has never been clean. Yet we remain an immigrant nation in actuality and in self-image. And while those who watch acrimonious arguments over this playing out daily on cable news may find it a surprising claim, we should not have a crisis of confidence in our ability to handle immigration issues. The fact is that this is difficult, but we in the United States—alongside similar immigrant nations like Canada (with a 22% foreign-born population) and Australia (24% foreign-born population, not including the UK)—are actually the best globally at handling this issue because of our unrivalled experience. Globally, migration pressures will grow in an emerging world. Ubiquitous information and communication flows will make information about and interactions with foreign countries and citizens easier. Automation and advanced or additive manufacturing can untangle supply chains and their workers. Changing climates may set people alight from their existing homes or work, fleeing unsuitable areas or seeking new opportunities. Military conflicts through new weaponry may do the same. These flows will need to be managed. More importantly, they will need to be governed as they integrate into the broader society.
***
Our existing American diversity is a boon to this integration as it offers a "golden dome" under which new arrivals can find any number of suitable ways to live their lives as Americans. It also means that our governance institutions and procedures were designed from the beginning to encourage counter-balancing expressions of diversity around a common stake in the American creed.
In fact, the United States at its founding was probably the most conscious historical effort to set up a government over diversity. The historical background is instructive. Emerging from the Revolutionary War, the country faced a variety of internal regional challenges and remained more "states" than "united" under the loose Articles of Confederation. At the same time, the populace was heavily composed of immigrants and lacked a strongly hegemonic social default. James Madison and other founders in their contributions to the Federalist Papers wondered how the destabilizing tendency of various factions "united and actuated by some common impulse of passion, or of interest" would be managed in the hard-won Union.
In his review of the subject, Madison, for example, saw diversity—"a division of the society into different interests and parties"—as "sown in the nature of man." As a part of human nature, it would therefore be impossible to remove its causes, whether through oppression or through consensus. In any case, the young state would not be powerful enough to do so even if it wanted to. Instead, the American answer to managing diversity—as diversity is inevitable—would be more diversity:
The diversity in the faculties of men, from which the rights of property originate, is not less an insuperable obstacle to a uniformity of interests. The protection of these faculties is the first object of government. [Federalist 10]
This argued for creating a framework to allow as much diversity as possible. Not redistribution, not seeking of a common denominator, not compensation, but protection of the abilities of its citizens to express diverse interests. Allowing as many diverse interests as possible within a large and expanding Union would naturally create a political and social system more robust to dominance by any single faction or against the spread of extremism. Were the state to limit any one interest, however, it risked unleashing another. The development of a Constitution (and Bill of Rights) was therefore their way of distributing power so that this diversity could be recognized. Moreover, over time, and not without misstep, the country has learned how to effectively govern over that diversity.
The very structure of a limited federal government based on checks and balances and the ability of state and local governments to have regional authority over matters closer to home, which affect peoples' lives most directly, allowed the nation as a whole to maintain and represent a diversity of opinion, resilient against domination by any one geographic interest or extremist fad. In addition, through today, these "laboratories of democracy" encourage experimentation in governance, the voluntary formation of ad hoc relationships, and help to improve government performance through a form of regional competition. Of course, protections of religious freedoms was also central to the early identity of the country at a time when the Church of England's monopoly on faith demonstrated the impossibility of forced commonality. As Thomas Jefferson reflected in his own letters, "Divided we stand, united we fall."
The country did not always uphold these values. Women's rights and later civil rights for blacks nearly split the nation. Even then, parts of society continued to try to keep the lid on and oppress the interests of millions of U.S. citizens for decades. When the pressure that built up eventually boiled over in the 1960s, it served as a stark lesson of the continual failure to effectively govern over this diversity. Though this has been recognized in the years since and was enshrined in the Constitution, it is worth considering the years of opportunity lost for not only the oppressed but also the nation as a whole had the value of this diversity been earlier enabled.
To return to our topic at hand, a diversity mindset can also be argued to have applied to the development of the U.S. economy. Acknowledging that some will be more economically successful than others, and allowing them to personally benefit from the value they create (while at the same time protecting the least well-off in society through a safety net), has meant that Americans have long had the chance to be rewarded for their own risk-taking entrepreneurship. Unlike some other modern societies, the existence of wealth in the United States—and a shared opportunity to realize it—is generally regarded as beneficial and not something morally corrupt to be appropriated by the state or stamped out through excessive redistribution. This underlying sense of responsibility has preserved a strong incentive for self-betterment across a diverse society. And it will help Americans at all strata of society to find creative ways to take advantage of this century’s emerging new technologies for their personal and community benefit.
The United States of course remains a nation of immigrants, and the foreign-born share of the population since records were kept in the 19th century has stayed in a consistent 11–14% range. As an example close to home here in the San Francisco Bay Area, nearly as many residents of the region were born outside the United States as were born in California, and English is the primary language at home of less than half its residents. It may be surprising to learn that native-born Americans make up less than one-third of Silicon Valley tech workers. But somehow, it all works. Regional labor productivity now exceeds $200,000 per employee and has grown 50% faster than the U.S. average since the turn of the century, and its share of U.S. patents granted has doubled. Other countries with less modern experience in this realm are now figuring out how to get there. The going can get rough, and U.S immigration today has problems, too. But to take the longer viewpoint, this is in our gut. America's ability to incorporate newcomers while maintaining the diversity they brought with them is unrivaled, and that will help us to make the best of—and provide needed global leadership in—what may be an even more chaotic global migration landscape going forward.
