Indeed, political scientists who have studied the conditions under which democracy is feasible have emphasized one insight more than any other: The more elites fear that democratic politics is going to appropriate their wealth, the more likely they are to turn against democracy. As the scholars Daron Acemoğlu and James Robinson have argued, for example, elites weigh the costs of accepting democracy against the costs of repressing prodemocracy activists. If democracy allows them to prosper, and opposition to autocratic rule is strong, they favor democracy. If democracy will likely cost them a large share of their wealth, and opposition to autocratic rule is easily quashed, they favor autocracy.

In the context of an automation revolution, that is a terrifying insight. The costs of tolerating democracy will almost certainly rise because the gap between the living standards of the rich and the living standard of average citizens will continue to widen. As that gap increases, the number of citizens who support redistributive policies will rise, and the shrinking ranks of the rich will have more to lose from those policies. Both factors make elites in comparatively unequal societies less tolerant of democratic policies than they are in comparatively equal ones. In fact, there are already first indications that, as inequality has grown in North America and Western Europe, some members of the economic elite have begun to turn against democracy.

At the same time, the cost of repression is likely to fall in an increasingly automated economy. Historically, elites have relied on a large police force and a sizeable army—institutions which are expensive to maintain and difficult to keep loyal. In an automated society, the need for human labor in the provision of security will dwindle just as it does in other sectors of the economy. The rich will live in secure compounds, protected by drones and connected by driverless cars. Meanwhile, ever-smarter surveillance technologies will help to monitor the activities of malcontents. The uncertain calculus of pacifying the army has been solved: no longer requiring the allegiance of flesh-and-blood officers, it can be accomplished with no more than electricity, metal, motor oil and cleanly written code.

To be sure, the elites will always need to rely on at least some physical labor to protect them—even if it is just somebody to monitor all the signals. But this need is likely to be small compared to the ever-growing supply of labor. With their ample resources, elites will easily be able to make the few jobs in elite protection appealing enough that the likelihood of defection will be much lower than it was in authoritarian societies before the rise of the automation age.

To make things worse, it’s not just that the costs of physically repressing dissent are falling; it’s also that the ensuing social unrest will be less and less likely to harm the economic interests of elites. At the moment, the prosperity of the wealthiest members of society is deeply dependent on some degree of social harmony. They need their middle managers to show up to work, their secretaries to organize their calendars and their janitors to make meeting rooms look respectable. The further automation progresses, the less the prospect of social disorder—or even a general strike—matters to the world of the rich. So long as the robots can be relied upon to do their masters’ bidding, the economy can chug right along even in the face of an intensely hostile population.

For now, the material interests of wealthy Americans remain bound up with the maintenance of political order to a considerable degree. Support for democracy remains strong, even among elites (though the level of that support has declined steadily in recent decades). But as inequality gets worse, that shared investment may fade. In an increasingly automated economy, the many are not only poor; from the perspective of the rich, they are also—for the first time in human history—dispensable.

As automation advances, the elites’ costs of suppressing democracy will fall, while the costs of tolerating democracy (with its increasing risk of redistribution) will rise. With each passing day, they will have a stronger incentive to rebel against democracy rather than to accept that much of their wealth might be redirected to the increasingly large ranks of the destitute.

Ultimately, then, society faces one of two scenarios: Either the political system will figure out a way to redistribute wealth in a way that preserves a middle class, or the political system will lose the capacity to redistribute wealth because it becomes completely controlled by elites. It is possible to preserve the middle class through redistribution, preserving both widespread wealth and democratic stability. But the time window to do so is short—and it is already starting to close.

 

IN A 2014 speaking engagement at Google, Hillary Clinton was asked about the possibility that new technologies might vastly increase the degree of structural unemployment. “I don’t have a quick, glib answer to give you,” she said in response. “There are no easy answers.”

It’s not only Clinton who doesn’t have a quick, glib answer. Nobody does. These issues are only starting to enter the public conversation, and few political leaders seem eager to confront them. Perhaps this is because politicians recognize how difficult these questions are. Or perhaps it is because genuine solutions challenge each party’s dominant economic narratives—which would also explain why, even in an election cycle seemingly obsessed with America’s economic decline, technology-driven job loss has not yet become a campaign issue.

The dominant Republican explanation for the recent economic stagnation has been simple: too much government. High taxes rob innovators and entrepreneurs of the incentive to start the businesses that create new jobs. Excessive regulations create unnecessary roadblocks that sap their creativity.

Granted, this story has a grain of truth in it. Some government regulations really are unnecessary, or counterproductive. Many more are confusing, or overly costly to comply with, in part because they require extensive paperwork and expensive lawyers. To make things worse, the influence that corporations and large donors have gained over the legislative system ensures that some of these regulations serve to protect established market actors, not to pursue important social goals.

And yet, the conservative impulse to oppose any redistribution of wealth through taxation leaves the movement unable to grapple with the basic problems posed by the automation revolution. For close to four decades, conservatives have argued that taxes are always too high, that government should always be smaller, that markets can solve all problems. But among those who have looked at the automation revolution, the growing consensus is that government will have to get involved. Tellingly, even the authors of a recent report on automation by the right-leaning Baker Institute admitted that they had not “been able to find any solutions based on the free market. All our proposals involve government intervention.” And while Donald Trump’s unexpected candidacy has, at times, provided a rhetorical challenge to the conservative consensus that it is always better to have less government, his actual policy proposals center on extreme and highly regressive tax cuts. He also talks about bringing back manufacturing jobs through aggressive nationalist trade policy. But the problem is that it’s the robots, not the Chinese workers, who are now running the factories.

The debate about technologies that already disrupt established labor practices is instructive. As Uber has become a political issue, Republican presidential candidates have been quick to embrace the company. (In characteristic rhetorical excess, Ted Cruz styled himself the Uber of politics.) Rather than seeing Uber as a threat, Republicans have generally cast the service as an optimistic story of renewal. Such technologies, they claim, are empowering workers, not undermining their traditional livelihoods. Isn’t it great, they ask, echoing the company’s rhetoric, that we are creating opportunities for our drivers to work as much or as little as they want? And won’t similar transformations soon follow in other parts of the economy, setting office workers who now spend their days in dull, repetitive, carpal-tunnel “information work” free to found small businesses—whether it be knitting ironic scarves, pet-sitting or giving massages on app-powered demand?

What this story misses is that, barring significant changes, there are likely to be few winners in the headlong race towards the gig economy. While some creative souls may indeed thrive in this environment, it is hard to see how this is a road to widespread prosperity. If technology eliminates a large number of jobs, many people will be out of work. If they all rush into service and craft-based entrepreneurial activity, supply will vastly outstrip demand. For all those who aren’t particularly talented or especially lucky, it will be very hard to make a living wage. Banks and financial institutions that loan people money to start their own businesses will do well. So will platform owners, manufacturers of 3D printers and perhaps some consultants. But on the whole, even more people will live without stability or security, relying more and more on chance and luck to generate rapidly dwindling incomes.