For three centuries, the broad direction of political history has been characterised — maybe even driven — by technological change and liberalism reinforcing each other. At least in the rear-view mirror, it has seemed a remarkably dynamic equilibrium. Technical progress increased productivity without leading to concentrated authoritarian power. The expansion of civil and political rights did not arrest efficiency and prosperity. On the contrary, the stability and openness provided by liberalism increased incentives to invest; and many technological developments broadened prosperity, reducing the capacity of exclusive elites to dominate their societies. So much has this miracle been taken for granted that the rule of law, applying to the rulers themselves, is regularly advanced as a precondition for enduring economic success. David Hume said much the same for republican government (under which heading he included constitutional monarchy). As Adam Smith observed of his mentor:

Commerce and manufactures gradually introduced order and good government, and with them, the liberty and security of individuals, among the inhabitants of the country, who had before lived almost in a continual state of war with their neighbours, and of servile dependency upon their superiors. This, though it has been the least observed, is by far the most important of all their effects. Mr. Hume is the only writer who, so far as I know, has hitherto taken notice of it.

Approaching the matter more systematically, Smith himself offered a four-stage conceptual history of our political-economic development: from hunter-gatherers through pastoralism and agriculture to commercial society, in which each of us became in some sense free to sell our labour and to buy goods and services from others in markets. Smith’s is not a historian’s history, nor even a conjectural narrative. Rather, it offers general equilibrium accounts of four distinct ways in which groups of people have lived together under different technologies. In each, the economics, politics and ethics (norms) are mutually reinforcing. The way of life must be not only incentive compatible but also, I want to stress, values compatible; or, pulling incentives and values together, a society’s core values must have been internalised to the point of becoming part of most relevant actors’ motivational psychology.

By way of illustration, one way of thinking about the unravelling of the old Soviet system is that the incentives set up by its institutions were at odds with the values those same institutions supposedly animated. Looking back, the Soviets lived in dynamic disequilibrium; a rather poor result given Marx’s teleology. Secrecy — the habitual enemy of efficiency — had to increase to mask the underlying tensions. On this story, the most grotesque rulers — Stalin and Mao, to name just two — are products of system contradictions that only a credible tyrant can keep from blowing things apart.

Smith perhaps says less than we can about the transition between equilibrium political-economic systems. The obvious exogenous shocks come from outside a society in the form of war, ideas, and technology. The obvious endogenous or internal shocks come from surges of resistance, invention, educational reforms, and perverse institutional dynamics. Among those, looking briefly at resistance helps makes sense of why, so far, technical progress and liberalism have made such a good team.

The simplest point is that, as a matter of empirical fact, technological change and, more specifically, the technical progress that increases aggregate productivity has tended, on the whole, to broaden the capacity for meaningful resistance, meaning resistance — active or passive — that reduces the (risk-adjusted) returns to ruling groups from ruling. The most obvious examples concern offensive weapons: in our world, notably firearms. But the same thought is found in Aristotle’s observation in the Politics that the advent of hoplites (with their vast shields) broadened who counted in the polity (crudely, infantry as well as cavalry).

Slightly more subtle examples include innovations that enabled the emergence of a merchant class, who became able to press for rights that both underpinned their interests (their motivation) and promoted prosperity-generating innovation. Later, when commercial society morphed into industrial society via the innovation known as the factory system, the notion that each person could freely contract with employers became fanciful but, gathered and acting together, workers could press for group rights. When granted, they provided the wherewithal to press for wider political rights; namely, the vote, enabling millions of people to overcome coordination problems in opposing (or supporting) their rulers.

The purpose of those reflections is to underline how significant it would be if we were headed into a world where technological change might serve power without liberty. There are reasons for thinking that, in an extraordinary reversal, authoritarian government and technological change might indeed already have become mutually reinforcing. But before turning to that alarming prospect, I shall sketch three examples of recent technological changes that have affected the distribution of power in democratic politics.

The first is that the new technologies lend themselves to networked services, in which very broadly the winner takes all. Google and Facebook are poster examples. Although now much more prevalent, the phenomenon is hardly new. The old telephone monopolies were obvious examples half a century ago, as the railways had been before that, and by and large they operated in Europe as publicly owned monopolies. When from the 1980s those and other ‘natural monopolies’ were ejected into the private sphere, it was on terms that included economic regulation designed either to reduce barriers to entry or to curtail the abuse of economic dominance. The focus, in other words, was economic efficiency: how to harness to the public good the private incentive to get rich.

But the justification of those anti-trust regimes might just as validly have been cast in terms of impeding the accumulation of private political power, as Germany’s post-second world war ordo-liberals understood after the travesty of big businesses allying with the Nazi party during the 1930s. The question for us, now, is whether the increased importance of network-based businesses will see concentrated economic power morph into concentrated private political power, with legislators becoming modern day equivalents of the British parliamentary clients of the Duke of Newcastle, who in late-18^th and early-19^th century Britain controlled who was elected in multiple parliamentary constituencies, making him a great power in the land.

When one looks at lobbying and campaign finance in the United States, the conjecture is not entirely idle. The Chicago School anti-trust policy, making prospective consumer welfare the measure of all things, cuts across what some had thought to be the liberal and republican political pre-conditions for a liberal economy: dispersed power. The solution might lie in constraining or balancing the potency of concentrated riches in politics. But if it is left to curtailing economic power itself, we should expect enemies of liberal politics to oppose anti-trust reform, arguing their case in the interests of an uninhibited liberal economics. Those struggles are underway.

The second example is in a somewhat similar vein, except now not about entrepreneurial tycoons but super-rich managers. The labour market for chief executives, and for other so-called C-suite executives — relatively safe and routine jobs — is clearing at prices that could make sense only for stars. This during decades when aggregate productivity growth has been tepid, and when much of the rise in the stock market has been down to a handful of tech companies. While not the fault of the individuals concerned — except to the extent they think they deserve their riches — it seems that superstar rewards are being handed out for what, for the economy as a whole, is mediocre performance. Politically, those whose incomes have not kept up might come to think of the system as rigged. Donald Trump makes that criticism, but might have been looking in the wrong places. The technology of modern labour markets might not be rich enough to deliver a set of marginal prices that reflect the marginal product of bosses across different sectors and firms. Middling performances are unlikely to be fully reflected in low bonuses if incumbents can easily move elsewhere, or simply afford to retire with their riches. That leads to a collective action problem — an individual board will not risk rocking the system’s de facto pay norms given the large transaction costs and uncertainties in finding and embedding replacements.

With the returns to top management jobs in the private sector having become so massive, we should perhaps not be surprised when people lament the quality of modern government. Pay differentials have become so great that the economy of esteem that used to draw people to public service might have inverted, acting now as a deterrent, and so even creating an adverse selection problem. To the extent liberal economy relies on some governmental services — such as order, security, and the adjudication of disputes, to take only a minimal conception of the state — this might eventually prove perverse for the economic winners themselves. Liberal democracy cannot sustain itself without multiple sources of prestige, so we should worry if relative wealth is the measure of all things.

Third, whereas the possible political hazards of network-based economic power come from the top, the new technology also brings bottom-up risks to liberal democracy. It is a familiar point that the rise of full-franchise representative democracy during the 20^th century owed not a little to the emergence and growth of a white-collar clerical class. These were people who, in their many millions, had in just a few generations moved from tied agricultural roles through heavy labour to sitting in offices. They were the backbone of civil society, and contributed to civic virtue by their high propensity to vote. As has been obvious for some time, those jobs are in decline, replaced by local automation or by relocation to lower cost, poorer countries. If the lower-middle and middle-middle classes feel threatened or alienated, what are the prospects for democracy? Since electing political office holders functions partly as a check on administrative power, that possibility should worry those liberals who prefer the ‘liberal’ aspect of liberal democracy.

The point of those examples is to paint a crude picture of how recent technological changes have contributed to souring our politics without yet delivering the marvellous productivity improvements associated with earlier economic revolutions. So far, the innovations have gone, most notably, into consumer products rather the production process itself. That might only be a matter of time, of course. But if we need time for the true goodies to come through, releasing resources to help address our many problems, we will have to hope the new technology does not strike a more direct blow to our way of life, including via tipping the balance of power in the world.

Crudely put, the spectre is the surveillance society, where Bentham’s nightmarish pantechnicon — total visibility in four dimensions — becomes a reality. Research conducted by Harvard’s David Yang and co-authors on the People’s Republic of China has shown that not only does increased state surveillance reduce protest, but also that the prevalence of protest increases state investment in cutting-edge technological developments. Unsurprisingly, authoritarian states have been importing facial-recognition technology and AI from China. Crudely, the surveillance state has incentives to push out the frontier of surveillance technology. This is the nightmare scenario of technical progress and authoritarianism becoming mutually reinforcing.

Precisely because total surveillance relies on bringing together information on everything anyone does, an actor with a monopoly over coercive power is best fitted to fill the role. The most, perhaps, each individual can do to protect themselves in such conditions is continuously to film their interactions with everyone else. We see signs of that among cyclists in Western cities, but who knows whether authoritarian governments would permit it. In any case, even so equipped, the individual has no sanction other than via mechanisms, such as courts, provided by the state. That is feeble compared with the way China’s Party-led state backs its surveillance with a system of ‘social credit’, rationing access to services according to the correctness of a person’s conduct.

That is all relentlessly bleak. Can no good come from the new technology?  Of course it can, and in ways that might help advanced-economy liberal democracies mitigate a problem that seems internal to their very success. Known to economists as Baumol’s cost disease, it shines an extraordinary light on the way services (private and public) have grown to account for a large share of aggregate output (value added as measured by GDP) without achieving anything like the productivity growth familiar in capital-intensive manufacturing.

Baumol’s thesis, first presented in the 1960s, holds that, in order to remain viable in the labour market, firms operating in sectors with relatively low productivity growth will find themselves needing to offer wage increases that are not backed by efficiency improvements, so the prices they charge must rise relative to goods in higher productivity-growth sectors. Where demand for those services is sensitive to price, demand will fall, and the sector(s) will shrink relative to the economy as a whole. But where demand is (relatively) insensitive to price, the sector will become a bigger and bigger part of the economy. It is easy to think of examples of both. String quartets can hardly deliver productivity improvements — Beethoven’s extraordinary late quartets are not improved by playing them twice as quickly so that more of them get performed in a concert. So, other things being equal, if prices rise, demand falls, and the performance of chamber music accounts for a smaller share of economic activity.

By contrast, to take a straightforward example, education’s share in economic activity will tend to increase. That is because although it struggles to achieve productivity improvements (increased class sizes typically reduce quality), demand cannot fall because education is compulsory up to a specified age. In either type of case, if organisations buck Baumol’s prediction by maintaining stagnant wages, teachers, musicians and others become less prosperous, incentivising them to seek work in other sectors. That risks adverse selection in the labour market, which matters a lot in professions like teaching unless there are enough people with strong vocational motivations.

While the most frequently cited examples of these problems come from the public sector, casual observation suggests there are plenty of instances in the private services sector, perhaps especially mass retail services. Take retail banking. First, branches with staff who know you are replaced by call centres. Then, the call centres become less flexible in dealing with problems caused by the firm itself, perhaps because aspects of the service have been automated in ways that are opaque to the faceless helper. For the economy as a whole, the effect is subtle but profound. What can superficially seem like an efficiency improvement involves, in fact, a deterioration in service when things go wrong; an economist’s notion of production efficiency operates across all states of the world, good and bad, not only how things go when the sun shines. In principle, such quality deteriorations should be recorded by the national statisticians as hidden price increases. If they miss that, we effectively suffer disguised inflation.

If the moral of the story so far is that technology can sometimes deliver illusory efficiency improvements, the question here is whether the new technology, notably AI, can do better. A reasonable guess would be that it can, but subject to some caveats that, if recognised, could help us make the best of things.

For things like call centres, the guess is that AI could be used for routine enquiries, helping to reduce costs while improving service quality. Concretely, AI might massively reduce waiting times when people call into a call centre since one would not have to wait until a costly human being was free. But precisely because some types of problem occur rarely, and have idiosyncratic features, the caveat is that AI might be little better than a fixed-script human at helping with serious but very infrequent lapses in service. Hence, my guess is that devoting more people to intricate customer problems could improve quality in the adverse tail of the distribution of customer services. Unless something like that happened — involving a person with access to AI trying to solve tricky problems rather than sticking to a fixed script — I suspect the net experience would be worse for many people.

An element of what is at stake here is the corporate cynicism involved in disguising price increases. This was brought home to me a few years ago when I heard an apparently kindly American businessman respond to a perceptive question about the hazards of entering retail services with words close to, just let the customers look after themselves. Whatever one’s personal response to that, for the economy as a whole it matters whether our official statisticians manage to recognise subtle but meaningful quality deteriorations when splitting growth in nominal output (measured in money) into real growth and inflation.

Quite how these tensions play out will vary across jurisdictions according to the nature of their political system and culture. Whether we like it or not, I suggest it is not far off to say the US has become an aristocracy, France a monarchy, and Britain a republic despite itself. Taking just the US example (because it is both the home of so much innovation and also still the most powerful state in the world), it has settled into an equilbrium not dissimilar from ancien regime France in so far as it has an aristocracy with two branches: one of wealth, and one of high culture, variously competing or cooperating to sponsor those holding government office.

On the face of it, one branch of the US aristocracy will put riches before liberalism, and will seek to tame the cultural elite’s rival ideological agendas using the lever of donations. So perhaps one’s prior should be that, net, the US elite will prioritise technological progress and power over the health of the Constitution’s core institutions, and it has looked that way for a few decades now. But in that case, the interesting question becomes how American elites will square their riches-versus-liberalism trade-off with their widespread worries about the rise of China, which is set for a century-long contest with the West and democratic East Asia. Like Britain’s long-18^th century struggle with France, the struggle will be in everything, everywhere.

In particular, will the US aristocracy learn from their early-1990s’ absurd ‘end of history’ moment, when key western capitals, led by Washington DC, assumed economic progress in China and elsewhere would lead inexorably to political liberalism? Now the stars reveal the opposite prospect: that liberal democracy might wither at the hands of technology. If, then, we want to prioritise ‘hanging on to who we managed to become’ — the provisional title of my current book project — we had better try to persuade those among us who prioritise above everything their own untrammelled freedom to become rich that, in all sorts of ways, they are relying on the hard-won freedom of the rest of us, just as we rely upon them. That project needs to be embarked upon without an antagonism that falls into the trap of treating friends as enemies, and without oversimplifying either our predicament or our manifest opportunities.

As Hume taught us at the beginning of the commercial age, even though the origins of moral norms and political institutions lie in the mutual advantage of those capable of harming each other, our capacity to learn to occupy something like a common point of view and to internalise some social norms as values helps detach us from the crudest calculus of personal interest. But the free-riding knave is always with us, and perhaps never more so than when technological change alters the balance of power within and between societies.