‘A well-ordered society is . . . one that is effectively regulated by a public conception of justice’
John Rawls, A Theory of Justice (1971)
The pursuit of justice can inspire us to great acts of sacrifice and devotion. The sight of injustice can move us to unspeakable sadness and rage. The political theorist John Rawls called justice the ‘first virtue’ of social institutions.1 Martin Luther King taught that it lay at the end of the arc of the moral universe. The ideal of justice has never been the exclusive preserve of politics or law. Alone among political concepts, it has a sacred standing, forever inscribed in the commandment of Deuteronomy: ‘Justice, and justice alone, you shall pursue.’
Justice has been the subject of a great deal of political theory but it does not lend itself easily to dispassionate analysis. In our day-to-day lives it’s most often present as a sense or a feeling, even an instinct. Its home is the gut, not the brain. For those struggling at the sharp end of the world’s injustices, even the finest treatises on the subject must seem curiously cold and detached. War and famine, squalor and hunger, ignorance and disease, hatred and bigotry—these are matters of the heart as well as the head. The richest eight people on the planet possess the same wealth as the poorest half of humanity.2 One per cent of the world’s population controls half its assets.3 No political theory, however eloquent, can make full sense of the towering injustice that exists in the world. No scholarly argument can match the primal urgency of Marx’s old battle-cry: ‘I am nothing and I should be everything.’4
Yet political theorists are not entitled to the luxury of getting carried away. There’s already enough bluster and confusion in the world. Sometimes the cause of justice is better served by an open mind than a clenched fist. It can be studied calmly, even clinically. It can even be studied algorithmically. The aim of a theory of justice, as Rawls put it, is to ‘clarify and to organize our considered judgments about the justice and injustice of social forms’.5 My aim in this part of the book is similar, if more modest: to clarify and organize our ideas about social justice from the viewpoint of the digital lifeworld.
The next three chapters are all about what I call algorithmic injustice. The argument is that questions of distribution and recognition—the very soul of social justice—will increasingly be settled by algorithms contained in code. Chapter seventeen is about technological unemployment: what does political theory have to say about a world without enough work? Finally, chapter eighteen looks at the risk that wealth in the digital lifeworld could increasingly funnel into the hands of a tiny group of people and corporations, and asks how a change in our understanding of property could stop that from happening.
The precise meaning of the term social justice is contested, but broadly speaking it’s used to describe the idea that everyone should get what they’re entitled to—no more and no less. It can be distinguished from its sister concepts in law—criminal justice (the punishment of those who break the law) and civil justice (the resolution of legal disputes between parties) although they all spring from the same conceptual pasture.
Justice and equality are not necessarily the same thing. Many ancient and medieval thinkers believed that humans were inherently unequal, not only in talent and strength but in their fundamental worth and moral status. On this view, serfs, slaves, and women should not be treated equally: rather, they should live quietly under the yoke of their masters and betters.6 That’s what justice meant in those times. The principles were said to be derived from tradition, or from the ‘law of nature’, or even from the word of God. ‘It is thought that justice is equality,’ trolled Aristotle in the Politics, ‘and so it is, but . . . only for those that are equal.’ 7
These days, by contrast, social justice is closely associated with the idea of equality. Egalitarianism is the belief that people should be treated as equals in important respects, although egalitarians themselves don’t always agree on what these are. Some will say that justice requires an equalization of wealth, for instance, while others will say it only requires equal opportunities to acquire wealth. Most theories of social justice share at least one egalitarian belief: that all humans are equal in fundamental worth. Few nowadays would agree with Aristotle that some lives are worth less than others. This consensus, rare among political theorists, is called the egalitarian plateau.8 In the next few chapters I do not depart from the egalitarian plateau. I take it as self-evident that all humans are essentially of equal cosmic value, and ask what the consequences might be for social justice in the digital lifeworld. As we’ll see, justice doesn’t always translate to material equality. There may be good reasons, just reasons, why inequalities in wealth or status should still be permitted. And justice certainly doesn’t mean treating everyone the same.
There are two main ways to think about social justice. Justice in distribution, as the name suggests, is about how assets should be distributed in society. Justice in recognition is about how humans see and treat each other, including our systems of social status and esteem. We begin in this chapter with distribution.
What gap, if any, should there be between rich and poor? How much tax should the ultra-wealthy pay? Who should benefit most from government spending? These are classic questions of distributive justice. They prompt a range of possible responses.
Let’s say that you don’t mind about the gap between the rich and the poor but you do believe that no one should go hungry or live on the streets. Everyone need not have the same, but everyone must have enough.9 If this is your view, you’re a sufficientarian.10
Many would argue that the neediest in society have a stronger moral claim to resources than the rich. So if the government has half a million dollars to play with, better to build a youth centre in a deprived area than a leisure complex in a prosperous suburb. Some go a step further, arguing as a general rule that the less well-off you are (even if you are not genuinely needy), the stronger your moral claim to society’s resources.11 This kind of belief is usually called prioritarian. Like sufficientarians, prioritarians are not concerned with the overall gap between rich and poor, as long as the less well-off get priority.
Others say that distributive justice demands yet more, namely equality of opportunity, meaning that irrelevant factors like gender, sexuality, or ethnicity should not count against people in their pursuit of valuable things like a job or education.12 Equality of opportunity is closely linked with another conception of justice, justice as desert. It means that people should get what they deserve. If you work hard and play by the rules then you should be rewarded. But if you laze around all day then you should get a smaller share of the pie. ( Justice as desert differs from a less well-known principle, justice as dessert, in which everyone receives a pudding of their choice.) Combining the equality of opportunity and justice as desert approaches, you might say that a just distribution of resources is one in which everyone has an equal opportunity to get what they deserve.
Many egalitarians question whether qualifications or other ‘objective’ indicia of merit are actually just guides to what people deserve. Consider a brilliant and hardworking kid from the wrong side of the tracks who leaves high school with patchy grades, despite her best efforts. Her less-gifted cousin might graduate from a private school with better grades and a range of extra-curricular baubles. If both apply for the same college place, the rich student is plainly better qualified, but does she really deserve the place more? In this instance, true equality of opportunity requires more than merely abstaining from racism or bigotry. It calls for affirmative action, taking into account the socio-economic conditions in which the cousins’ qualifications were earned (or not earned, as the case may be).
Another perspective on desert is to say that instead of looking at merit to decide what people deserve, we should try to reward morally deserving or socially useful conduct, as opposed to work whose value is only valuable in commercial terms. On this view, nurses and teachers should earn just as much as financiers and corporate lawyers.
From a still more radical position, we might question whether any of us actually deserve the qualities that determine our financial success in life. The fact that I am ugly and talentless is no more my fault than the fact that you are beautiful and gifted. It’s also bad luck for me that the market, left to itself, will tend to distribute more of the good stuff to beautiful and gifted people than to ugly and talentless ones. The luck egalitarian tradition sees this as a matter of injustice: we shouldn’t be punished or rewarded for matters beyond our control. That includes genetics and upbringing. If the luck egalitarians are right, then the genetically well-endowed should be stripped of their unearned riches and the less fortunate should be compensated for their bad luck in the lottery of life.13
One objection to luck egalitarianism is that it fails to take into account different work ethics. If you and I are equal in talent, but I work harder to make the most of myself, then surely I deserve more than you! Perhaps. But only if you believe that the desire and capacity for hard work are themselves things we choose or deserve, rather than attributes we are born with or the product of upbringing. If they too are a matter of luck, then (on the luck egalitarian argument) neither of us deserves more than the other.
So does this mean that the talentless and lazy should receive the same share of society’s resources as the talented and industrious? Surely benefits and burdens should still be allotted according to choices for which people can be held morally responsible. If I take a gamble and it pays off, I should receive the rewards of that risk—but likewise I must be held responsible for my bad bets and pay the price when they go wrong.
Alternatively, it might be said that if talent and hard work are to be harnessed for the benefit of society, then they need to be incentivized with higher rewards even if that generates some inequality. In the absence of economic incentives, it’s often said, people would be less inclined to work hard, take risks, or innovate. The economy would stagnate and we would all be condemned to receive an equal but slender slice of a very stale economic pie (that’s a metaphor for the wealth of a society, not a return to justice as dessert). John Rawls made a claim of this type in A Theory of Justice (1971), arguing that social and economic inequalities are permissible so long as they are to the greatest benefit of the least advantaged. It’s okay if the gap between rich and poor grows, but only if the result is that the poor get richer faster than any other group. Rawls’ principle brings us back to the first issue touched on in this section—whether distributive justice is more concerned with absolute deprivation or relative inequality. There are certainly reasons why relative inequality matters. It can erode the common ties of humanity that bind us together. It can cause the less well-off to feel misery and shame. And it can prevent the poor from participating fully in social and political life.14 That’s why many radical egalitarians say that to ensure equal opportunity for children, the state must equalize outcomes for adults, whose wealth determines the privilege of their offspring.15 They seek a society in which everyone has the same.
On the other side of the political spectrum, many are hostile to the very idea of egalitarianism, at least insofar as it requires the redistribution of social goods. One view, advanced by Edmund Burke, David Hume, and Friedrich Hayek, is that the pursuit of equality would require a disastrous amount of state intervention in our lives.16 In Hume’s opinion:17
Render possessions ever so equal, men’s different degrees of art, care, and industry will immediately break that equality. Or if you check these virtues, you reduce society to the most extreme indigence; and instead of preventing want and beggary in a few, render it unavoidable to the whole community.
Other non-egalitarians, like Robert Nozick, say that any artificial redistribution of wealth unjustly interferes with individuals’ overriding right to their own property. For Nozick—possessed of a remarkable mind and equally remarkable eyebrows—the justice of any distribution of assets depends only on how it came about, and has nothing to do with how much anyone has. The only question is whether a particular resource was properly acquired and freely transferred (that is, given and received in exchange for something, or as a gift). If so, then the state has no right to take it from me and give it to you, no matter how deserving you may be. ‘We are not in the position of children who have been given portions of pie by someone who now makes last minute adjustments to rectify careless cutting.’18 (Like all good theorists, Nozick enjoyed a dessert metaphor—but would have felt queasy at the thought of an egalitarian gateau.) Blockchain technology, the most advanced system yet known for securely recording the complete ownership trail of a piece of property, would probably have fascinated Nozick. The lawful chain of ownership, recorded immutably on the Blockchain, would itself demonstrate that justice had been done—irrespective of how much everyone ended up with.
How do questions of distributive justice play out in the real world? By what mechanisms do we decide how social goods are actually distributed? The answer is through the interplay of the market system and the state.
In modern times the default way of distributing resources is using the market system. In a pure market system there is no central decision-maker directing where resources should be allocated. Instead, the work of distribution is done by every participating individual. Resources are privately owned, swapped for other goods, traded for money, and exchanged for labour. As Adam Smith famously observed in Wealth of Nations (1776), the market system works on the basis that people will generally pursue their own self-interest: ‘It is not from the benevolence of the butcher, the brewer, or the baker, that we expect our dinner, but from their regard to their own interest. We address ourselves, not to their humanity but to their self-love.’19
When it comes to distributing economic resources, the market is reasonably efficient. It allows for resources to find their way to the places where they are wanted most, at least as indicated by how much buyers are prepared to pay. (This is, of course, a imperfect measurement of demand: the billions who lack electricity in their homes, and would like to have it, would say that the market system does not really reflect their demand.) The market gives people a degree of freedom to do as they please with the resources at their disposal. The tragic flaw, however, is that markets generally distribute resources efficiently but not necessarily justly. There is nothing in the logic of the market to ensure that the poorest get enough, or that they get priority, or that there is equality of opportunity, or that people get what they deserve (whichever principle of distributive justice you prefer). Allocation according to how much people are willing to pay will not mean that resources get to where justice requires that they go. Some kind of external intervention in the market system is always needed to ensure a semblance of egalitarian justice.
That’s where the state comes in.
One of the state’s main economic functions is to regulate the market, specifying what can be owned, bought, sold, and under what conditions (including, for instance, the rule that people should not be bought and sold). This prevents some of the most egregious injustices. The state often does a lot of its own spending too, enabling it to direct resources to the places required by justice (for example: welfare payments for the unemployed or sick).
One of the great political debates of the twentieth century concerned the extent and nature of state intervention in the market economy. At one extreme, some favoured a ‘command economy’ in which the state did most of the spending and determined what people earned, what was produced, and at what price. At the other end of the spectrum were those who supported a ‘night watchman state’ that did as little regulating and spending as possible. There are a range of options between these poles.
What is the future of distributive justice? I suggest that, in a substantial departure from the past, algorithms will play four vital roles. But we should recap briefly before I introduce them. Code, you may recall, refers to the instructions given to hardware—the physical stuff of technology—telling it what to do. Digital systems can only operate according to their code. An algorithm is any set of instructions for performing a task or solving a problem. An algorithm can be expressed in code, that is, in programming language that digital systems can understand. Machine learning algorithms are algorithms capable of ‘learning’ knowledge and skills, including how to find patterns, create models, and perform tasks. Some have the potential to act in ways that their human creators did not foresee- and many are high-performing in ways that do not necessarily mimic human intelligence.
First of all, in the digital lifeworld algorithms will increasingly be used, along with the market and the state, to determine our access to important social goods. This makes algorithms a new and important mechanism of distributive justice.
Let’s begin with access to work. Work is (for now, at least) the main way that most of us get the money we need to live and flourish. Even now, applications for jobs are routinely decided by algorithms. Seventy-two per cent of resumés ‘are never seen by human eyes’.20 Instead, algorithms are used to scan applications and determine which candidates have the requisite skills and experience. Other algorithms are used to make determinations about candidates’ character and ability outside the formal application process. They do this by chomping through whatever personal data is available and drawing their own conclusions. The data used may have no direct connection with employment. A candidate’s online browsing activity, or the ‘quality’ of their personal connections on Facebook, may determine whether their application succeeds or fails.21 Machine learning algorithms can also be used to discover personal traits—habits, addictions, weaknesses—that human recruiters assessing paper applications could never possibly have known about. Algorithms, in short, will decide whether millions of people access the most precious thing the market has to offer: a livelihood.
Once a person has a job, it will become more common for their work itself to be measured, monitored, and assessed using algorithms. They’re already used to predict when employees are likely to resign; it’s expected that they will be used, more and more, to scrutinize general performance and determine pay and promotions.22 The world’s largest hedge fund, Bridgewater Associates, is building an AI system to run the day-to-day management of the firm, including the scoring and dismissal of staff.23
Algorithms will also play an increasing role in determining our access to other resources of critical importance. They’re already used to decide our credit scores and whether we would make good tenants.24 Insurance companies use algorithms to estimate when we are going to die. Automated predictions of health risk (perhaps based on data relating to our food consumption) could come to determine how much we pay in premiums.25 Data on our driving skills might be used to determine our access to motor insurance.26
Work, loans, housing, insurance: these are not luxuries. They’re essential social goods. All rational citizens want them, regardless of whatever else they want.27 Those who have them live comfortable and prosperous lives. Without them, life can be very difficult. How they are distributed, and according to what criteria, are of central importance to the future of distributive justice.
Using algorithms and data to make these decisions is not inherently a bad thing. On the contrary, it’s possible that carefully crafted algorithms could eliminate the biases and prejudices of human decision-makers. With regard to work, for instance, affirmative action algorithms could be used to broaden the pool of successful applicants from beyond the usual colleges and institutions. When it comes to loans, housing, and insurance, algorithms could be used to widen access for those who need or deserve it most. My point, at this stage, is simpler: code (embodying algorithms) is an increasingly important mechanism of distributive justice. It demands close political attention.
As time goes on, we’ll learn more about how algorithms interact with the market and the state, formerly the key mechanisms of distributive justice. Algorithms plainly won’t substitute for them but they will affect their functioning in interesting and important ways. In the examples already laid out, algorithms determine the terms of access to social goods that were previously distributed according to market forces or state intervention. This is a significant change.
Next, algorithms intervene as participants in the market, buying and selling (among other things) billions of dollars’ worth of financial products. This has distributive consequences: the share of overall wealth enjoyed by financial institutions, for instance, has ballooned since the introduction of automated trading.28
Third, algorithms are increasingly used to determine the information available to purchasers. In the United States, more than 80 per cent of consumers search online before buying a product. The results they are shown directly delimit the set of options from which they choose in making their purchases.29 Often algorithms will apply class distinctions: online shopping platforms routinely display advertisements for payday loans to less well-off groups.30 Every algorithm of this kind will benefit some groups more than others. The question is, whose interests should they prioritize? The seller or the buyer? Rich or poor? These are quintessential questions of distributive justice.
Finally, algorithms increasingly intervene in the most fundamental machinery of the market economy: the price mechanism. Customers may be charged more or less depending on where they live (outlets like Staples charge differentially for the same product depending on the postcode of the buyer),31 the time of day (gas stations charge more in hours of peak demand),32 and the weather outside (vending machines can use algorithms to vary prices according to the heat).33 These are simple examples but the possibilities are much more radical. Research suggests that if Netflix took into account its customers’ online behaviour (5,000 variables, including how frequently they visit IMDB and Rotten Tomatoes) it could increase its profits by more than 12 per cent.34 An extreme end result would be ‘person-specific pricing’ whereby algorithms are used to charge customers the precise maximum they are prepared to pay at that moment.35 This kind of practice was previously impossible. Vendors never had enough information about buyers and it wasn’t feasible to change prices so rapidly. Not so in the digital lifeworld. When I buy my groceries using a digital application it might be hard to know if the price displayed has been tailor-made for me, based on my past spending habits and whatever else the vendor has learned about me.
Algorithmic interference with the price mechanism raises profound questions of distributive justice. Is it just for people to pay different prices for the same thing? The market already fails to guarantee that the least well-off will get enough or receive priority in the allocation of scarce social goods. Nor does it ensure equality of opportunity, or that people get what they deserve. Person-specific pricing could make this worse. A sick man might be prepared to pay his entire life’s savings for a simple operation to save his life, but does that mean that’s what he should pay? Perhaps algorithms could be used for redistributive purposes, with the rich paying a little more, and the poor paying a little less, for the same goods.
This could also been seen as an issue of liberty. The price of goods determines, to a meaningful extent, what we can and can’t do in life.36 According to the old quip, the Ritz is open to rich and poor alike—but we know that, in reality, only the rich are free to enter because they have the resources to pay. When we mess with prices, we also mess with liberty.
For centuries, we have debated the economic rules that ought to apply to the market and the state. We must now do the same for code. What should be seen as acceptable, what should be regulated, and what should be forbidden? More on that in chapter sixteen.
Now, the second type of justice: recognition.