‘Nearly all men can stand adversity, but if you want to test a man’s character, give him power.’ Abraham Lincoln
How do the strong dominate the weak? By exerting power: the bedrock of political order and the godfather of political concepts. Any serious effort to understand the future of politics must involve an investigation of what power is, what forms it will take, and who will wield it. That’s the work of the next five chapters.
I suggest that power in the future will take three forms. The first is force (chapter six). The second is scrutiny (chapter seven). The third is perception-control (chapter eight). Digital technology will increasingly be the main source of all three. This means that those who control the technologies of force, scrutiny, and perception-control will be powerful; and those who don’t will be correspondingly powerless. My contention is that, over time, power will grow more concentrated in the hands of the state and large tech companies.
In this chapter our aim is to clarify what we mean by power and take a closer look at the relationship between power and digital technology. The vision that unfolds may seem a little heartless in places, focused on the brute fact of power rather than its legitimacy. Don’t worry: the rights and wrongs come later.
Power is a difficult concept to define, and it’s tempting to plump for the definition once given by a judge to the term hardcore pornography: I know it when I see it. But while this sort of laxity might pass muster in a US court of law, political theorists try to be a little more precise, particularly given that power is not always something you can see. We begin, instead, with the simple distinction between having power over someone and having power to do something.1 Power over is the power of a boss who tells her subordinate to perform a task, or a schoolteacher who orders her pupils to sit quietly. It’s about a ruler getting the ruled to comply. Power to, by contrast, does not imply a system of rulers and ruled. It refers to the ability, facility, or capacity of a person to do something—to walk down the street, for instance, or lift a heavy barbell. It is a broad definition. With power to, the fact that one person or group is powerful does not require that someone else is correspondingly powerless.
Both of these conceptions of power are linguistically acceptable. Neither is wrong. But at this stage of our inquiry, power over is more relevant than power to for two reasons. To begin with, if we want to understand the future of politics then we need to know who will be powerful over whom, what forms that power will take, and to what ends it might be directed. Our concern is with the power relations between people rather than the ability of each individual to do as he or she pleases. Second, power to is similar enough to the distinct concept of liberty that we can more sensibly tackle it under that heading in Part III.
Adopting power over as our primary focus, then, what does it mean to say that person A has power over person B? There are several possible answers to this question, but it’s worth starting with the intuitive one offered by the great political scientist Robert Dahl: ‘A has power over B to the extent that he can get B to do something that B would not otherwise do.’2 This definition covers a variety of situations, from the power of the state to lock up its citizens to the power of a mother to order her child to bed. Implicitly, it also covers situations where A stops B from doing something that B would otherwise do. But Dahl’s definition is not without its problems. If, through my erratic driving, I inadvertently force you to change lane on the freeway (which you would not otherwise have done), have I exercised power over you? Or would we say that I only exert power when I intend to? What about if, in a fit of rage, I demand that my girlfriend returns the birthday gift that she kindly bought for me—even though I like the gift and my welfare would be better served by keeping it? Do I still exert power when it’s not in my interests to do so?3 These are questions on which people may legitimately disagree and I don’t propose to get bogged down in them. We should, however, ask whether power is always a matter of getting someone to do something they would not otherwise do. I have no intention of stealing your wallet, but does that mean that the criminal law has no power over me because it only prevents me from doing something I would never have done anyway? This seems strange. For this reason, it would seem sensible to loosen Dahl’s definition to say that A has power over B to the extent that he has the capacity to get B to do something that B would not otherwise do.
So what makes a person or entity powerful? That depends on three factors. First, if their power has a wide range, in terms of geography and the number of people and issues it affects. Thus a national government, whose power extends over all the people within a territory across a wide range of matters, is more powerful than a headteacher whose power is confined to the boundaries of the school and issues relating to its administration. Second, a powerful person or entity is one whose power pertains to matters of significance. The power of a judge to determine the liberty or rights of others is greater than the power of the cafeteria server who determines how much potato mash you receive at lunchtime (important though that may seem to you in the hunger of the moment). Finally, the stability of power is important. Your power to pin me to the floor until I wriggle free is less stable than the power you would obtain by chaining me to the wall in a dungeon.4 Pulling together the threads:
A person or entity is powerful to the extent that it has a stable and wide-ranging capacity to get others to do things of significance that they would not otherwise do, or not to do things they might otherwise have done.
This definition is important. We’ll refer to it throughout the book. Together with the understanding of politics set out in chapter four (‘the collective life of human beings, including why we live together, how we order and bind our collective life, and the ways in which we could and/or should order and bind that collective life differently’) it’s broad enough to include new forms of power that might not traditionally have been thought of as ‘political’.
Power comes in a number of forms. To illustrate this, imagine that Matt intends to shoot and kill his friend Larry, but a third person, Kim, stops him. In one scenario, Kim wrestles Matt to the ground and confiscates the firearm, allowing Larry to escape unharmed. This form of power we call force: where Kim strips Matt of the choice between compliance and noncompliance. In another scenario, Kim tells Matt, ‘If you murder Larry, I will burn down your house.’ Matt decides not to shoot Larry because he does not want his house burned down. In this instance, Kim’s power takes the form of coercion: Matt complies because he fears punishment or deprivation. Another alternative is for Kim to persuade Matt that shooting Larry would be morally wrong. If she does so successfully, then Kim has exerted influence: securing Matt’s compliance without resorting to a threat of deprivation. Next, imagine that Kim is a religious leader who sees Matt on the verge of slaying Larry and cries, ‘Don’t shoot!’ Matt complies, not because he fears deprivation or even because he believes that slaying Larry would be wrong, but because he respects Rabbi Kim’s authority as a source of moral commands. Finally, imagine that Kim stops Matt from murdering Larry by falsely promising him a million dollars if he refrains. In this situation, power takes the form of manipulation—where Matt complies because he has not been shown the true basis of the request being made of him.5
Force, coercion, influence, authority, and manipulation are helpful ways of thinking about different forms of power. But we shouldn’t think them inherently right or wrong—that influence and authority, for instance, are morally preferable to force or manipulation. All forms of power may be used for good or ill. Force can be legitimate (a police officer who handcuffs an escaped convict), influence can be malign (ISIS propaganda aimed at young Muslim men), and authority can be abused (paedophile priests). Even manipulation can be morally justifiable: think of an undercover agent who deceives to infiltrate a criminal enterprise.
We turn now to the relationship between digital technology and power.
All digital systems run on software, also called code. Code consists of a series of instructions to hardware—the physical stuff of technology—telling it what to do. It is written in programming language, not natural language of the kind understood by humans. But the word ‘language’ is apt to describe it, because it also has its own formal rules of grammar, punctuation, and syntax.6 One important difference between code and natural language is that code is meant to be much more precise. It seeks to generate unambiguous commands with no grey areas or room for interpretation.
The key thing to remember about digital technology is that it can only operate according to its code. If you ask a calculator what 5 + 5 is, it can only answer 10. This will be true regardless of how much you might want it to give another answer. The wishes of human users are irrelevant to the functioning of code unless and until it is programmed to take those wishes into account. For the same reason, you can’t use a calculator to draft an email.7 Asking a particular technology to do something it’s not programmed to do is like walking into a closet and asking it to take you to the fifth floor.
Code and algorithms are often referred to interchangeably, but they’re not strictly the same thing. The word algorithm can be traced back to the ninth-century Persian mathematician Abd’Abdallah Muhammad ibn Mūsā Al-Khwārizmī. The translation of his name, algorismus, came to describe any mathematical approach to reasoning, calculating, and manipulating data.8 Today, the word algorithm describes a set of instructions for performing a task or solving a problem. It need not be written in computer code. A set of driving directions is one form of algorithm, specifying what to do under various conditions: ‘Go down the street, turn right at the post office and then left at the lights. If the traffic is bad, take a right at the garage and take the second exit at the roundabout . . .’ Relationship advice often takes algorithmic form: ‘If he continues to lie, break up with him. But if he apologizes, cool off for a bit, see how you feel, then try to talk things through . . .’
When we talk about digital technology, the algorithm is the formula and the code is the expression of that formula in programming language. A great deal of code contains algorithms, whether for making decisions, learning skills, finding patterns, sorting data, or predicting events.
What’s this got to do with power? Well, when we interact with digital technologies we also necessarily submit to the dictates of their code.9 To take a simple example, you can’t access a password-protected document unless you enter the correct password: the machine has no choice or discretion in the matter, and neither do you.10 It makes no difference that the document contains vital medical information that could save your life. It also doesn’t matter that the only reason you don’t know the password is because you forgot it. The code won’t allow you to do what you would otherwise do.
We regularly come up against the constraints of code. Think about digital content such as music, films, and e-books. In 2009, President Obama gave the visiting UK Prime Minister Gordon Brown a gift of twenty-five classic US films. Returning to London, however, the most powerful man in the land found that the movies wouldn’t play on his UK DVD player.11 Why? To the untrained eye, it would seem like a bug or glitch. But the opposite was true. The prohibition had been coded into the DVDs by their manufacturers and distributors, to protect their commercial interests and enforce the law of copyright. This is usually referred to as Digital Rights Management (DRM).
Because of code’s ability to direct our conduct in a finely honed way, many distinguished thinkers, following the pioneering work of Harvard professor Lawrence Lessig, have argued that code is law (or at least that code is like law).12 Thinking about the DVD example, it’s clear that code exerted a kind of force: Mr Brown was entirely deprived of the choice whether or not to play the DVD. But it could just as well have been coercion. Imagine that you send your sister an email containing a digital music file that you downloaded from iTunes. That music may well refuse to play on her computer, because of DRM rules prohibiting the duplication of copyrighted content. Again, a kind of force would prevent your sister from doing what she wanted. Now imagine that iTunes did not actually prevent her from accessing the content, but instead that it punished her for trying to do so, by locking her (or your) account for 24 hours thereafter. If you knew this was going to happen, you might think twice before downloading and sending the music, and your sister wouldn’t try to listen to it if you did. In this example, you both behave a certain way because you fear punishment. That’s coercion rather than force. In this instance, code certainly shares some features with law.
But to say that code is law, or even like law, is no longer enough.
Code can also be used to influence and manipulate people in ways that don’t resemble the workings of the law. As we’ll see, this happens when tech is used to scrutinize people and when it is used to control their perception of the world. A more helpful formulation than code is law is that code is power: it can get us to do things we wouldn’t otherwise do by means of force, coercion, influence, and manipulation. And it can do so in a stable and wide-ranging way. That’s why code will play such a crucial part in the future of politics.
In 2006 Lessig wrote that ‘Cyberspace is a place. People live there . . . and then at some point in the day they jack out and are only here . . . They have returned.’13 Typically prescient at the time, this way of thinking is now outdated. In the digital lifeworld ‘cyberspace’ will be a less helpful way to talk about our interaction with technology. As we’ve seen, our lives will play out in a field of connected people and ‘smart’ things with less distinction between human and machine, online and offline, virtual and physical. Whether we want to interact with digital technologies, or even whether we are aware of them, will matter less and less.
I suspect that the distinction between ‘cyber’ and ‘real’ is already losing most of its psychological resonance for younger readers, for whom chatting on WhatsApp is just chatting, not ‘cyber-chatting’, and buying stuff on Amazon is simply shopping, not ‘e-commerce’. (We long ago stopped speaking of Amazon.com.) In the long run, the distinction between cyberspace and real space will lose its explanatory value. There’s nothing ‘virtual’ about the digital pacemaker keeping you alive. No one would say that a passenger in a self-driving car was travelling in ‘cyberspace’. While in 2000 you might have been able to leave code’s empire by logging off and shutting down, in the digital lifeworld that will be much rarer. Trying to escape the reach of technology will be like trying to escape the law: possible in theory by fleeing to the wilderness, but highly impractical if you want to live a normal life. And as virtual reality (VR) systems grow in popularity, some of us will spend time in universes entirely constituted by code—where code is not only power, but nature and physics too.
Much of the code that surrounds us in the digital lifeworld will be able to reprogram itself and change over time as it learns to recognize patterns, create models, and perform tasks. As we saw in chapter one, machine learning and AI systems more generally will become more autonomous and ‘intelligent’ in ways that do not necessarily mimic human intelligence or the intentions of their original programmers. They will also be much more common. This will profoundly affect the way that code exerts power. The academic literature has traditionally treated code as a kind of stable architecture, setting inflexible side-limits on what we can and can’t do. This is reflected in the structural metaphors we use to talk about code: ‘platforms, architectures, objects, portals, gateways’.14 In the future, code will be a much more dynamic, sensitive, and adaptable referee of our conduct—capable of changing the rules as well as enforcing them. Sometimes it will seem ingenious or bewildering. Sometimes it might seem irrational or unfair. But that’s power.
Digital technologies exert power on us when we interact with them, by defining what we can and can’t do, by scrutinizing us, and by controlling what we perceive of the world. In the digital lifeworld, such technologies will be everywhere. The code animating them will be highly adaptable and ‘intelligent’, capable of constraining us in a flexible and focused way. Therefore certain digital technologies will provide a means of exerting great power in the digital lifeworld, and those who control these technologies will exert great power through them. Even if no particular person or group can be said to be ‘in charge’ at a given time, humans will be constantly subject to power from many different directions, constraining and guiding their behaviour. That’s the essence of power in the future, and it’s the subject of the next four chapters.