The Invidious Hand

Social justice in the age of control

This is the first essay in a three-part series for a project with the Joseph Rowntree Foundation, Social justice in a digital age. You can read more context here.

When Facebook launched, we thought little of it. I remember signing up one day at university. A few clicks, an account created. At the time, we just thought of it as an online service, like Google before it. The fact that it was free was a curious bonus.

Like the fable of the frog being boiled slowly, the effects of digital platforms have been easy to miss. As Shoshana Zuboff reminds us, when the car was first invented we called it a horseless carriage, misconstruing a revolutionary invention as nothing more than a modification of an old one. Today, we do the same thing with digital platforms; Amazon is an online store, Uber is an app-based taxi service. The modifiers once again reveal our complacence.

In this essay I want to see if we can describe what has changed with the rise of platforms, so that we can better evaluate our predicament. My interest is in what platforms mean for social justice, so I’ll focus on two implications, for consumer markets and work. And since I’m interested ultimately in making things better — understanding our situation so that we can improve it — I’ll end with what this means for the way we govern.

Before we get there, though, we have some foundations to lay. For all the attention we pay to big technology platforms, we rarely stop to ask a basic question. What is Facebook? What does a platform do?

When we talk about the internet, we have a habit of eliding two concepts. We refer to Google, for example, as if it’s one thing when really it’s two: an organisation and a platform. The same can be said of Uber, Amazon, or Airbnb.

Take Uber as an example. It’s an organisation, with the legal status of a private company: a group of people who are collaborating in order to turn a profit. But Uber is also the thing these people are building: a digital platform — software that allows us to do things. Just as we would never blur the idea of Ford, the company, with the cars that Ford makes so we need to distinguish Uber the organisation from Uber the platform. As with Ford, the two have different social and economic implications.

If we look at the company side of platforms like Uber, Netflix, or Google what we see is a new practice of production: a cluster of complementary methods, mindsets, and organisational forms that are designed to make the most of digital technology. These are today’s equivalents to the practices and forms that were developed by people like Frederick Taylor, Henry Ford, and Alfred Sloan in the early-to-mid twentieth century to make the most of their generation’s technological frontier. Back then, the revolutionary practice of production was the assembly line (exemplified by Ford) and the new organisational form was the modern corporation (exemplified by General Motors), with its hierarchies, waterfall processes, and silos. Today, the new practice of production is agile product development and the new organisational form is the agile company in which the unit of delivery is an autonomous, multidisciplinary product team.

The emergence of a new practice of production is a development of world-historic significance. It is coming to define the 21st century just as mass-production defined the 20th century, shaping the era’s social character, aesthetics, and approach to problem-solving. Mass production eventually changed almost every aspect of society and the same is happening with digital methods. Ultimately these new ways of working will influence everything from art to architecture to the way we think about ourselves.

So what about the platforms that these companies spend their time designing, building, and optimising? The platform is a thing so new that we don’t yet have the words to describe it. That I’m using the word ‘thing’ shows that even the category is unclear. Is a platform a tool? Or a service? Or an institution? We could argue that it’s all or none of these. And so the best we can do, to avoid the trap of the horseless carriage, is to use metaphor to grasp the kind of thing a platform is, and what its ramifications might be.

I like to think of a digital platform as a portal: a means of accessing an alien dimension, with different laws to our own. This is precisely what is powerful about a platform; it creates a space for us to function in the alien dimension, so that we can make use of its alien laws. So we might say a platform lets us enter Digitalland, and walk in Digitalland, breathing its alien, digital air.[1]

What are the laws of Digitalland? They can be hard to pin down because they’re so different to our own, but some of the basics are clear. Digitalland doesn’t have space in the way Physicalland does, or at least space is insignificant there, because of the possibility of near-lightspeed travel. Digitalland also doesn’t have ownership in the same way as Physicalland. Or, to be more precise it doesn’t have rivalry, because in Digitalland two people, or five people, or a million people, can use the same thing at the same time. Most radically, Digitalland is a place where the substrate of reality itself is open, like text in a book. If you know how to read that text, you can see anything that’s happening in your part of Digitalland. And if you know how to write the text, well, then you can change reality itself. And so Digitalland affords some people a form of omniscience and omnipotence that in Physicalland we could make sense of only by reference to God.

So a platform is a space in which things can be done; things we can’t do — or at least not do as well, or as easily, or as cheaply — as we can in Physicalland. The platform we call Amazon, for example, lets us buy and sell things without being near them. Google lets us do something similar for information, putting all of human knowledge within our reach without us physically leaving the house. This model only works thanks to that lack of rivalry, which means millions of people can use the same information simultaneously. Facebook uses Digitalland’s spaceless qualities to put everyone in the world instantly in touch with each other. Netflix and Spotify, meanwhile, have used the absence of rivalry to assemble libraries of films and music that anyone in the world can use at the same time. And of course all these platforms exploit those powers of omniscience and omnipotence, the implications of which we’ll come on to.

These metaphors help us see what kind of thing a platform is. But their real power lies in combination with one other fact: with a few rare exceptions, the spaces we’re building in Digitalland are commercial endeavours, so they’re architected, calibrated, and optimised for profit.

The commercial status of spaces in Digitalland isn’t a big deal on its own. After all, many spaces in Physicalland are run for profit too. The big deal comes when you bring in those alien laws, particularly the possibility of omniscience, which means a company can know everything that happens in their space, and omnipotence, which means a company can re-architect and recalibrate their space without limit, instantly, and relatively cheaply, which in turn means they can optimise the space in rapid cycles of iteration. This all makes possible, and therefore in a sense inevitable (because how can a company ignore a route to much higher profit?), entirely novel business models.

It’s these business models that explain why markets work differently in Digitalland, opening a new chapter in the history of capitalism. It’s a point in the plot at which the balance tips — subtly at first, and then decisively — to something more malign. With each year that passes, firms put a little less effort into making great products and services, and a little more into predicting, shaping, and constraining our behaviour. It’s a world remade slowly but surely by a powerful force: the invidious hand.

What is it about Digitalland that summons the invidious hand? And what are the consequences of its growing power? To answer these questions, we need to start by defining those new business models.

The simplest and least disruptive business model for a space in Digitland is the one used by digital services like Netflix or Spotify. They use the rather quaint approach of charging people to access their space, which is not all that different to the pre-internet business model of building an attractive space in Physicalland, like a theme park, and then charging people a fee to enter.[2]

Netflix, for example, currently charges UK customers £10.99 a month for unlimited visits to the space they’ve created, which contains films and TV programmes that anyone on earth can watch at the same time. Their incentive is therefore to get people to spend as much time in the space as they can. They do this in some nice ways that would have made Adam Smith happy, for example by filling the space with great films and TV. But they also do it in some invidious ways, such as by designing the space like a flytrap for the mind, so that it’s easy to enter and hard to leave.

The second business model is one that more directly exploits the omniscience possible in Digitalland. These spaces bring to mind a Jorge Luis Borges short story about the world’s biggest junkyard, or an infinite bazaar. In these spaces, you can find anything you want — holidays, life partners, books, builders, leftover food — by finding someone who’s willing to give or sell you that thing. The companies that build these spaces — eBay, Amazon, or Airbnb for example — make money when matches happen there, in that they work on the basis of a transaction fee, or a tax, a bit like a government. These companies are therefore incentivised to maximise the number of matches that take place in their space, or rather to maximise the tax they can get away with charging. Again, they do this in some nice ways, such as by giving people glimpses into their omniscience, to help them find matches. And they do it in some invidious ways, for example they make it hard not to use their space, such as by making people leave any matches they’ve found at the door on the way out, so that they have to come back.

The third, most consequential — and most literally invidious — business model was pioneered by the biggest spaces of all, Google and Facebook. It’s proven so profitable that it is spreading across Digitalland, so that by now it’s less a business model and more just a feature of the way all business is done in Digitalland.

This model works by building a space like a cattle pen, into which you entice people with the offer of a free service that exploits the laws of Digitalland. For Google, the service is being able to access the world’s information. For Facebook, it’s being able to communicate with pretty much anyone, anytime, for free. Once people are in the cattle pen, the service then holds their attention, while the space owner captures their likeness in data, trying to understand their deepest motivations and desires. The space owner then uses these insights to build and refine what we might fancily call an architecture of behaviour modification but might more simply call puppet strings. From then on, each time we enter the space, and sometimes even when we leave, they fit us with the puppet strings so that they can shape our behaviour. More pertinently, they rent out the puppet strings to other companies for a fee and, in time, whether or not the space was originally set up for this purpose, the creation and refinement of these puppet strings becomes its whole raison d’etre.

To name this all so directly feels almost indecent, exposing the naked workings of Digitalland. It’s also hard not to feel bashful at the breathlessness of it all. Can this really be true? Isn’t this just an absurdly overblown caricature of what is really just advertising? When I first read the literature on digital platforms, that’s what I thought. But over time, I’ve come to accept the profundity of the changes underway. I now think there’s less risk that we overstate the significance of digital platforms and more risk that these changes sneak up on us under the cover of old words and concepts, until one day our stomachs fall as we realise what’s happened.

So these days, when someone says they feel uneasy about the power of big tech, I can’t help but hear the frog asking plaintively: ‘is it me, or is it hot in here?’ We’d be wise, I think, to trust our instincts, or at least grab a thermometer to check them.

To get the measure of our predicament, it’s useful to focus not on where we are now but on where we’ll be soon. So let’s talk about the future.

When we look ahead, we can distinguish between two types of change: catch-up change, whereby ordinary companies apply methods that are already used at the digital frontier, and new research and development, which pushes the frontier forward.

Most companies in the economy today fall far short of the capabilities I’ve described. So the main driver of change in the next few decades will be companies closing the gap with where the frontier is now, applying capabilities that are already used by the owners of the most sophisticated spaces in Digitalland.

This process of catch-up sounds boring but, since most of the economy operates years, in fact decades, behind the frontier of what’s possible, catch-up is the main flavour we’d taste if we bit into technological change. It means that, even if we stopped all new inventions now, technological change would still make our lives unrecognisable in a few decades’ time.

Catch-up happens partly through a ‘change or die’ dynamic, as businesses find they have no choice but to adopt new business models or be crushed by competitors that do, and it happens partly through new business creation. Since Digitalland is still a recent discovery, its territory is still largely unexplored, so vast tracts lie open for the taking.

As these processes of creative destruction play out, what seems like incremental change year-on-year compounds into a revolution. Over the last 20 years in the UK, for example, the annual birth rate for businesses has been 12% and the death rate 11%. As this inflow and outflow run with a steady insistence, most of the water in the tank of the economy is soon replaced. Around eight in ten private firms that do business in the UK today, accounting for around half of all turnover, didn’t exist 20 years ago. And so an old economy is, every few decades, all but swapped for a new one.[3]

This takes us to the more tantalising question: where will the frontier go next? It might feel like we need to get out the tarot cards at this point to engage in pure speculation. And it’s true that technology can be hard to predict. It’s a fool’s errand, for example, to guess what new services will be developed using today’s breakthrough technologies; when the iPhone was launched, not many people predicted Uber. But what we can do is place some directional bets, using a tried and tested method: follow the money.

According to technology analyst Benedict Evans, two neologisms dominate the frontier of technology investment in the early 2020s: Web 3.0 and the Metaverse. And when it comes to the latter — which is the more relevant to this essay — people are pursuing two moonshot areas of innovation.

The first moonshot is all about addressing the shortcomings of Digitalland relative to Physicalland. This is important because, when you zoom out, the main thing you notice about the spaces we’ve built so far in Digitalland is just how lame they are. Today’s spaces are impressive only really because they give us access to the alien laws of Digitalland, and those laws are very liberating. But as constructions in themselves, the spaces are rudimentary. It’s as if we’d just discovered an alien planet and started out by building a simple, temporary base there, with tiny cabins, no windows, and no decent toilets or showers.

I find this image helpful to visualise how early we are in our exploration of Digitalland. Picture the first few years after we discover the alien planet. At first, we’d just want to be able to go there and survive long enough to do some things. A simple base would let us do that; we could visit to take advantage of the way things work there — maybe we’d do some farming to exploit the long summers, or maybe we’d enjoy trampolining to make the most of the low gravity. The visits would be task-based; each time we finished, we’d come back to earth. And on the occasions when we stayed longer than intended — maybe we were having too much fun, or had some work to finish — we’d get back to Physicalland feeling all unwashed and unwell, promising ourselves that, next time, we wouldn’t stay so long.

The spaces we’ve built so far in Digitalland are like the alien base but with one key difference: they’re much worse. At the moment, when you spend time in Digitalland, you lack something more basic than nice toilets; you miss deep aspects of what it is to be human. You don’t fully experience spatial dimensions, for example, and you don’t have a multi-sensory and tactile experience. Think of all those Zoom calls, as you stare hour after hour at a flat screen. And consider how you can’t taste an Instagram post, or touch a product on Amazon.

There’s even a literal sense in which the spaces we’ve built so far in Digitalland are cramped and uncomfortable. Just look at the epidemic of back pain we’re experiencing in Physicalland from people craning their necks and slouching their backs during the time they spend in Digitalland.

So you can see why putting this right, and levelling up the spaces in Digitalland so that they at first resemble, and then later mirror, and ultimately surpass, the quality and texture of human experience we enjoy in Physicalland, is a top tier priority at the frontier of capitalism. In fact, this gives us a nice way to characterise our present historical moment. We’re living in a painful interregnum — a time after we’ve discovered the alien planet and found things we really want to do there, but before we’ve been able to make the planet comfortably habitable. Hence our excitement at all the liberating things we can now do but also that background hum of physical, emotional, even existential, unease.

All of which makes the latest public video address from Mark Zuckerberg seem less weird. He dedicated a 30-minute video to Meta’s top research goal, which is an attempt to pass the visual Turing test, by which he means creating a space in Digitalland that is experientially indistinguishable from Physicalland. Once this goal is reached, which is surely just a matter of time, Digitalland will pretty much trump Physicalland. It will have all the comfort of earth while also giving us access to those alien laws. There will be, for example, a full sense of spatial dimensions and the spaceless ability to move anywhere in those dimensions instantly for free. At this point we might be tempted to call time on the war of dimensions; Digitalland wins.

So that’s the first moonshot goal. And the second is no less audacious. If anything, because it will come about in a more creeping and incremental way, the second goal is where I’d put my money if I had to bet on a reason our grandchildren’s lives will be unfathomably different to our own. It is the attempt to better integrate Digitalland into Physicalland, or rather to turn Digitalland into an intermediate layer through which Physicalland is experienced.

This second area of innovation is sometimes referred to as augmented reality, which is a neat turn of phrase if you want to bake in the assumption that it will improve or enhance reality, as technology companies like to do. An alternative phrase would be subverted reality, which feels apt because the goal of this work — not for any nefarious reason but simply because it expands the scope, and therefore profitability, of the business models of Digitalland — is to subvert or flip the order of things, so that Digitalland becomes life’s dominant mode.

Subverted reality is being pursued through two main enabling technologies, the first of which seeks to create a kind of imprint or echo of Physicalland in Digitalland, which is sometimes called a ‘digital twin’. These data replicas — imagine a digital model of your car, or of a subway system, or of your dad — sit in Digitalland but in a way that connects invisibly back to their equivalents in Physicalland; it reminds me of the way a voodoo doll links invisibly to its subject. They are a key component in those puppet strings, in that they make it possible to effect reactions from the twin back in Physicalland.

A digital map is a good example of this technology at work. Its superficial goal is to mirror a physical place, so that it can represent that space in Digitalland. But the deeper intent of a digital map is to change the space in Physicalland using the twin, for example by changing how we move around it, or by privileging some places or businesses within the space over others, so that they grow. Similarly, a digital twin of a person, like the ones that are built of each of us by companies like Google and Facebook, is used to predict, and thereby change, the person’s behaviour.[4] And if we think back to those invidious business models, we remember that the point isn’t just to make puppet strings; the point is to rent them out, so that other companies can use them too.

The second technology that enables subverted reality seeks subtler ways to insert the digital layer into our lives, or to slip the layer between us and physical reality. This accounts for the interest in wearable technologies and thin interfaces like eye-glass / retina displays, and now-familiar interfaces like voice.

The thing to watch out for are instances of subversion: moments when Digitalland becomes life’s dominant mode or dimension. When I say dominant here I don’t mean dominant in the sense that our experience in these moments feels primarily digital (although, depending on our progress toward the visual Turing test, this will also happen more and more). I mean dominant in the sense that, in these moments, Digitalland is the dimension in and through which our choices in Physicalland are made. It is, in this sense, the substrate in which our will is situated. And so, as this work progresses, Physicalland will slip into a form of submission, becoming life’s minor or servant mode.

I know, this all sounds a bit dramatic. But remember the frog, whose friends no doubt think she should relax and enjoy the warm water. Because when you stop to think about it, quite a lot of our lives already work in the way I’ve described.

Think of something as mundane as searching Google Maps for a restaurant. It feels normal and totally non-revolutionary, right? But the map is a digital twin — a representation of the physical world that exists in Digitalland. The map is of course not a neutral ‘copy’ of the world developed as a public service. It’s designed to shape our behaviour; that’s how it’s funded and is the reason it’s built. And so while, for now, Physicalland still feels like life’s dominant dimension (after all, most experiences, like having a meal, are still mainly physical), Digitalland is increasingly the domain in which we make the decisions that determine those experiences. Just ask yourself when you’re next out for dinner: why am I sitting at this table, in this restaurant, eating this food, with this person? The answer is probably a version of: ‘because the restaurant shared an appetising picture on Instagram, so I searched on Google Maps to see where it was, and then messaged a friend I’d seen in my Facebook feed, and made a reservation on OpenTable.’ All of which are spaces in Digitalland, the home of the invidious hand.

If that’s where we are now, where are we headed? The clearest glimpse we’ve had of a subverted future was in 2016, when the video game Pokemon Go took off, expanding so fast it surpassed Twitter’s entire American user base in seven days. Pokemon Go situated digital things — animated Pokemon and their paraphernalia — into a digital map of physical places, and people then ran around those places chasing the digital things, causing real physical problems. One man pleaded with Nintendo to remove a digital artefact from his garden after his house was inundated with Pokemon-hunters; restaurants saw sales double after they realised ‘if you can’t beat ’em, join ‘em’, and paid Nintendo to put Pokemon treats in their establishment; and in the US a total of 256 people died — yes, died, in Physicalland, where you don’t get extra lives — after virtual Pokemon distracted them from real life hazards.[5] The whole episode was nothing more than a flash, a subliminal message, gone before it registered in our collective consciousness. But still, it left a quivering mark on the retina, a sign of things to come.

Ok, so let’s pause, take a breath, and turn to implications. Why is any of this relevant to social justice? And, to the extent that it is, how should we respond?

I’m going to look at this from the perspective of two roles we play in a capitalist society: consumer and worker. These roles, and the institutions that form them, mediate our experience of capitalism. We can think of them as ways in which individuals dock into the system. They determine how life in the system feels to individuals, for example whether it is secure, dignified, and fulfilling. And they also determine how our individual behaviours aggregate into social outcomes, deciding, for example, whether society as a whole is stable, dynamic, and fair. And so understanding how these roles are changing is a good way to understand how the work of social justice needs to change.

Let’s start with our lives as consumers. The inescapable and really rather consequential implication of everything above is that markets work differently in Digitalland to the way they work in Physicalland.

I’ve made this point at length before so I won’t relitigate the whole argument here (see my essay The Fable of Bees, if the bees were all on Facebook). For now, I’ll just recap the headline, which is that consumer choice and competition function differently in Digitalland, and this undermines our old approach to economic regulation, which was a pillar of the social democratic settlement.

Consider first how choice works in Digitalland. If we reflect on how it feels to be a consumer in the early 2020s, I think we can tell that something’s off. Notice the vibe of compulsion that now permeates our lives as consumers, and the never ending struggle to avoid binges and impulse purchases. I know I’m not the only person who’s run out of tactics to stop myself doom-scrolling. I give you, for example, the phone safe, a product designed not to keep your phone safe but to keep you safe from your phone. And think of the way we now have to be constantly on guard against the latest behavioural wheeze. Think, for example, of all those times you’ve spotted an unintended subscription in your bank account, or found an auto-renewed bill that’s higher than the rate you signed up to. And if you’re anything like me, this prompts a trawl for other accumulated subscriptions, which you clean out from your bank account as if you’re clearing barnacles from a boat. A lot of this stuff flows from the way companies are now unprecedentedly sophisticated and determined in the art of behaviour-shaping, and seem to find behavioural manipulation, rather than product optimisation, the most profitable way to go.

If we picture ourselves making a decision in a space in Digitalland, we see why things feel this way. Imagine I’m trying to buy something in the infinite bazaar, or looking at a digital map, deciding where to eat out. And let’s recall a few pertinent facts about spaces in Digitalland: the information that I do and don’t receive is decided by the company that runs the space; the choices with which I’m presented are framed by the same company; the company is omniscient within the space, knowing more about me than my closest friends do; the company is also omnipotent, meaning they can change my mental context at will; and one of the main things the company is trying to do — and sometimes the main thing — is to shape my behaviour, often to the extent that they’re obsessively optimising their ability to shape my behaviour so that they can rent this ability out.

So what can we say about the choices we make in Digitalland? Back in Physicalland, one thing we felt we could say was that our choices ‘revealed our preferences’. For example, if I bought a hotdog, this revealed that I wanted a hotdog. This process of revealing preferences sat at the heart of the Adam Smith worldview; it was the mechanism that made markets uniquely well-placed to solve information and coordination problems, because the aggregation of revealed preferences guides the market about what we collectively want. In a sense, it is the invisible hand.

Do the choices I make in Digitalland reveal my preferences? It seems to me the answer is complicated. It’s not quite no, because clearly we still have free will, even in Digitalland. But it’s not quite yes, either, is it? There are, after all, companies that are worth hundreds of billions of dollars, whose main business model is to rent out puppet strings. And there is a general frenzy of enthusiasm as firms across the economy try to develop and hone these capabilities.

All of which raises some interesting questions. If our choices don’t reflect our preferences, or at least don’t reflect them as well as they used to, then what do they reflect? Do the choices we make still conjure up the magic of markets? Or do they conjure up something more sinister?

Let’s turn to competition. There’s a mature academic literature on the competitiveness of markets, so here we can draw more directly on economic data. A wide range of studies show rising market concentration, slowing rates of start-up activity, rising inter-firm gaps in productivity, and the emergence of what economists call the winner-takes-all economy.

If you think about a platform as a space in which things can be done, it makes sense that competition between platforms works differently to competition between physical products and services. If you want to find someone who has something you want, and you want to go into Digitalland to do it, why would you enter a space without many people in it? Why wouldn’t you swell the already-bustling crowd in the infinite bazaar? Or imagine if you wanted to start your own space in Digitalland. How would you persuade people to join your space when the owners of other spaces force them to leave their valuable (digital) possessions behind when they leave? And think about the information asymmetries that exist between big established spaces and small new spaces. If you want to persuade people to join your new space, you’re stuck peering through the glass at your potential customers in another space, trying to figure them out. Meanwhile, the owner of that space knows every last thing about them. This explains those winner-takes-all dynamics; big spaces get bigger, small spaces struggle to scale.

Competition, like choice, is of course integral to the idea of a market; in a sense, it is the market, or at least is a source of its power. And because competition and choice are fundamental, the character of markets — and, therefore the character of a capitalist society — is highly sensitive to changes in the logic by which these dynamics function. When firms notice, for example, that behavioural tricks are now the best route to profit, they divert their energy into manipulation, turning innovation into its evil twin, so that it drives not better products but subtler ways to trick us. And when people notice that competition is now a winner-takes-all game, entrepreneurs don’t seek sustainable ways to scale their businesses; they seek an exit to an established platform. Meanwhile, investors sink distorting sums of money into unprofitable firms in the hope of being first to the number one spot, while leading firms work hard to amplify those anti-competitive forces — they try, as Google puts it, to deepen the moat.

So that’s our role as consumers. What about the other role we play in capitalism, as workers? What is it like to work in Digitalland? The simplest answer to that question is: it depends on whether or not you have access to the underlying text of Digitalland.

What I mean is that workers in Digitalland fall into two classes (and I use that word advisedly): people who have access to the means of control and people who don’t. As a worker, you’re either part of the effort to design, build, and run a space in Digitalland, or you work in a space and are subject to the way it’s built.

It seems to me that this distinction so heavily determines a worker’s economic power in Digitalland, and their sense of agency and status, which together form so much of our identity and position in society, that the word ‘class’ is the right one to use.

Let’s see if we can characterise, albeit crudely, the two classes of worker in Digitalland.

First, imagine that you work in a space in Digitalland so that you’re subject to, and not in control of, the way that the space has been built. An example would be a person who works as a courier for a platform-based delivery company. Another example would be a warehouse operative whose work is tracked, modelled, and optimised through a workforce management platform.

For work like this, I think it’s fair to say that Digitalland is the dominant dimension. As per my definition above, I don’t mean dominant in the sense that these jobs are literally digital; they’re not pouring drinks as an avatar at a bar in Minecraft. What I mean is that all the defining decisions are made in and through the digital dimension: it’s where pay is determined, tasks are allocated and performance is managed, and it’s where a worker’s visual identity and reputation with customers is decided. So people doing these jobs are the first generation in history to work on the other side of the portal; they work in Digitalland.

How does work in Digitalland differ to work in Physicalland? For one thing, the companies that own the spaces within which people work can see almost everything they’re doing and saying. They can also change almost any aspect of the work environment instantly and iteratively, without notice and with few constraints, from the amount they charge people for the right to work in the space to the way people’s work is perceived and surfaced to customers.

How does this feel to workers? We have a good sense thanks to ethnographic and qualitative research. On the one hand, people say it’s stressful to know that their every action is being tracked and optimised. In one study with Amazon warehouse workers, for example, a person described “the mental stress of being held to the productivity standards of a robot“, which made them feel “a constant buzz of low-grade panic”. On the other hand, many people find working in Digitalland paradoxically liberating because, although they are watched, they often have more freedom than they did in Physicalland, particularly when compared to low-paid Physicalland jobs. For example, people can now often decide when and where they work.[6]

All of which raises a pertinent question: how would we characterise the relationship between a person working in a space and the company that runs the space?

What we can say for sure is that it’s not the same relationship as the one between a worker and the company they work for in Physicalland.

In Physcialland, working relationships are codified in the legal institutions of employment and self-employment and their attendant social norms, the formative development of which took place in the late nineteenth and early twentieth centuries. But working in a space in Digitalland is different to this; it feels more like you’ve gone into a space that is engineered for work, in exchange for which you give the space owner a cut of your takings. So it feels to me intuitively more accurate to say you work in the space than that you work for the space. Certainly, it would be a weird coincidence if the old categories of employment or self-employment were the right way to conceptualise and govern this entirely novel relationship.

What about the second class of people, the elite of Digitalland, who aren’t just subject to the alien laws but who also put those laws to use? I’m thinking of jobs like software engineer, data scientist, product owner, or UX designer. What is this work like?

When I imagine these elite roles I find it hard not to think of the control room in the film The Truman Show, from where the Director and their staff watch Truman and shape his environment and behaviour.

So how does it feel to work in the Director’s room? Let’s not be coy about it; surely it’s better than being out there, on the receiving end of things. In fact, I would imagine that at times these elite roles give people an empowering, even thrilling, sense of agency.

I’m sure it’s not all fun, though, running spaces in Digitalland. For one thing, even if you’re in the elite, so much of your work happens in those rudimentary spaces, with their two-dimensionality and lack of tactile experience. It’s a bit like being in the first generation of people sent to work a stint in the alien space base; it must be exciting, and I bet it would be really well paid, but the environment is inhumane. So for all their hefty pay cheques, it’s no surprise that surveys show even elite workers in Digitalland often feel burned out.[7] And is it a surprise, either, that among elite tech workers we’re seeing a rising sense of moral angst? It’s as if workers at the big tech platforms are starting to look at the people, out there in the spaces they’re optimising, subject to their efforts to track and manipulate, and feeling increasingly icky about the whole damn thing.

Ok, so this is a reductive and impressionistic depiction. But I do think it helps us make sense of how work is starting to feel. The return of eye-watering inequalities, not just of money but of status and agency; the mix of freedom and insecurity that characterises jobs at the digital frontier; the way that top jobs in the digital economy feel, to most people, so out of reach; and the red eyes and frayed nerves of the burnout generation.

All of which brings to mind our historical experience of what happens when work is reconfigured by technology. In 1845, as the Industrial Revolution gathered pace, Disraeli’s novel, Sybil, described a society that was splitting into “two nations” between which people’s lives were so different that they felt “as if they were dwellers in different zones, or inhabitants of different planets”. It’s hard to think of a better description of where we’re headed if we keep misgoverning work in a digital age.

The more I write about technology and social change, the more I empathise with the frog. Her plight just nails so well the gradualism paradox: changes that play out slowly are more likely to take us by surprise.

This is one of those patterns, like the golden ratio or fractals, that appears again and again, in many different domains of life. A marriage sours quietly and then lights up in argument; our bodies thicken into middle age until we catch a shocked glance in the mirror; mud acretes on a hillside and then, liquified by rain, washes a town away.

My sense is that institutions exhibit this behaviour too, but in mirror image. It’s the environment around our institutions that changes slowly while they stay structurally the same. Then, one day, we rub our eyes and realise we’re in new surroundings. We end up feeling like the occupants of a car, stuck with its wheels spinning, in a swamp. We can rev the engine all we like, and fight over the steering wheel, but we won’t get anywhere until we change the vehicle’s design.

This essay is part of a project with the Joseph Rowntree Foundation, Social Justice in a Digital Age, designed to help us avoid the plight of the frog. We’ll be working with a range of organisations, from the Institute for Government to the British Academy, to create space for discussion. The goal is to understand the ways in which capitalism is changing, so that we can better reconfigure and reimagine our institutional settlement.

In this essay, I’ve kicked things off by looking at one such change, the transition to a society that runs on digital platforms. To avoid the horseless carriage trap in which big changes hide under the cover of old words, I’ve used metaphor to see what kind of thing a platform is and to understand the implications platforms have. I’ve then reflected on what this means for two roles we play in a capitalist society, as consumers and workers.

What should we do in light of all this? I’m afraid this isn’t the kind of essay that ends with a policy manifesto. In fact, one implication of my argument is that some, though not all, of the social problems we face today sit at a deep level of abstraction; they can’t be solved with policy tweaks. To borrow a phrase from the philosopher Roberto Unger, the problem goes ‘all the way down’.

It’s this depth of problem that makes our present predicament analogous to the one we faced in previous technological revolutions. As I write in End State:

‘Society didn’t survive the Industrial Revolution by tweaking mediaeval policy instruments. We didn’t expand the guilds, freeze the charges for turnpikes, and increase funding for the knights of the garter. We replaced the old system with a new one.’

Fast forward fifty years from today and my bet — or maybe my hope — is that we’ll be able to say the same thing about the digital revolution. We won’t get through this thing by reclassifying some workers as employees, or by putting benefits up or taxes down, or by using industrial-era laws to break up technology monopolies. The work goes deeper than this. We need new institutions, practices, and mindsets for a digital age.

So rather than pretending to know the answer, let me wrap up by suggesting how I think we should approach this work, and by sharing some reflections and questions to explore.

I see Digitalland as a new frontier for the work of social justice; it’s territory that is as yet ungoverned, or rather that is mis-governed, in that it’s governed with institutions, practices, and mindsets that were designed for Physicalland.

As a result of this mismatch of governance, what we see at the frontier are novel injustices, many of which we haven’t encountered before and are only just beginning to understand. But what we also see are new opportunities — opportunities that are so significant that a part of our problem (and I think a surprisingly big part) is that we struggle to believe, or to visualise, how good things could be if we put Digitalland’s full potential to use. So we’re experiencing a failure of imagination as much as anything else.

Take consumer markets as an example. Despite admirable work from forward-thinking regulators like the UK’s Financial Conduct Authority and Competition and Markets Authority, deep down we’re still stuck in the old mental model of Smith’s benevolent baker. It’s a worldview in which choice and competition are seen to all-but guarantee good outcomes in markets, so that as long as we keep an eye on bad behaviours and address externalities like pollution, all will be well. It’s this underlying worldview that explains why we cast the state in a role like the referee at a football match: the government is there to watch the game, check for infractions against a list of banned behaviours, and keep its whistle at the ready.

As markets move into Digitalland, there are malign ways to win the game within the old rules. Bad outcomes aren’t just down to bad apples; they’re systemic. An industry can easily, for example, fall into a cycle in which firms compete to design the stickiest contract, or the most doom-scrolly content, so that innovation means not making better products but making our lives worse. This dynamic isn’t unique to a platform economy but it is amplified and made more common by the alien laws of Digitalland. In the old world of the baker, crunchy bread was normally the best route to profit. Now, if you’re a digital platform, your best bet is often to refine your subscription trap; to shakedown your customers for data to refine those puppet strings; or to engineer a more subtly misleading checkout experience.

The response, crudely speaking, is for the state to care more about outcomes. We need to stop assuming that a competitive market will serve consumers well, and bring a more open mind to the question of whether or not a market is delivering the goods. This idea might chill the blood of economic liberals but it shouldn’t; it doesn’t mean the state delivering outcomes, it needn’t mean the state intervening more, and it probably means having fewer, not more, prescriptive rules.

A well-functioning market, in which competition drives healthy innovation, is still a uniquely effective institution. But the state must take the health of that institution more seriously. This could mean, for example, tracking important outcomes, such as price differentials between groups of consumers, or the stickiness of consumer behaviours, or consumer sentiments like trust and regret, to gauge the health of the market and, where necessary, making corrective interventions.[8] In general, it means relying less on a checklist of prescriptive rules and more on outcomes and principles, like treating customers fairly. In Digitalland, firms are just too quick, and the dark art of those alien laws is too opaque, to be regulated by a bureaucrat checking off boxes on a clipboard.

If anything, the shift to Digitalland has even starker implications for competition policy, and we need to upgrade our mental model here too. The old approach to competition policy was similar in spirit to the one we used for consumer policy: to monitor and intervene by exception. In competition policy, though, the state behaved less like a referee and more like a teacher scanning a playground for bullies. This owes a lot to the origins of competition policy in the 1800s, when we developed a new policy settlement in response to industrial-era monopolies. The whole approach was built around the idea of bigness — the regulator gets involved when one big company tries to buy another — and the main exceptions to competition were reserved for the natural monopolies of physical infrastructure like railways, electricity networks, and water systems.

In Digitalland, things work differently. Market power isn’t just about big kids pushing little kids around; it’s more systematic than this, as if the layout of the playground itself encourages bullies. We still need an anti-bullying policy, for sure. But twenty-first century competition authorities need to care more about playground design. They also need to be alive to the risk that the bullies will try to re-landscape the playground to their advantage. And if we’re going to resist that re-landscaping, we’ll also need to rediscover the political courage that characterised historic fights against monopoly.

This all means working harder to make digital markets less prone to concentration. So how do we do that? An example is using interoperability policy to force big technology platforms to adopt common standards, so that people can move between spaces in Digitalland. This would allow people to take their data with them when they leave a space and it would let people message friends in another platform without having to join it. The ultimate goal is to make it harder to wall off Digitalland into private estates and to diminish those winner-takes-all effects.

All of which opens up bigger questions about the way we govern private markets in Digitalland. How should we treat ownership in a world of non-rivalry? Do we need a radically permissive approach to intellectual property? And what about Digitalland’s territory? Should parts of the territory be classed as a digital commons, open to anyone? Should we, for example, treat the data we generate when we live and work in digital spaces as a public good, open for entrepreneurs and researchers to use? Could those rules of interoperability and open data amount to a digital right to roam, so that we can move more freely across Digitalland? We need to answer these questions in a way that is unconstrained by the old paradigms of Physicalland.

What about work? It’s another area where we see fresh injustices and opportunities side by side. So how do we make work fair and secure in a world of platforms without denying ourselves the powers of Digitalland?

What I find most striking about work is the way we’ve struggled to free ourselves from old ideas. Think of how long we’ve spent arguing over whether workers in Digitalland — such as Uber drivers — are employees of the spaces in which they work or are self-employed. It seems to me clear that Uber drivers are neither of these things; why on earth would legal categories that were invented decades before the internet be a good way to conceptualise and govern work enabled by platforms?

To be fair, I can see why we’ve got stuck with old approaches. Confronted by the fire of insecurity in the gig economy, we reached for the nearest policy solution available, throwing our pint on the flames. But if we’re going to stop the fire from returning and spreading, we need a more sustainable solution. We need to stop dining out on the intellectual work of previous generations and develop a settlement of our own; a new legal framework to govern work in the digital age.

What does that mean? As with economic regulation, it means asking some basic questions. For example: how should we define the relationship between a person who works in a space in Digitalland and the owner of that space? On the back of this, we can then debate the responsibilities that flow from this novel relationship, so that we can codify these responsibilities into the hard bones of law and the soft tissue of social norms. To what extent should a space owner be allowed to limit the autonomy of a worker in their space? Should they, for example, be allowed to stop the worker from also working in another space? Should they be able to stop a worker from taking their data with them when they leave? What minimum standards do we expect for the working environment in a space, for example with respect to accessibility and equity?

As well as new laws and norms, we’ll also need new institutions to fill the gaps that are left when the old institutions of Physicalland don’t survive the transition to Digitalland. Think, for example, of the social and economic role that was played historically by the physical workplace. What mix of institutional innovations from civil society, progressive businesses, or even the state, could fill this gap? Or think about how workers can train and progress in a world without line managers or employers, and in which workers like Uber drivers may never even meet their coder colleagues, let alone share an office with them. Who will help workers get ahead in Digitalland? If we think back to the example of worker insecurity in the late 1800s, what happened then was that a new institution, social insurance, emerged, at first as a patchwork of voluntary schemes run by progressive businesses and charities. Only later was social insurance formalised as a new arm of the state. My hunch is that we’ll see a similar process play out now, as new institutions take shape first in the charity sector or in localised experiments before they’re regularised as pillars of a digital-era state.

Finally, we need to answer some broader questions about the civic responsibilities of organisations that run big spaces in Digitalland.

I chose to focus this essay on two implications of the rise of platforms, for consumers and workers, because they both bear directly on social justice and because they’re areas of public policy where the effects of platforms seem particularly advanced.

Really, though, we do more than shop and work in Digitalland. We also read news, debate, and share information; make and lose friends; and form opinions, which in turn shapes how we vote. These civic processes can make or break social justice, which depends on our ability as a society to reach decisions that are deemed legitimate and that arise from a pluralistic and evidence-based testing of ideas. Yet it now seems clear that the business models of Digitalland aren’t compatible with the healthy functioning of these processes.

I won’t get too deeply into the political implications of platforms, which is a hot mess of its own and a topic on which I’m no expert. One quick reflection, though, is that our policy response to these big social questions is currently being developed in quite a haphazard way, and it might be useful if we were more intentional and organised about how we’re having these debates.

As an example, a lot of the best thinking about platforms is happening in economic regulators. This is mainly because consumer markets are on the frontline of these trends, so regulators have little choice but to confront them. Yet the effects of platforms run far beyond consumer markets into issues like freedom of speech, and regulators are not well-placed to address these wider ethical and philosophical questions. It’s problematic, for example, that Ofcom, the laggard among UK regulators when it comes to new ways of understanding markets, is being tasked with regulating much of the internet economy. So my sense is that we’re not setting ourselves up for success.

Here’s another example of how this plays out, and of how we could course correct. In 2022, the CMA proposed that the biggest digital platforms should be given a new legal designation, Strategic Market Status. The idea was to label big platforms with this status and then attach new responsibilities to it, beyond those that apply to normal companies. This seems to me a radical and necessary step. Radical because, for the first time, it acknowledges that platforms are different. Necessary because, until we acknowledge this fact, we can’t define the responsibilities that come with being a platform.

The CMA’s work is in the process of being diluted after effective lobbying from big tech companies. Still, it strikes me that when politicians pick this idea back up — as they will eventually have to — we could build on the idea of giving platforms a distinct status. Maybe in time, a new legal status for big platforms could become the basis for a new legal regime, similar to the regime we built around the limited company — a way of giving these institutions the freedom they need to innovate while also being clear about their civic responsibilities, for example with respect to freedom of speech.

The more I think about it, the more I think we face some historically significant choices about the way we govern platforms, and that what we’re missing is a policy vehicle that is capacious enough for these discussions. Maybe what we need is something like a series of landmark Platform Acts, akin to the nineteenth century Factory Acts; a way to surface the big choices we face and facilitate a public argument over the moral duties of platforms with respect to workers, consumers, and citizens, and a means to codify these duties in law.

These are just some initial thoughts; some seeds of ideas that may or may not take root. The whole point of these essays is to open up conversations, not close them down, so I’d love to hear responses and critiques.

As part of the project, we’re going to host forums to discuss these issues. To stay in touch, follow me or JRF on Twitter. And if you’re more of a writer than a talker, here’s an open version of this essay where you can add comments and read contributions from others.

And, as always, you can follow my writing on Medium and on Substack, and buy my book, End State.

Footnotes

1. Credit to W. Brian Arthur, whose wonderfully pithy book The Nature of Technology led me to the way I describe digital technology in this essay. Also to Ulrich Dolata and Jan-Felix Schrape, whose paper Platform Architectures, emphasises the distinction between platforms and the companies that run them. And to Shoshana Zuboff whose book The Age of Surveillance Capitalism prompted many thoughts and avenues for further reading.
2. One big difference is that in a theme park your rides have a maximum capacity, which in Digitalland, thanks to non-rivalry, isn’t the case. Or at least, the capacity of a digital space scales much more easily than a space in Physicalland.
3. With thanks to Toby Nangle for these estimates, which as Toby would emphasise come with very high levels of uncertainty.
4. I’m using the phrase ‘digital twin’ here in quite a broad sense, to refer to the way data can be used to model, make predictions about, and thereby influence, a physical system.
5. All the ‘ins’ in these sentences should really have air quotes. What does it mean to say the treats were put ‘in’ the restaurants? This speaks to one of my favourite challenges for future public policy: what rights do you have over a digital representation of something you own in Physicalland? If someone builds a digital map, and puts a virtual prize in your garden on the map, and you get inundated with people trying to reach the prize, what right do you have to tell the company to stop? And what happens when the value of physical places, like your home or restaurant, is pushed up or down by these decisions?
6. This strange juxtaposition of no freedom and lots of freedom was summed up well in the title of a BBC investigation: ‘Uber’s paradox: Gig work app traps and frees its drivers’. Also to note, for another day: another world-changing fact about Digitalland, which we still seem to be in denial about, is that it obviously has no geographical borders. So for work that doesn’t have to be done in-person, immigration policy loses all purchase. Which is of course another source of discontent among workers in the developed world, as wages find a new level in a single global pool for labour.
7. For example, one report suggests that as many as 80% of workers at big tech companies like Google and Facebook have symptoms of burnout.
8. What kinds of intervention can regulators make? A nice example, which I explore in End State, is the UK Government’s ban on letting agent fees. At first, the Government tried to keep fees low by using an orthodox approach: they gave renters more information, so that they could shop around. This failed because renters don’t shop around for letting agents, they shop around for homes and then get lumped with the letting agent the landlord has already chosen. Eventually the Government clocked this and made a market-correcting intervention; they banned fees for tenants, which shifted fees onto landlords, and landlords do shop around for letting agents, so the fees now face pressure from competition. The intervention made the market more competitive, not less.