THE NEW OIL?
In April 2018, Mark Zuckerberg was summoned to Washington, DC to appear before the US Congress and answer questions about the Cambridge Analytica scandal. At home in London, I sat up late into the night on two consecutive evenings and watched the livestream of the hearings. It was sober viewing, though there was one enjoyable moment of comedy when Orrin Hatch, an octogenarian senator from Utah, asked Zuckerberg, ‘How do you sustain a business model in which users don’t pay for your service?’ The Facebook founder, wearing a dark suit and blue tie, had answered questions respectfully, even when they were motivated by party political point-scoring or displayed a weak understanding of technology, but this level of disconnection from the modern world was too much for him. After a pause in which he struggled to suppress his amusement, Zuckerberg drily replied, ‘Senator, we run ads.’
Advertising-based business models have been around for hundreds of years and form the commercial basis of many newspapers, magazines, radio stations and TV channels. They are an accepted part of life and are not generally regarded as controversial. News, opinion and entertainment are provided for free, financed by the sale of advertising space to companies that wish to promote their products to readers, listeners and viewers. The value of this advertising space depends on how many potential customers it is expected to reach, and how inclined those customers are to spend money on the advertisers’ products. The business model of Facebook – and of other owners of digital advertising space, such as Google and Twitter – is basically the same: spaces in the user interfaces of websites and apps are reserved for ads and auctioned to advertisers, which generates revenue; services like social networking, messaging, email and search can therefore be provided to users for free. Where this business model differs from that of traditional media companies is in its use of data that enables advertisers to target specific audiences. The claim of surveillance capitalism theory that the business model of companies like Facebook and Google is ‘illegitimate’ therefore rests on claims regarding their use of data.
At the beginning of this book’s introduction, I mentioned Ray’s German shepherds. He hadn’t told me about them, but I’d noticed his ‘I heart GSDs’ key ring. If he’d been wearing a claret and sky blue scarf, I might have asked him about West Ham United instead. Until I asked, I didn’t know for certain that he had German shepherds. There could have been other plausible explanations for the key ring – it might have come into his possession by chance, or he might have borrowed the van from a friend. But I made an educated guess that turned out to be correct, and we had a warm, meaningful conversation about his dogs as a result.
Statisticians call this a probabilistic inference. Ray made one about me too when, without consciously thinking about it, he put together the time of year, my destination in central Cambridge and the high proportion of books in the small load, and calculated that I was probably a student. Probabilistic inferences like these are what is happening when you are targeted with Facebook ads; advertisers define which audience they would like to see their ad (for example, women who like yoga, German shepherd enthusiasts or West Ham fans), and Facebook uses data to make educated guesses about who those people might be.
Neither Facebook nor its advertisers really know these things about you, and they don’t need to. If a sports retailer shows you an ad for West Ham merchandise and you’re not a West Ham fan, it doesn’t really matter – you scroll past the ad in half a second and then it’s gone. When Facebook guesses correctly it can seem uncanny, which is why some people are convinced that Facebook must be listening to them through the microphone on their laptop or mobile phone. We might say, ‘I said to my friend this morning that I was thinking about getting the new West Ham shirt, and now there’s an advert for it on my Instagram feed!’ A more likely explanation would be that the conversation and the ad have the same underlying cause: West Ham releasing their kit for the coming season. Retailers will want to sell the kit, and fans – many of whom will have liked the team’s Facebook page – will be talking about it. Most of the time, though, you barely notice that ad targeting is happening.
One of the key claims of surveillance capitalism theory is that this kind of targeting is intolerable because it’s an invasion of privacy. But privacy is subjective – what we consider private varies from society to society and from person to person. Did I infringe Ray’s privacy by asking him about German shepherds after spotting his key ring? I don’t think I did, and I don’t think Ray thought I did, either. The keys were in the ignition, so the key ring was in plain view of me or any other client in the passenger seat of Ray’s van. It was in the public domain. Perhaps this might have been different if we’d been in a country where people have different expectations about interactions with strangers – in Sweden, for example. My Swedish alter ego might have noticed the vallhund key ring but decided it would be impolite to comment on it. It might also have been different if Ray preferred to keep his work separate from his private life. If that had been the case, he probably would have chosen a neutral key ring and kept Charlie’s muzzle out of sight, instead of hanging it from the rear-view mirror.
Ray’s choice of key ring is an analogue version of what we are calling profile data – the kind of data that’s expressed by your Facebook profile. Similar examples of actual profile data would be liking the page of the German Shepherd Dog Community on Facebook (which has nearly 2.5 million followers), or joining the German Shepherd Owners group (which has over 100,000 members). Pages you like and groups you join appear on your profile and are visible to other Facebook users – including advertisers. They are in the public domain unless you choose to hide them, and probabilistic inferences about your interests can be made from them.
Is it an invasion of your privacy for Facebook to show you an ad based on inferences from your profile data? I want to suggest that this is a matter of personal preference and not – as surveillance capitalism theory argues – a matter of ethics. To bring this to life, let me tell you a story about the early days of Bought By Many, the tech company where I worked between 2012 and 2018.
Why Facebook Ads?
In the initial start-up phase, before they know what their product is or who their customers are, entrepreneurs will generally do whatever they can to save money. One obvious way to do this is to economise on office space. My two co-founders and I agreed on a mutually convenient location for our first office – Farringdon in central London – and started viewing the cheapest places we could find online. Eventually we settled on a single whitewashed room in a serviced office building. The compromise was square footage: it was officially a two-person room, but we decided that with small desks we could all fit in. There was just about enough space for Bought By Many’s only physical assets: a basic printer and a coffee machine. When we had a visitor, we would borrow a chair from another office and install them behind the closed door. The first person to visit us was our seed investor, John. He had put £300,000 into Bought By Many, so obviously we wanted to be hospitable and make a good impression. We made him a latte and placed the borrowed chair behind the door. My co-founder Steven launched into an update about what we’d achieved in our first couple of weeks as John sat down. As he put his weight on the chair, the back right chair leg went straight through the floorboard and into the void below. John was tipped suddenly and violently backward. Luckily, he managed to keep hold of his coffee cup: we were mortified enough as it was, without having to administer first aid and arrange dry cleaning. When we made our first hires and moved into a five-person room a few months later, our tiny office with the hole in the floor was turned into a stationery cupboard.
Another way we conserved funds was by paying ourselves as little as possible. Steven retained a part-time interim CEO job, which meant he only needed a minimal salary from Bought By Many but also that he couldn’t be in the office all the time. Guy and I managed on 50 per cent of what we were used to by reducing our living costs. Some of the time, rather than paying the train fare to come into London from Maidenhead, Guy would work from the shed at the bottom of his garden, which meant there were often days when I’d be in the office on my own. I prefer being in physical proximity with people I’m working with, so those days were quite tough. The office had a high ceiling and a big Crittall steel window: these gave it character, but also meant it was very cold in the winter. Much as I would do six years later in draughty corners of Cambridge libraries, left by myself I would sink into melancholy doubts about the choices I had made. It was during those times that I learned what makes Facebook so valuable.
At that time, Bought By Many had a hypothesis that if we got people who shared the same need for insurance together in an online group, we could use the group’s collective buying power to negotiate a better deal with insurers. While Guy wrote code for the website and Steven negotiated agreements with insurers, my focus was on coming up with ideas for groups and then testing them out with real people. The first idea was to create an insurance policy that would pay out a million pounds if a young person suffered a life-altering injury while playing amateur rugby. In search of people who might be interested, I started following school rugby results obsessively so I could work out who the top teams were. On Monday mornings I would individually email the coaches at ten or fifteen schools to congratulate them on Saturday’s result or commiserate with them. Then I would tell them about Bought By Many’s insurance. My efforts were met with indifference. Eventually one of the coaches kindly explained that there were two major problems with our proposition. Firstly, the schools were reluctant to draw parents’ attention to specialist insurance products, as it would imply that rugby wasn’t safe and that their existing insurance arrangements were inadequate. Secondly, there was an expectation that the players’ union, the RFU and the wider ‘rugby family’ would take care of anyone who was seriously hurt playing the game. That was what had happened when the England under-21 prop Matt Hampson was paralysed after a scrum collapse. I went to interview him at his home in the Rutland countryside and he spoke movingly about the financial, practical and emotional support he had been given after his accident. We had managed to pick a problem that people were hesitant to talk about openly and where there were already solutions that – while not immediately obvious to outsiders – were more than good enough.
Our second idea was equally flawed. A home insurer wanted to acquire new customers in Yeovil in Somerset and agreed to offer a group discount if we signed up one hundred customers. I tried various tactics to get the town’s residents excited about clubbing together with their neighbours to get cheaper home insurance. I ran a disastrous Google Ads campaign, in which I paid £26 for a single click to our website. Then I persuaded Steven and Guy that we should decamp to Yeovil for two days, with the vague idea that we’d be able to meet ‘community influencers’. We wandered the streets talking to people at random, but it was hopeless – nobody was interested. The only useful thing we learned was that a lot of people found our messages about Bought By Many’s ‘new’, ‘different’ and ‘innovative’ way of getting insurance alienating, and that they valued the established ways of doing things that they understood and could trust.
It was while I was smarting from these lessons one November afternoon in the little office, that I turned to Facebook. At a networking event for start-ups, Chris, an entrepreneur who had founded a social network for PAs, had told me how he had acquired his first 500 users. He had logged on to LinkedIn, joined every group he could find that was aimed at PAs and started answering people’s questions. Which hotel in Kowloon had the best gym? And was it feasible to have a meeting with a supplier on a Eurostar train? After a couple of weeks, he felt he had enough credibility to start posting links to articles on his website that were relevant to the discussion. Other members of the group followed them and found them useful enough that they were willing to sign up for Chris’s network. I wondered if I could do something similar with my new idea for Bought By Many: travel insurance for people with diabetes.
There aren’t many diabetes-related groups on LinkedIn, but there are a lot on Facebook. With some help from two data scientists called Steve Johnston and Liam McGee, Guy and I had built a huge database of anonymous internet searches for insurance, which also showed how many pages of content there were on the internet about each search topic. Analysing this data revealed that thousands of people were searching for travel insurance that covered the complications of diabetes, but almost no insurance companies included information about these conditions on their websites.
I felt like I had a valuable insight into where the insurance industry was failing to serve the needs of people with diabetes. Could I join the Facebook groups for people with diabetes, get involved in the conversations and pick my moment to mention Bought By Many’s travel insurance, in order to repeat Chris’s success? Having browsed the groups, I decided that I couldn’t. Firstly, unlike Chris, I didn’t have anything useful to contribute apart from one highly specific insight. Secondly, most of the groups didn’t allow participation by commercial organisations; even where this wasn’t explicitly prohibited, it was clear that it wouldn’t be appropriate for me to join in. These were conversations between people whose lives were directly affected by their condition: advice about glucose monitors was being shared, stories about awkward insulin injections told and consolations about pain offered. There is a lot of scepticism in academia about using the language of ‘community’ to describe online groups, but that was exactly what I was observing. I suppose I could have pretended, but it would have been disrespectful – a violation, even. I needed to think again.
This was what brought me to Facebook ads. I had some understanding of them, shaped by working on the integration of Techlightenment into Experian, the data company I was working for at the time. Techlightenment’s software made it easy for retailers to generate thousands of ad variations for their large product catalogues. As that wasn’t relevant for Bought By Many, I’d deprioritised testing Facebook ads. But now, having seen how lively the conversation about diabetes on Facebook was, they seemed to be the obvious answer. I loaded up Audience Insights, Facebook’s tool that shows you roughly how many people you can reach based on the criteria you set for your marketing campaign. I set the geographic filter to the UK, typed ‘diabetes’ into the ‘interests’ filter and hit the return key.1.5 million people. Wow! I typed ‘diabetic retinopathy’ and hit return again. 216,000 people. ‘Diabetic foot’ gave me 66,000 people. It was incredible – analysing the anonymous internet search data had given me an important insight about an unmet insurance need, and now I had found a way of telling the people who would benefit from knowing about it without violating the norms of their online community. Clearly this was better than emailing school rugby coaches and wandering around Yeovil. Within an hour I had set up some simple Facebook ad campaigns for our Travel Insurance for People with Diabetes group. Within a week, Steven had identified an insurance broker who could cover diabetic complications and negotiated a 12.5 per cent discount for Bought By Many members. Within a fortnight, we had over a thousand members – we were in business.
The Rights and Wrongs of Micro-targeting
The kind of targeting I was doing when I set up those initial Bought By Many campaigns is known as micro-targeting, which simply means defining an audience with a high degree of specificity – for example, people aged over eighteen in the UK who appear to have an interest in diabetic retinopathy, based on their Facebook data. Since the Cambridge Analytica scandal, micro-targeting has become a byword for everything that’s regarded as being wrong with data-driven advertising. So you might expect that the people who saw my micro-targeted ads for diabetes travel insurance would have been appalled. But in fact, in the vast majority of cases, they really liked them. Why was that? I think partly because they recognised a challenge people face as a result of having complications of diabetes – access to affordable travel insurance – and offered practical help with it. Even more than that, people valued the creation of a space for the discussion of that specific challenge. The comment threads on the ads became a place where the pros and cons of travel insurance were debated, and experiences of different insurers and different countries’ healthcare systems shared. Just as Ray wasn’t offended by me asking him about German shepherds, most people who responded to the ads didn’t mind Bought By Many speaking to them about diabetes; in fact, they actually welcomed it. In both cases, something new and of mutual value was created as a result of starting a conversation based on an inference from data.
We can break the minority of people who didn’t like the ads into two groups. The first, less than 1 per cent of everyone who saw the ads, consisted of people who loathe all advertising so much that they are willing to devote time and energy to making their displeasure known. This involved using the comment thread to post insulting messages and gifs. For these people, it seemed like privacy was not the issue; they were affronted by advertising per se, and regarded targeted ads using their profile data and undifferentiated spam as the same thing. The second group, larger than the ad rejectors but many times smaller than the group that valued the ads, consisted of people who had concerns about being targeted. They expressed this in private messages to the company Facebook page, or by posting comments like ‘How does Facebook know I’ve got diabetes?’ To understand why targeting can be an emotive issue, it’s worth considering their perspective in some detail.
Bought By Many now employs over two hundred people, including teams dedicated to replying to questions by Facebook Messenger, email and phone. Back then, however, it was just me – I was a one-man marketing, product, operations and customer services organisation. There were significant downsides to this: for two years until we could afford to hire Heidi, our first social media manager, my weekends were spent communicating with strangers in the comment threads of Bought By Many Facebook ads. I would answer questions while eating breakfast on Saturday mornings, and from the sofa during Antiques Roadshow on Sunday evenings. If I went out for the day, I would answer questions from bus stops, restaurants, shops and museums. It wasn’t great for my quality of life, but there was a huge upside: I came to understand how the people I was advertising to thought and felt. I gained so much insight from it that for the next three years I made everyone in the company take turns answering questions on our Facebook comment threads when Heidi was on holiday. A lot of these insights were about insurance, but I also learned a lot about people’s attitudes to and understanding of data.
I had some very interesting exchanges with the minority of people who objected to being targeted. They typically didn’t dislike targeting that was based on things discernible from their profile – it was targeting based on behavioural data that they thought was inappropriate. You’ll recall that Ray’s German shepherd key ring was an analogue version of his profile data. By contrast, the things that enabled Ray to conclude that I was probably a student were analogue versions of my behavioural data – the kind of data that is generated as you live your digital life on Facebook apps and across the wider web. Examples of my actual behavioural data might be liking multiple posts by students at Jesus College, Cambridge or spending hours on the University of Cambridge admissions website at the time of year when applications for graduate programmes are made. It is one thing to be shown a Facebook ad for diabetes travel insurance when you’re openly following the Facebook pages of the major diabetes charities; it’s quite another if you’re targeted with it having avoided putting anything about diabetes on your profile. Similarly, I didn’t feel Ray had infringed my privacy when he asked what I was going to study, but I might have felt differently if he’d been driving me to an address near Addenbrooke’s Hospital and asked what I was going in for.
What I’m suggesting here is that data-driven advertising isn’t inherently privacy-infringing, as surveillance capitalism theory argues – people sometimes like targeted ads and get genuine value from the conversations that can develop around them in social media. Even when they’re indifferent to ads, most people aren’t troubled by those that speak to interests they’ve publicly disclosed in their social media profiles; where ads do feel invasive, it’s typically because something that you regard as private has been inferred from your behavioural data and directly addressed. There are two ways to mitigate the risk of that happening. One is through law – we’ll discuss that a bit later. The other is by taking control of how your data is used. And contrary to what you might have heard from proponents of surveillance capitalism theory, it’s not at all hard to do.
You Are Not the Product
‘If you’re not paying, you are the product’ is a hugely popular claim among Facebook’s critics. It’s one of John Naughton’s 95 Theses about Technology, Niall Ferguson made the claim in The Square and the Tower, his book about the power of networks, and Zeynep Tufekci made it in a TED Talk. It’s an idea that seems to resonate because it draws attention to the fact that Facebook’s paying customers are advertisers rather than its users, as well as highlighting the role that users’ data plays in creating value for advertisers. It’s become a commonplace in academic scholarship and political debate, as well as in tech journalism, but unfortunately it’s unfounded.
As a statement to be taken literally, ‘you are the product’ doesn’t accurately describe what is sold by Facebook and bought by advertisers; Facebook’s ‘product’ is the advertising space in its apps. Understood metaphorically as a claim about the relationship between the company and its users, it suggests that Facebook treats you as if you were its product – in a disempowering way that doesn’t respect your rights or your wellbeing. A typical example is the assertion by the UK Department of Digital, Culture, Media and Sport Select Committee in their inquiry into fake news that Facebook ‘make it extremely difficult, in practice, for users to protect their data’ through user interface design and ‘complicated and lengthy terms and conditions’. If this were true, there would be a case for saying that Facebook mistreats you, but it’s all wrong. In contrast to the vast majority of legal terms, Facebook’s data use policy is designed for readability. Its ‘Ad Preferences Center’ gives a user transparency and control over if, when and how their data may be used to target advertising. Users can opt out of profile data about their relationship status, job title, employer, education and interests being used. They can opt out of their behavioural data being shared across different Facebook-owned apps and third-party websites. They can opt out of lists of customers that companies who already have their contact details – like their bank or mobile phone provider – have uploaded to Facebook. These opt-outs are accessible from the main Facebook settings page and via the ‘Why am I seeing this?’ link that appears in the menu of every ad. They are simple to use, and activating them won’t degrade the social networking features available. If users want a more detailed explanation, Facebook even provides visual guides to explain how its ad targeting and privacy controls work. Short of getting rid of ads altogether, an idea we’ll examine in Chapters Five and Eight, it’s hard to think what more they could do.
What’s more, if these controls are news to you, it seems that you’re in the minority. A Reuters survey in April 2018 found that 69 per cent of Facebook users knew how to change their privacy settings and that 39 per cent of users had recently done so. If you know about the privacy settings and haven’t done anything to change them, it’s worth pausing to reflect on why that is. If the truthful answer is that you can’t be bothered, how egregious can the use of your data in ad targeting really be?
Navigating from an ad to Facebook’s Ad Preferences Center.
If, on the other hand, you have changed your settings, you might have noticed that the quality of your Facebook and Instagram feeds has suffered. Behind the above image of how to navigate to the Ad Preferences Center is an ad for the ‘all-new Chilli Beef Loaded Fries Box’, apparently available at my local BP petrol station. It is taken from my own Facebook feed, but since I don’t eat meat or own a car, it’s hardly an appealing prospect.
Facebook’s explanation for why I was shown this ad indicated that BP wanted to reach people aged over eighteen in Cambridge. That’s a good example of ‘broad targeting’. Where micro-targeted ads are specific, broad-targeted ads are generic. As they need to appeal to a wide audience, they tend towards the lowest common denominator. It’s unlikely that meaningful conversation will develop around them in the way that it did on our ads at Bought By Many that talked about the challenges diabetic retinopathy presents for getting travel insurance. When my friend Jim, who I mentioned in this book’s introduction, opted out of data in his Gmail account being used for targeting advertising, he began to see more ads for payday loans and hookup apps. Jim and I might have had more privacy by opting out of data-driven targeting, but it negatively affected our experience of the internet.
The alternative way to mitigate the privacy risks associated with the use of data in advertising is through law. Recent legislation like the European Union General Data Protection Regulations (GDPR) requires companies to obtain consent before collecting behavioural data from consumers. Most behavioural data relies on cookie technology, and GDPR’s requirement that you consent to cookies is why you now see so many pop-ups asking you to click a button or check a box when you visit a new website. The main drawback with this approach is that it interferes with everyone’s experience of browsing the internet – regardless of your personal preferences about data privacy, if you’re in the EU you must either accept or decline cookies at every website you visit. Ironically, this can be just as irritating as the ubiquitous pop-ups that characterised online ads in the bad old days, when advertisers relied on cheap tricks instead of targeting to try and catch your attention. The point is that when laws like GDPR are implemented, we are no longer able to choose which trade-offs to make between privacy and convenience.
Data is the New Manure
Perhaps I’ve persuaded you that what surveillance capitalism theory has to say about the privacy implications of targeting may be too simplistic. But what about the argument that Facebook has stolen your data and sold it to advertisers for a vast profit? Surely that’s unjust, even if targeted advertising isn’t so bad? Let’s explore it with the help of another proponent of surveillance capitalism theory, the musician and entrepreneur will.i.am, who has captured the logic of this argument rather well:
The ability for people to own and control their data should be considered a central human value. The data itself should be treated like property and people should be fairly compensated for it. As a musician, I benefit from the copyright system that attaches ownership rights to my lyrics and instrumental tracks. Why should the data that I generate be handled any differently? It makes no sense that the information is used as the raw material to produce billions of dollars of income for massive ‘data monarchs’ yet is of no financial value to me [ . . . ] all one gets is a ‘free’ account bursting with advertising, faux news and lame ‘sponsored content’.
For will.i.am, data is the fruit of your labour, in the same way that ‘lyrics and instrumental tracks’ are the fruit of a musician’s creative endeavours. Data is therefore your property, and it follows that when it is used as a ‘raw material’ by ‘data monarchs’ like Facebook or Google, you are being exploited – the services these companies provide in return aren’t fair compensation. In other words, your property is being stolen.
There’s a fundamental assumption underpinning will.i.am’s argument: that your data is economically valuable. That is what people are getting at when they call data ‘the new oil’. The Economist, the magazine in which will.i.am’s op-ed appeared, famously depicted the big tech companies as offshore oil rigs on its cover in 2017, calling data ‘the world’s most valuable resource’.
But here’s the thing: data is nothing like oil. Metaphors matter a lot for how we think about intangible concepts, so it’s worth unpacking a few of the reasons why this particular metaphor misses the mark. Firstly, oil is scarce, plus it’s difficult and expensive to extract. By contrast, data is superabundant, and it’s produced effortlessly and in unimaginably large quantities. Secondly, oil is gone once it has been burned, while data can be used again and again. Thirdly, and most importantly, oil is a commodity. It’s homogenous: one barrel of crude oil is worth the same as any other. Data is the opposite. Most of it, including things that might matter deeply to you, like your gender identity or religion – is economically worthless, as it expresses little about your preferences as a consumer. Some of it – including things that are entirely banal, like the fact that you left four pairs of jeans in your Amazon basket without checking out – can be very valuable indeed, though only to a small number of organisations at a specific moment in time. Oil is traded in a marketplace operating on the basis of supply and demand, but you won’t hear the BBC economics editor reporting that the price of data has dropped below $100 a byte, or Mark Zuckerberg announcing that Facebook is going to limit their production of data. That’s because selling data simply isn’t the business Facebook is in. That may seem like a surprising claim if you’ve watched The Great Hack, the Netflix documentary about the Cambridge Analytica scandal, but let’s dig into Facebook’s business model to see how it really works.
The cover of The Economist in May 2017.
As we’ve touched on already, Facebook sells advertising space to advertisers – not your data. Data is just one of many components in its advertising-based business model. The other components include the post format, the virtual canvas on which ads are displayed and commented on in the newsfeed. They include ‘Ads Manager’, the sophisticated software that enables marketers to create and manage Facebook ads and campaigns. And most importantly, they include Facebook’s reach – 2.5 billion people, who look at Facebook-owned apps for an average of almost an hour a day.
It is all these components working together, rather than data by itself, that makes Facebook ads so valuable. What excited me on that quiet afternoon looking at Facebook’s ads system was not the ability to reach people with diabetic retinopathy – if Bought By Many only had something to say to that one community, I could have contacted Diabetes.co.uk about sponsoring a thread on the retinopathy forum, and it probably would have worked just as well as a Facebook ad. But Bought By Many also had things to say about travel insurance to people with other medical conditions, like Crohn’s disease, angina or fibromyalgia. And we had things to say about pet insurance, too, not just to German shepherd owners like Ray, but to owners of pugs, cockapoos and giant schnauzers. What excited me most was the ability to speak to all these different communities instantly, through a single piece of software and in the familiar environment of a newsfeed. I didn’t need to research the most popular internet forums for each medical condition or dog breed, haggle with different webmasters or cobble together imagery in multiple formats to fit the requirements of different websites. From my desk in the tiny Bought By Many office, it felt like I could reach more or less anyone in minutes.
So the data gets its value from this context. It’s sometimes suggested that you can calculate what Facebook ‘owes’ you for your data by dividing its ad revenue by its number of users. Based on Facebook’s financial results for 2018, that calculation comes out at just under $25 per person per year. But that massively overstates what the average Facebook user’s data is worth, as it doesn’t recognise the importance to Facebook’s revenue of having developed the software and apps that have become part of daily life for billions of people.
Rather than oil, a better metaphor for data might be manure. Like most data, manure is a mundane by-product of life, and there are businesses that have built the logistics to collect it on a large scale and process it into something useful: fertiliser for their crops.
Let’s imagine a business that specialises in growing blackeyed peas. It spreads the manure-based fertiliser on the fields, and the plants grow tall. Eventually the pods are harvested, and the peas are picked, sorted, canned and shipped. You happen to live near the farm, so your household must have contributed a small amount of the manure. You see the peas on the shelves of your local supermarket and think of a recipe you’d like to try, so you buy a couple of tins. Now, let’s say will.i.am shows up while you are sitting down to enjoy your black-eyed pea salsa with some tortilla chips and a cold beer. He tells you that your effort to produce the manure that went into the pea fertiliser was work, that the manure was your property and that you should receive a share of the agribusiness’s revenue as compensation for it. I suspect he would have a hard time persuading you. It would be obvious that manure was only a part of the overall picture, and that its economic value was contingent on everything else the agribusiness did. And besides: you’d have your salsa to enjoy and other things you’d rather talk to will.i.am about.