IT-Bosses tell us that all our problems can be solved by collecting information. The more you give, the more you get. But then there is only „social physics“ instead of politics left.
Even the mavens of Silicon Valley are occasionally right. And so: the production, accumulation, and analysis of traces left by the digital devices does produce real benefits. Following the logic of the Peace Dividend – a popular slogan of the early 1990s, which held that decreased military spending would promote economic growth – we can speak of the Surveillance Dividend: the idea that the Internet of Things and Big Data and the inevitable disruption of the entire universe by a handful of Californian start-ups will yield economic abundance, political emancipation, universal prosperity.
Thanks to the increased trackability of everything, we can design better, optimize better, govern better, know better. The Surveillance Dividend boosts efficiency. It saves money. It extends lives. Its benefits are real. The right question to ask, then, is not whether the Surveillance Dividend allows us to govern better or know better. No, it’s to ask: better than what?
To answer that question, it helps to analyze how the proponents of the Surveillance Dividend tout its benefits across different domains. In his new book Social Physics, Alex Pentland, a professor at the MIT Media Lab, an adviser to the World Economic Forum in Davos, and a very important man (according to his web-site, he dined “with British Loyalty and the President of India”), describes an experiment called FunFit that he ran in Boston.
The goal was to get members of a local community to be more physically active. In the past, the study might have deployed a social advertising campaign about the benefits of health. Or it might have paid individuals to stay fit. Pentland, however, chose a different strategy: everyone in the study was paired with two peers from the community. They might have known them only slightly or, on the contrary, be very close to them. These two members were provided with small cash incentives for getting their shared contact to increase their physical activity, tracked by accelerometers in the smartphones provided by the study. Thus, if you walk around more than usual, your peers – not you! – get the cash.
The results were astonishing: the scheme worked almost four times more effectively than the traditional approach of paying individuals directly. Moreover, if your peers were people with whom you interacted a lot, it worked eight times better. Pentland, thus, announces the birth of a new discipline: “social physics.” Thus, by studying our existing social relationships and using that knowledge to provide specifically crafted incentives to individuals, we can finally address the long-neglected social problems.
Observation solves problems
Pentland provides another example: during the 2010 congressional elections in the US, American academics conducted a study on 61 million Facebook users divided into two groups. Both saw messages urging them to vote but the first group saw a generic and unpersonalized message while the second group saw a personalized message that featured faces of their friends who had already voted. The laws of social physics held up: more people from the second group actually voted. For close friends – as opposed to mere online acquaintances – the results were particularly impressive: four times more people voted after seeing the personalized message.
Systems based on social physics work because they know us: not just our daily movements and communication patterns but also our friends and the nature of our relationships. Social physics has mind-boggling implications. With enough data-mining, one can find the right neighbors to convince us to cut energy consumption, the right friends to tell us not to eat junk food, the right colleagues to remind us not to slack off during work hours. It’s all about finding the right people at the right time and getting them to send the right messages.
The granularity and trackability of our digitally mediated social relations makes it possible to turn them into yet another tool of what Michel Foucault called governmentality. Instead of appealing to the well-being of the community or the self-interest of the consumer in the marketplace, one can regulate individual behavior by using friendship itself as a tool of governance, selectively exposing us to the various parts of what technology companies call our “social graph.” Pentland proposes some institutional solutions for resolving privacy concerns but such solutions ought not to concern us here. What matters is that the Surveillance Dividend is real: continuous observation of individuals can solve problems.
The living labs
Facebook’s get-out-the-vote experiment is a randomized control trial, a popular type of scientific experiment, originally popularized in the medical context, where experimenting subjects are randomly divided into two or more groups. One group receives some intervention – e.g. they see photos of their friends who already voted – while the other group doesn’t. Such studies are increasingly popular among social scientists and services like Facebook – with their millions of users and easily adjustable settings for what those users see – are ideal experimental grounds, full of unwitting guinea pigs (that is, us).
The furor over a recent study, where Facebook showed some happy users posts that were positive while unhappy users saw negative posts, seems rather naive. As one of Facebook’s data scientists put it a few months before the scandal, “at Facebook, we run over a thousand experiments each day. While many of these experiments are designed to optimize specific outcomes, others aim to inform long-term design decisions.” Translation: better worry about the thousands of daily experiments that we did not tell you about!
As evidence-based and results-oriented policies are much in demand these days, Facebook supplies us with ideal intellectual infrastructure for testing which interventions work and which ones do not. The Surveillance Dividend again: the more Facebook tracks us, the more effective are the policies that truly change the world – and in real-time rather than two years later. Pentland even wants to “revive the social sciences by constructing living labs to test and prove ideas for building data-driven societies.” James Fowler, the co-author of the Facebook election study, pushes it even further, claiming that “we should be doing everything we can to measure the effects of social networks and to learn how to magnify them so that we can create an epidemic of wellbeing.”
From Observation to Explanation
A recent article in Foreign Affairs reveals other benefits of the Surveillance Dividend. Since the poor are under “the constant pressure...to spend their money on immediate needs,“ the authors, who work for the Bill & Melinda Gates Foundation, laud the potential of mobile phones to “nudge” the poor to make regular saving deposits. To save or to spend is not the only big decision facing the poor; in developing countries, such decisions – about vaccination, education, crop insurance – are many and they are not always taken under the best of conditions.
Why, then, not turn cellphones into the poor man’s Siri or Google Now – the two popular virtual assistants – and have them continuously monitor what the user is doing, tracking environmental constraints and suggesting the right decisions? The Surveillance Dividend yet again: thanks to constant tracking, otherwise vulnerable people can become more resilient and resourceful in tackling their problems. One day, with better smartphones, we can even teach them how to code!
Such ideas look appealing for two reasons. First, the persistence of social problems, from climate change to poverty to obesity, has produced a nearly universal consensus that more drastic measures are in order. Thus, openly paternalistic methods that would previously be taboo are now up for discussion. Academics keep churning out books with titles like Against Autonomy and Epistemic Paternalism: A Defense, which emphasize the necessity of interfering with the decision-making of individuals, either in the interests of the community or for their own good.
The continued appeal of behavioral economics, which aims to correct what it takes to be naïve assumptions about human rationality made by neoclassical economics, is another factor. Behavioral economists want to account for how people behave in the real world and not in some fancy theoretical models. To that end, they – and especially those academics working on global poverty – go out into the field and, after carefully observing the poor, conduct randomized control trials to see if their hunches are correct.
App against poverty
Such hunches do not always yield theories or basic causal explanations: if the researchers see, say, that a rural school with one schoolteacher does a better job at educating students than a school with two, such a discovery becomes “actionable” even without theory. There is a certain similarity here to the can-do, results-oriented attitude embraced by technology companies: Facebook doesn’t need to know why happy stories make users click more in order to use this knowledge. The end of theory, predicted by Chris Anderson in Wired in 2008, reached this field somewhat earlier: when so much can be observed, studied, and tested, extensive theoretical and philosophical debates only get in the way.
One of the underlying assumptions shared by many behavioral economists is that we don’t always behave in our best interests due to specific reasons that can be identified, categorized, and rectified. In their recent book Scarcity, Eldar Shafir and Sendhil Mullainathan, two prominent economists who have pioneered the application of behavioral economics to the study of poverty, suggest that poor people become so overwhelmed by anxiety imposed by always having to worry about money, that they end up making decisions that are not in their own interest. Poverty, they argue, follows from cognitive scarcity, which “rather than a personal trait... is the outcome of environmental conditions... that can often be managed” – a perspective which, they claim, provides “a radical reconceptualisation of poverty.”
In other words, poor people make bad financial decisions because their other concerns steal their “cognitive bandwidth” much like using Skype or Spotify might be stealing your Internet connectivity. On this view, if only the poor got the right text message at the right time, they might actually end up saving more. To fight poverty, then, we must “scarcity-proof our environment” so that poor and irrational decisions are excluded or minimized through an always-on monitoring system of some kind (Mullainathan and Shafir compare it to a smoke alarm).
Poverty, then, becomes an information program that can be fought with the kinds of information tools that yield the Surveillance Dividend. Consider a smartphone app called BillGuard. It will not only alert you once your intended monthly spending limit has been exceeded but it would also search the Web for coupons that will lower your bills based on your spending patterns. Or take the iBag - an actual bag, equipped with sensors and connectivity, which automatically locks itself – and, presumably, your wallet – when it believes you might be close to overspending. Constant monitoring is what makes such innovations possible.
Such apps might lift some people out of poverty. They might even make their developers rich. But what is the cost of “informationalizing” poverty? And is it how we – and “we”here refers to that almost forgotten entity, a community of citizens, not clever venture capitalists or all-disrupting entrepreneurs – want to fight it?
Similar acts of informationalization – whereby a problem is stripped of its material and political dimensions and is presented simply as a matter of undersupplied or delayed information – can be observed in other domains. Max Levchin, a co-founder of PayPal, hopes to apply machine learning and data mining to solve health problems. “Health is a big information problem waiting for data analytics and wearable sensors,” he said on announcing Glow, his app for helping women to conceive. Glow does this by tracking their sexual activity (including positions) and menstrual cycles and sending them various alerts (“fertile window starts!” or “woohoo! You’re ovaluating!”).
Levchin might be driven by noble motives but whether health – or anything else for that matter – constitutes an “information problem” is not a question to be taken lightly. Silicon Valley, for the most part, has already answered it for us – in the affirmative. Give them any problem – and, a few apps later, an “information” solution is magically found. Recast that way, the problem inevitably leads to the invocation of the Surveillance Dividend and its unquestionable benefits. But shouldn’t we also inquire what happens once health, education, poverty are recast as problems presumed to be solvable via information?
Problems and Surveillance
The growing fascination with the potential of smart energy is another potent example of such informationalization in action: reduced to apps, smart thermostats and smart meters, energy is decoupled from the vast complex of political and economic networks responsible for its production and is presented solely as an information problem that omnipresent feedback loops can solve. As the Australian academic Yolande Strengers writes in her new book Smart Energy Technologies in Everyday Life, the consumers are imagined as “data-driven, information-hungry, technology-savvy“ while the provision of data” is the only means by which [they are] understood to operate and change.”
That there are alternative, non-informational, more political routes to tackle energy is not something you would learn by starring into your smart meter; it does not develop much literacy and expertise about energy consumption so that, as Stengers notes, “consumers begin to equate ’energy management’ with ’data management.’” But energy literacy is not just a factor of individual preferences and feedback loops; it also requires some reflection about energy efficiency standards, building design, consumption habits, indoor cooling practices.
Even more disturbingly, the problems that the Surveillance Dividend can help solve – climate change, obesity, poverty – are increasingly recast in the language of national security – and, once this rhetorical step is made, the terrified public accepts even the most drastic measures. This national security connection is no overstatement: there are more and more studies that purport to show the links between climate change and the likelihood of civil wars, the level of poverty and the degree of youth radicalization and so on: the military industrial complex knows how to extend its tentacles into seemingly non-military domains.
Freedom of Information Laws
The proponents of the Surveillance Dividend know this. Here is how Pentland links apps, public health, and matters of national security: „an app on a phone could quietly look for uncharacteristic variations in behavior and then figure out if an illness is developing, “ he writes in Social Physics. Thus, “the ability to track diseases such as the flu at the level of individual should give us real protection against pandemics, because we could take steps to reach infected before they spread the disease further.“ Given that the security paradigm still dominates most policy debates on both sides of the Atlantic, such arguments will find support in all the right three-letter agencies.
Thus, not much will be achieved by questioning the benefits of the Surveillance Dividend. Social physics, randomized controlled trials (RCTs), nudging: they aren’t useless. Proponents of the Surveillance Dividend frame their benefits as self-evident and apolitical: we are told that informationalizing problems just makes them more knowable and more manageable. But there’s nothing self-evident or apolitical about the tools and methods of the Surveillance Dividend. In reality, they only see what they want to see and they only know what they want to know. What they often don’t know and don’t want to see is their own politics.
We live in an age of profound epistemic asymmetry. The hyper-visibility of the individual citizen – trackable through all sorts of smart devices – is matched by the growing hyper-invisibility of all other players. Governments continue to classify more documents, outsourcing their functions to private companies that don’t have to comply with freedom of information laws. Corporations sow confusion about the real impact of their activities, deliberating manufacturing ignorance by funding dubious pseudo-scientific research. Wall Street churns out financial instruments so opaque as to defy any efforts at comprehension.
The open data movement might address some of these challenges but its greatest success to date has been getting governments to release data that is mostly of economic and social utility. The thorny political data is still closely guarded. There’s no “social physics” for the likes of Goldman Sachs or HSBC: we don’t know the connections between their subsidiaries and shell companies registered in tax havens. Nobody is running RCTs to see what would happen if we had fewer lobbyists. Who will nudge the US military to spend less money on drones and donate the savings to the poor?
The tools of the Surveillance Dividend work at just one level: that of the individual citizen. They render the citizen fully transparent and manipulable, creating a semblance of “problem-solving” while leaving governments and corporations free to pursue their own projects. To paraphrase Foucault, we have all become eminently trackable and eminently nudgeable. Our bad habits can be detected, analyzed, and corrected in real-time, thus dissolving many of the problems that are currently overwhelming the social services. Thus, the very notion of politics as a communal enterprise morphs into individualistic, consumerist-friendly spectacle, where solutions – we call them apps these days – are sought in the marketplace rather than in the public square.
This individualization of politics is not very surprising, for the methods that give us the Surveillance Dividend have deliberately abandoned any systematic search for factors and causes of social change that transcend the individual. In trading causal explanations for “actionability,” their proponents have, in fact, abandoned theory and thus have to feign ignorance or naivete every time they are faced with a problem that cannot be easily reduced to individual decision-making. Do we really have to run a randomized controlled study to know what it is that lobbyists or bankers do all day? The world might be maddeningly complex – but it’s also embarrassingly simple: corporations still want to make money, governments still want to build bureaucratic empires, security agencies still want to grab power. “Theory” might have ended but why dispense with the obvious?
Dispensing with causal explanations can no doubt be a profitable business strategy. A car dealer can profit from knowing that used cars that are orange are more reliable than used cars in standard colors – a typical (and real) correlation revealed by Big Data analytics – without having to explain why. Transporting this model from business to politics, however, entails making far too many assumptions about the scope and the purpose of politics as well as the distribution of blame between players.
Shrinkage of political imagination
The individualistic lens through which the tools and methods of the Surveillance Dividend approach social problems simply take certain questions off the table. In what is perhaps the most stinging critique of behavioral economics in the realm of development, the economist Sanjay G. Reddy points out just how this quest for empirical and evidence-based solutions empties debates off many critical issues. “The larger questions once asked within the discipline, regarding the effect of alternative economic institutions and policies (such as those concerning property arrangements, trade, agricultural, industrial and fiscal policy, and the role of social protection mechanisms) … have been pushed to the background in favour of such questions as whether bed-nets dipped in insecticide should be distributed free of charge or not, or whether two schoolteachers in the classroom are much better than one,” he writes.
The Surveillance Dividend reduces politics to knob-twiddling as if society was just a radio to be finetuned. Worse, when the information-based solution is immediately available – which would be the case once everything is digitized and interconnected – whoever wants a non-informational solution faces the burden of having to prove why this less efficient route is better than deploying the Surveillance Dividend again.
However, a politics made of smart devices is not necessarily a smart politics. Recently, the Wall Street Journal featured a smart toilet that can “sync with users’ smartphone...and play their favorite music through speakers built into the bowl.” It’s trivially easy to make it run randomized control trials to see if the music makes users happy and nudge them towards a healthier diet after analyzing their, well, output. That such a gadget can plausibly pass for an instrument of contemporary politics – ideal for creating an epidemic of wellbeing – is a very sad testament to the shrinkage of our political imagination.
Hier können Sie die Rechte an diesem Artikel erwerben.
Natürlich ist das Kitsch, es ist maßlos und fern jeder Realität: Tom Cruise fliegt wieder in „Top Gun: Maverick“. Wer sich diesen Pathosformeln des Kinos entziehen kann, der hat das Hollywood-Kino vermutlich nie ganz verstanden.