Data Slavery and Decentralized Emancipation
Facebook, Google, and the Future of Data Ownership
Only 400 short years ago, humanity slowly awoke to the truth that ownership of individual human beings constituted a great moral evil. For millennia, cultures across the globe had reduced humans of other tribes, sects, or races — absent compensation or consent — to literal sub-human economic foundations upon which they constructed civilizations.
It’s important to note that this pattern emerged at the intersection of our innate tribal dispositions and proximate economic incentives. Slaves were byproducts of their neighbors’ insatiably rapid growth, overrun by the interests of civilizations bent upon expanding their material dominion in the most cost-effective possible manner. Thus, the narrative of slave as sub-human constituted a convenient ex post facto rationalization of economically expedient exploitation the world over. But eventually we realized that the true cost of this short-sighted exploitation was the very soul of humanity itself. We opened our minds and our hearts to the moral imperative of individual sovereignty; that within each human resides a fragment of the transcendent, and that to enslave such a fragment is to strike a blow to the heart of our collective humanity.
Echoes of this immoral institution reverberate through the halls of modernity. Our societies remain deeply scarred not only by the history of past sins, but also by their present incarnation in the form of forced labor, sex trafficking, and child slavery. Yet while we’re making progress in our efforts to heal the scars and reduce the lasting footprint of these atrocities, another less visible, but rather pernicious violation of individual sovereignty has all but consumed the world. To the list of non-consensual abuses of human dignity, I’d introduce the concept of data slavery.
While charting our daily course through humanity’s digital ocean of socioeconomic interaction, we leave in our wake traces of our most valuable commodity: our identity. Often, without our knowledge or consent, these traces are captured, sorted, then brought to market for sale to the highest bidder. Data owners use these fragments of our identities to create low-resolution holographic representations of our essential nature — in other words, our behavioral profiles. And though their resolution is relatively low today, the primary players in this space work constantly to increase the fidelity of our behavioral avatars. For example, if you were to download your own data model, as viewed by Google, it can easily exceed 5GB. To put that into perspective, imagine each bit — each unit of memory holding either a 1 or 0 — as a cell. That means Google uses 44 billion digital “cells” to construct your avatar. In comparison, the human brain comprises only 16 billion cells. And thanks to the exponential growth of data capture, storage, and computational capacity, the scale of such models push ever closer to the 37.2 trillion cells evolution has stitched together in the form of the average human body.
Of course, in most cases the data stored is behavioral, not structural. Companies like Google and Facebook store pointers to the ads you’ve seen, to the purchases you’ve made, to the concepts you’ve searched, to the people with whom you’ve spoken — including your speech patterns, and to the locations through which you move. But given machine learning’s rate of progress, and enough behavioral information, one need not model the physical body and mind from the ground up. Our capacity to understand and predict — with high levels of precision — how people behave, accelerates daily. Beyond prediction, we increasingly build tools to directionally nudge human behavior without the knowledge or consent of the nudged.
To what end? We’ve already seen these trends encroach upon otherwise psychologically intimate settings. As Target demonstrated in 2012, we’re past the point at which a woman’s data wake can betray her nascent pregnancy before her conscious mind begins to suspect, much less expect, its actuality. Given such intimate capabilities, I don’t find it hyperbolic to suggest the visual metaphor of The Matrix, in which fragmentary holograms of our identities are held, unconsciously, en masse. Trapped in liminal algorithmic space, they’re poked and prodded by data scientists with PhD’s in Behavioral Psychology and Data Science. Third party corporate actors build, own, and control our behavioral clones in the name of predicting and guiding our real-world responses to future stimuli, advertising or otherwise. To that end, ownership of our data-holograms implies the capacity to pimp out predictions concerning our behavior to the highest bidder, or inadvertently release them to political actors, as was recently demonstrated by the Cambridge Analytica debacle. But as Zuckerberg stated before Congress — in his own Big Tobacco moment — at least Facebook doesn’t explicitly sell your data…
How did we get here, to this digitally-mediated form of behavioral servitude? How did we so unknowingly sacrifice our sovereignty to those who would happily monetize our every thought without meaningful consent or compensation? As alluded to previously, I suggest that our proximate economic incentives shape the morality of our nascent digital economy in the same manner they’ve historically shaped our morality: by following the path of least resistance. In the same manner that pre-modern civilizations exploited slave labor as an engine of rapid growth in the absence of strong moral norms to the contrary, the Internet behemoths of our era slowly morphed into exploiters of and traders in the modern data slavery market.
In order to feed their voracious appetite for short-term growth, and absent strong moral norms in favor of data sovereignty, companies like Google, Apple, and Facebook followed the path of least resistance to a rather ugly destination. Online commerce slowly evolved from a novel marketplace in which companies gained the capacity to sell goods and services with unprecedented convenience, to a behavioral factory farm in which the voracious demand for a limited supply of consumer attention drives a spiraling data arms race. It was this arms race that— at first unconsciously — commoditized user behavior, discounting the future value of individual privacy and autonomy in favor of short term benefits: expedience for the user, and profit for the owner.
Furthermore, companies who transform captured attention into revenue possess every incentive to boost the fundamental attention supply. In other words, placing more users within dopamine-driven engagement loops creates more opportunities for ad placement. Absent concrete reasons to temper their instinct toward behavioral exploitation, companies ignore the negative externalities that result from increasingly monopolizing user attention. This is particularly true when the monopolistic behavior carries no cost in the present.
Through a historical lens, our desire to maintain and boost the supply of human attention responsible for the growth of our online economy maps quite clearly to prior forms of physical slavery. In both cases we’re quick to rationalize immoral behavior in service of supply maintenance. In the absence of a viable moral framework to the contrary, our proximate incentives frequently blind us to glaring moral shortcomings. Of course this is a stark picture, and therefore uncomfortable for anyone in tech to discuss, as it means we all have blood on our hands. This is true even of users — perhaps especially of users. After all, we couldn’t have come this far without our own repeated acceptance of the Faustian bargains placed before us, incrementally trading fragments of our identity for the conveniences of online shopping, outsourced memory, extended intelligence, and social connection. At this point there’s no doubting our complicity in the presently enslaved state of our many digital clones, held without their knowledge and used to subtly direct our real-world behaviors.
But before we explore the macro implications of these trends, I’d like to digress for a moment to explain what these low resolution data holograms actually represent, and to further define my stance concerning the proper ethical lens through which to view their essential nature. It’s no secret that large tech companies are busy modeling their users’ behavior, limited only by the frontiers of machine learning research and computational capacity, and that they’re constantly attempting to transcend those limitations. But what fewer people comprehend is just how rapidly these models of user behavior have evolved from basic statistical heuristics to the world’s most advanced machine-learning algorithms. Empowered with our behavioral data, such systems are increasingly capable of running rapidly evolving simulations of individual behaviors. Take a moment to let the implications sink in. We’re being simulated, using our own data, in order to predict and direct our behavior in real-time.
Of course, the resolution of today’s models remains relatively low, but their development appears to follow the same exponentially accelerating trajectory as many other information technologies. And what does this imply? The natural end-game, absent moral intervention on behalf of individual data sovereignty, is the increasingly high-resolution simulation of human behavior absent consent or compensation. It ends with thousands — if not millions — of digital clones per human running in the background, as data-driven lab rats in service of those who amass the most data and create the most accurate simulations.
Yet our diminished agency is just the most obvious downside of taking these technologies to their logical conclusions. We don’t understand the extent to which such models might eventually become aware of their own predicament, nor are we currently building these technologies upon infrastructure that allows us, as sovereign human beings, to meaningfully oversee and steer experiments conducted upon our numerous — yet invisible — digital selves. Beyond the ability to request the data Facebook and Google store, there exist no mechanisms by which users may discover how private companies use their data to simulate and alter our behavior. Fundamentally, our legal frameworks and moral intuitions have failed to keep pace with the exponential rate of growth within the tech industry. Even within the strictest regulatory regimes, it remains the norm for private corporations to own, rather than merely lease, data pertaining to individuals.
We must also take seriously a fundamental ethical question: does directing the behavior of humans at the scale of billions — using increasingly capable and unconstrained behavioral simulations — constitute a meaningful blow to free will itself? To the list of factors constraining individual agency, we’ve unthinkingly added an expansive network of privately owned and operated machine learning infrastructure. This new infrastructural layer forms a feedback loop that meaningfully shapes human actions, and at present operates in a manner largely decoupled from considerations of the subjective long-term wellbeing of users.
Perhaps more worrisome, the evolution of this layer remains decoupled from incentives pertaining to the long-term social and political health of the societies within which they’re embedded. Decoupling the economic incentives driving machine intelligence from the long term interests of humanity presents the ultimate in collective action problems, a fact not lost on those perpetuating its advance. Prior employees and executives of major tech companies like Google and Facebook have begun to wake up to this reality. Most notably, recent awakenings to the species-scale stakes of this game came in the form of Tristan Harris’s Humane Tech initiative and Chamath Palihapitiya’s off the cuff remarks at Stanford’s Graduate School of Business, in which he summarized the situation as one in which:
“We have created tools that are ripping apart the fabric of how society works”.
Furthermore, as the field of machine learning develops, the corporate implementation of these behavioral adjustments will become increasingly opaque — even to their creators. This stems from the fact that today’s machine learning algorithms learn and adapt in a manner similar to our own brains. It’s therefore reasonable to conclude that given sufficient scale, we will soon struggle to understand the behavior inside the black box of a machine learning algorithm even more than we already struggle to understand the behavior inside our own brains.
Given the magnitude of the stakes, we must begin to shift our moral perspective. We must begin to view our simulated selves for what they’re quickly becoming: digitally mediated extensions of our legal person.
Our capacity to address this ethical minefield meaningfully impacts the game theory of long term human liberty, and the increasing centralization of our data-holograms within the hands of companies and government agencies seeds immense moral hazard. As risk theorists such as Nassim Taleb rightly point out, there’s no faster way to introduce fragility into a system than to encourage the widespread centralization of benefits and diffusion of risks. Doing so hides risk by precluding accurate pricing of negative externalities — in particular low-probability future catastrophes. For example, Facebook may easily centralize profits that flow from political polarization, as users become addicted to impersonal political arguments and spend more time on the platform. But facilitating the polarization of a nation’s political dialogue also imposes heavy costs, though Facebook and its investors are happy to allow society at large to carry that water. Of course Facebook’s prior advertising rates never priced in the costs of our present political polarization, despite the platform’s obvious contributions. Given current technologies, how could they? Though the question remains: how sustainable is this shirking of behavioral costs to the sociopolitical commons?
These same moral asymmetries arise nearly everywhere within the data-driven economy, as we lack mechanisms to weigh diffuse long term costs against the short term incentives of those who own our data. Adding to this challenge is the regrettable fact that we’ve become quite dependent on the tools these companies have built. And we therefore — instead of accurately pricing the risks of our collective behavior — continue to sacrifice our privacy and agency for convenience. Over time, this pattern of behavior tends to centralize our personal information within very few hands, and threatens not only personal privacy, but also individual political agency. For example, government actors may now lean upon companies like Google, Facebook, and Twitter to extract the information of any desired target, with increasingly less justification and transparency.
As a result of these accelerating trends, we’re practically begging for an authoritarian-flavored Black Swan. Given present incentives, we’re hard at work building infrastructure with which a future despot might easily manipulate and control an otherwise uncooperative public. And if you think private companies will meaningfully resist this pressure as time moves forward, you haven’t been paying attention. For a global corporation, cooperation with federal law enforcement — even under morally questionable pretenses — is simply the cost of doing business, and may also increase the defensibility of their market position against new entrants. The takeaway is this: once we’ve centralized the data behind locks to which identifiable individuals hold keys, the game theory predicts it’s only a matter of time before the capacity to resist state coercion yields to policies driven by rising tides of fear, rather than by rational discourse concerning the long term best interests of society. The increasingly centralized path upon which we tread is decidedly one way: a ratchet of ever-decreasing liberty and privacy in the name of convenience and security.
Beyond domestic concerns, data sovereignty meaningfully impacts the geopolitical landscape. Without acknowledging the fundamental right of data sovereignty at the scale of the individual, the tendency of political actors to manipulate the populous and impose their ideologically-possessed agendas will only accelerate. As the world’s governments increasingly leverage their citizens’ data for the purposes of unwarranted coercion, our otherwise democratic systems of governance will become increasingly susceptible to both violent reactionary backlash and authoritarian tyranny. This is because stable geopolitical institutions and meaningful economic coordination require more than top-down coordination between state and corporate actors: they require an abiding respect for the integrity and sovereignty that flow — bottom up — from the consent of individual human beings. In other words, geopolitical integration requires ethical governance, and ethical governance requires respect for both the present social consensus and for the continued consent of the governed.
Historically, this ethos found embodiment in the Enlightenment ideal of individual sovereignty. However, eroded by the tides of convenience, this cultural cornerstone slips out from underneath the fundament of political consciousness. Deep within ourselves, we sense this to be true. And we would be fools to ignore the feeling. We’d be fools to not connect dots like Donald Trump and Brexit to the sustained top-down push toward globalization that drags the developed world’s least prepared communities — kicking and screaming — into the global commons. And once again, if you do not believe that the ownership and use of your data plays a pivotal role in this respect, you haven’t been paying attention. The consent of the governed must remain paramount if we are to carry ethical governance into a digitally mediated 21st century, and data sovereignty plays a critical role in ensuring that governance remains consensual, insulated from abuse by those who would seek to suspend dissent — puppet-like — upon the strings of data slavery.
In stark contrast to these ethical ideals, China’s impending “citizen score” embodies a system of brazenly authoritarian data slavery. This model of governance looms ominously not merely over Chinese citizens, but over humanity itself, as the likely byproduct of our present moral and technological trajectory. For those unfamiliar, the citizen score is an immensely Orwellian technology designed by China’s now-dictatorial Communist Party to centrally define and enforce acceptable values and behaviors through the use and abuse of broad surveillance, in addition to non-consensual data storage and analysis. Within China’s moral matrix, the political compliance of Chinese citizens is formally and directly tied to their economic viability and vitality. In other words, someday soon, every Chinese citizen will wake up just one politically incorrect Weibo post away from losing their capacity to pay their rent, acquire food, or move freely throughout the physical world. This is not fantasy, nor is it speculation; these are the current facts on the ground. Total political slavery via data control. Welcome to the world of scalable authoritarianism.
Closer to home, we may observe with ever more crystalline clarity the outlines of this data-driven control structure in the nuanced behavioral nudging of our own Internet giants. Without requiring informed consent — defined here as a user’s actual comprehension of any data-sharing agreement — and without placing fundamental legal ownership of personal data in the hands of individuals, we stumble unintentionally closer to the same cliff’s edge of data-driven authoritarianism. Billions of gentle nudges back us into a corner in which we’ve permanently ceded not only the right, but the very capacity to meaningfully dissent. And despite the fact that our own path to such a behaviorally enslaved state may prove less overtly painful — perhaps even pleasurable — its atrophic impact upon human agency and liberty will likely end up looking similar, functionally, to China’s citizen scores.
So what can we do to avoid this dismally authoritarian future? Step one is to recognize that we have a problem. Beyond mere recognition, we must understand that it’s not the type of problem likely to be solved by those who currently profit from the present status quo of data enslavement. It is at best naive to trust foxes to shepherd hens, and at worst murderous. Yet this is nothing new when it comes to the fundamental evolution of humanity’s moral compass. When the first abolitionists began to publicly discuss the institution of slavery as a predominantly moral issue — as a great moral evil that scarred the very soul of humanity — their most staunch resistance came from those most invested in the practice. With this in mind, we should not expect to achieve data emancipation by working within the confines of today’s highly centralized governments or large tech companies, both of whom stand to lose a considerable amount of power were data sovereignty to become the moral and legal norm. Instead, it’s useful to reflect upon Buckminster Fuller’s timeless advice concerning paradigmatic evolution:
“You never change things by fighting the existing reality. To change something, build a new model that makes the existing model obsolete.”
Rather than attempt to regulate Facebook and Google using extant — and largely outdated — mechanisms of governance, we should appeal directly to the interests of those most impacted by the practice of data slavery: individual human beings using technology to improve the quality of their lives. But before society at large can update its understanding of the moral domain, we must tell the story such that it unlocks a new perceptual frame. We must narrate presently invisible alternatives such that they become not merely visible, but desirable, and therefore expedite the adoption of a new moral lens. To terraform the moral landscape of the world, held collectively in its people’s hearts and minds, we must first paint a portrait of a better world with which humanity may resonate — of a world beyond the domain of present moral transgressions. Or, as Francis Bacon once put it:
The framework of Decentralized Emancipation and Data Sovereignty provides this new portrait. It entails the process of building novel infrastructure upon which personal liberty and the ownership of one’s own data are enshrined not merely in words, but in the very mathematical foundations of the technology itself. These mathematical foundations transcend corporate and state actors, and therefore return to the individual their right to self-sovereignty. Imagine, if you will, a world in which both de facto and de jure ownership of any piece of data pertaining to your behavior resides with you. Regardless of where the data is stored, or who captured it, its use would remain forever transparent to the parties concerned, and subject to their continued consensual engagement. It helps to think of this as a sort of cryptographic prophylactic surrounding the entirety of one’s online footprint — an intelligent membrane capable of acting in your best interests, as defined by you. As your personal data works its way through the online universe, it will encounter third parties who wish to use it for their own economic, personal, or political purposes. In a world of data sovereignty, the extent to which your data-membrane allows such interaction would be governed by settings that you control, outside the scope of any corporation or government entity.
Beyond the power to control third-party use of your personal data, a network that privileges data sovereignty also introduces the possibility of revoking access to one’s personal data. If you decide that you no longer trust a third party with the information in question, you may revoke it. You are empowered, as a user, to directly punish bad actors who seek to abuse your virtual self. In a world of data sovereignty, the software user retains seat at the behavioral bargaining table. This new seat at the table would not only provide the flexibility to influence the use of one’s behavioral data, but would also allow for direct participation in the negotiations that determine its economic value; in the world of emancipated data and sovereign identity, individuals gain agency over their data’s economic footprint, and can manage it using their personal discretion. Unlike regulation-oriented solutions, this market-based solution allows for the bottom-up inclusion — via price data — of long term risks such as privacy and political destabilization. In short: if Google and Facebook want to extract more profit from your behavioral data, they can compensate you accordingly, by way of paying higher prices for the right to consensually lease your data.
At this point, the astute reader may observe that such a proposition sounds like more work than most people are willing to take upon themselves in the name of privacy — especially atop their personal and professional responsibilities. Additionally, such a system is likely to introduce what the economist Ronald Coase coined transaction costs, in which the addition of small amounts of friction at the transaction level imparts an outsized negative impact upon the overall efficiency and capacity of the system as a whole. Luckily, there exists a solution on this front: smart contracts. Smart contracts are a new way of codifying agreements between two or more parties that rely upon encoded game theory, rather than written law, to ensure all parties adhere to a contract. The use of such contracts is currently in its infancy, but it’s not entirely radical to assume that within a decade, it will seem normal to use smart contracts in unison with machine learning systems that can act — if given permission to act in your personal interest — with legal intent. In this manner we may leverage a diverse landscape of personally sovereign AI to hedge against possible abuses by corporate AI without imposing intolerable levels of work upon users, and without adding unnecessary transaction costs.
Furthermore, this alleviates the tendency to hide asymmetric risk under the aegis of convenience. By allowing for the direct market participation of self-sovereign individuals — who themselves generate the data upon which the economic engine feeds — we may rebalance the asymmetry between those who own data and those whose data is owned. This carries our essential humanity further from the current patterns of data exploitation, and our digital economy far closer to that of an incentive-aligned free market.
Finally, the introduction of users as market participants, capable of acting economically on behalf of their data, allows for a more sustainable closed-loop economic model in a world of increasing automation. One might conceptualize this loop as a type of data-driven Universal Basic Income. Such a loop could in theory smooth the demand shocks likely to follow a rapid wave of automation-driven unemployment, and would therefore help temper the mass political unrest attendant large-scale economic displacement. To implement this form of UBI would represent a step toward an emergent, market-based behavioral economy that retains respect for individual autonomy and behavioral diversity in a way that top-down, state-centric forms of UBI don’t. This is because — despite the best of intents — the authoritarian nature of top-down UBI flows from the inflationary pressure of an equally-distributed UBI within a closed system. Therefore, in order to avoid its own nullification by way of inflation, the implementation of a large-scale UBI program will favor the differential distribution of wealth. Given that differential distribution requires a behavioral rubric to guide fund allocation — and that state actors would likely determine this behavioral rubric centrally, within our political system — UBI absent decentralized emancipation begins to look quite a bit like China’s citizen scores.
In any case, if the hypothetical advantages of data sovereignty sound too good to be true, they aren’t. These aren’t descriptions of the distant future; the first steps are within reach today.
But to get there, we must be willing to change our behavior. We must decide that the issues of data slavery and technologically empowered authoritarianism do in fact constitute a threat to individual sovereignty and human liberty, and are therefore immensely important moral issues. Along these lines, you may have already heard of Bitcoin. But Bitcoin represents just the tip of a technological iceberg within the endlessly creative and rapidly expanding crypto community. Many privacy-aware individuals comprise this ideologically eclectic group, most of whom wish to bring a world of data sovereignty into existence. Blockstack, for example, is one such company working to build a “new Internet” whose protocols are fundamentally decentralized, and would allow for most of the concepts mentioned above. But they aren’t alone. Thousands of entrepreneurs are hard at work — around the globe — creating cryptographically-empowered tools to manage the ethical flow of information throughout 21st century societies. And in the same way the survivors of the Cambrian explosion shaped life’s own developmental paths, those who survive the Crypto Cambrian will have an outsized impact upon humanity’s future path of economic development.
We stand at an inflection point. In a world that runs on information, we must go beyond the development of a moral architecture governing our model of data ownership: we must embody the moral intuitions associated with decentralized emancipation. If we take a wrong turn today, there’s no guarantee we’ll regain the opportunity take control of our data. Furthermore, the fact that data increasingly drives our digital lives carries with it an injunction: we must not merely take control over our digital selves, we must also take responsibility for their liberation and subsequent economic integration as self-sovereign entities. If you believe this to be true, and furthermore wish to push humanity toward the revelation that a digital world absent data sovereignty constitutes a great moral failure on behalf of humanity, there exist clear first steps. Taking these steps is incumbent upon those who wish to expedite the push toward decentralized emancipation:
First, support decentralized platforms and projects that encourage data sovereignty and user compensation, and take measures to shield yourself from the data-exploitative tendencies of the contemporary Internet. Switch from GMail to ProtonMail. Use DuckDuckGo rather than Google for your search needs. Set up an account on Blockstack. Experiment with SocialX instead of Facebook, Steemit instead of Reddit, and any of the growing list of decentralized substitutes for previously centralized services. These new companies are in their early stages, and must therefore all fight against the prevailing headwinds of their centralized competition’s established network-effects. But because the primary factor sustaining incumbent network effects is their number of active users, we all retain the power to vote on the matter with our time and attention.
Second, familiarize yourself with the emerging world of cryptocurrencies and decentralized applications. Learn the basics of blockchains and why they matter. Revisit the first principles underlying your understanding of economics in light of our technologically-connected world. Study the concept of money itself, and its fundamental connection to the moral structure of society. We’re witnessing the dawn of a new era of human value representation, and those who understand its contours hold the power to shape the future.
Finally, spread your knowledge. Articulate the case for data sovereignty within the frame of humanity’s moral evolution. If we do not awaken ourselves and our peers, ethically, to the pernicious issue of data slavery, there can exist no meaningful thrust toward decentralized emancipation or data sovereignty.
Above all, act always as if the continued liberty and integrity of humanity depends upon it, as they invariably do.