The starting point
We don’t need conspiracy theories to see that something profound is happening to the ways we interact with technology. The verifiable facts are sufficient.
Around the world, corporations and governments are recording our movements, purchases, interactions, faces and voices in ways unknown in any previous era. These records feed machine‑learning models that filter job applications, determine the price we pay for insurance, decide which transactions appear suspicious, trigger airport screenings, and shape every aspect of the information we see online. Legal frameworks in many countries explicitly permit state access to telecommunications and cloud data under broad security or law‑enforcement powers. Commercial actors build parallel infrastructures for advertising, fraud detection and “customer insight” using the same raw material.
Facial recognition has moved from laboratories and police experiments into airports, city streets, shopping centres and smartphones. Large‑scale data breaches and scraping operations have shown that billions of images and records can be harvested without informed consent and then reused in ways that individuals neither expect nor approve. Automated decision‑making has been adopted in credit scoring, welfare eligibility, employment screening, border control, policing and content moderation. In many of these domains, appeal mechanisms are weak, opaque, or non-existent. A person can be blocked, flagged or deprioritised by a scoring model they have never heard of and cannot interrogate or challenge.
These are not speculative observations. They can be validated from government documents, corporate disclosures, investigative reporting, academic research, technical standards, court cases and the statements of those who build and supervise these systems. The unresolved questions lie elsewhere: What kind of civilisation emerges when such systems become the default grammar of public and private life? Who benefits, who loses, and according to whose worldview? At what point does administration slide into governance by unseen metrics?
I want to explore both the advantages and the dangers of this emerging architecture without succumbing either to fashionable alarmism or to the kind of complacent boosterism that confuses novelty with progress. The point is not to rehearse clichés about “technology” as if it were an autonomous force, but to ask whose purposes are encoded into these systems, and whether a different pattern is still possible.
The promises of life with data
To start with the attractions is important. If there were no tangible benefits, we can safely assume that these infrastructures would already have been abandoned. People accept pervasive data collection in part because it appears to work.
One self-evident benefit is efficiency. Automated decision‑making, when it’s well designed, can process volumes of information far beyond the reach of individual human minds. Without machine learning models to detect anomalies in financial transactions, identify routes for ambulances, allocate scarce hospital resources, or filter spam and automated attacks, many existing services would collapse under their own complexity. Automated risk scoring, albeit imperfect, does prevent some fraud, catch some money laundering, and reduce some types of crime. In that sense, digitised surveillance and scoring build on older practices of record‑keeping and bureaucracy rather than replacing them.
Another attraction lies in the promise of personalisation. Data‑driven systems can, in principle, adapt to an individual’s preferences, constraints and abilities. A student might receive targeted remedial material. A farmer might obtain customised weather, soil and market forecasts. A migrant sending remittances might benefit from lower fees because their risk profile is modelled more precisely. In these situations, data‑driven inference can widen opportunity rather than narrowing it.
A third argument, especially prominent in lower‑income contexts, is inclusion. Where formal identification systems or credit histories are weak, digital records can, at least in theory, provide an alternative route into banking, welfare, healthcare and education. The rhetoric surrounding certain large national digital ID schemes leans heavily on this claim: that without robust identification and data, poor citizens remain invisible to the state and to markets, locked out of services others take for granted. Whether this rhetoric is borne out in practice in a fair and non‑coercive manner remains an open question, but it can’t be dismissed without scrutiny.
There is also the question of safety. Many people welcome cameras in public spaces, monitoring on public transport, and stricter checks at borders and stadiums, particularly where actual experiences of violence or organised crime are part of everyday reality. Parents may welcome social media age‑controls or monitoring of their children’s online behaviour. Communities facing kidnapping, terrorism or recurrent riots may judge that the trade‑offs in privacy are justified by reductions in risk.
To acknowledge these gains doesn’t mean we must accept the present direction as inevitable or benign. It does, however, remind us that the emerging “metric society” is not an aberration imposed from outside. It’s created, validated and normalised through choices made at every level – from governments and corporations to families and individuals – under specific historical and cultural assumptions about security, progress and control.
Worldviews, world-systems and mindsets
If we look only at technical features, we miss the deeper story. Data infrastructures are not just tools; they’re crystallisations of a particular worldview.
One recognisable worldview treats society primarily as a system of risks to be managed and resources to be allocated. From this point of view, individuals appear first as data points: potential customers, potential threats, potential patients, potential voters. Data collection is justified as a means to reduce uncertainty and optimise outcomes. The ideal is a world in which nothing important is left to chance because the relevant variables are all being tracked and modelled.
When that particular worldview hardens into a world‑system, we get the expanding apparatus of identification, scoring and surveillance. The system is expressed in laws that require telecoms to retain metadata, in standards that encourage interoperability of IDs across sectors, in procurement contracts for facial recognition at transport hubs, in credit scoring models that integrate social and behavioural data, and in platforms whose revenue depends on predicting and nudging user behaviour at scale. None of these components is decisive by itself; together, they compose an environment in which to be a citizen increasingly means to be legible to machines.
Within that environment, people adopt mindsets that are shaped by both promise and threat. Some become fluent in gaming algorithms, curating their online personas and transaction histories to avoid being flagged by systems they don’t fully understand. Others, fatigued by constant prompts and disclosures, submit to a quiet resignation, signing terms of service without reading them and assuming that nothing they do can genuinely protect their privacy. A few attempt to step outside these frameworks altogether, reducing their digital footprint, paying in cash, shunning social media. But even that stance becomes a signal within the system, potentially counting as a risk marker in itself.
This pattern is not confined to one region or ideology. Authoritarian regimes may use the same infrastructures overtly to enforce loyalty and suppress dissent. Liberal democracies cloak them in the language of choice, convenience and consumer empowerment. Corporations couch them in the jargon of “customer experience” and “risk management”. Yet beneath these differences runs a shared assumption that whatever can be measured and predicted should be measured and predicted – because not to do so would be irresponsible or uncompetitive.
At this level, the distinction between East and West, public and private, security and marketing begins to blur. The core question becomes whether any society that builds its governance around data and prediction can avoid sliding towards a situation where human beings are primarily known and treated as aggregates of behavioural signals.
The shadow side: opacity, dependency and dispossession
The disadvantages of this drift are no longer abstract. They show up in daily frictions that are often experienced as personal misfortune rather than as symptoms of systemic design.
Opaque decision‑making is the first of these. When a visa application is rejected, a welfare payment delayed, a digital wallet frozen, a job application ignored, or a social media account suspended, the person affected is rarely given a clear explanation. The formal reason, if one is given at all, will cite a breach of vague rules or the need for “security checks” rather than the specific pattern in their data that triggered the action. Behind the scenes, a model has translated fragments of behavioural history into a score. The person is judged without being heard, and without even knowing the criteria by which they are classified.
This opacity matters because power is being exercised whether or not it’s acknowledged. When human officials were the primary gatekeepers, bias and arbitrariness were still present, often brutally so. Yet there remained at least a theoretical space for dialogue or appeal. When we shift the locus of judgment to machine models, the criteria can become more consistent, but they also become more obscure and less negotiable. Accountability evaporates into technical jargon and corporate secrecy.
The second disadvantage is a deepening dependency. As critical infrastructures – payments, healthcare, communications, education, transport – become entangled with digital identification and data‑driven risk assessment, the ability to function in society without accepting these systems shrinks. To decline biometric registration, to refuse a smartphone app that proves vaccination status, to avoid social media accounts that are increasingly demanded for background checks, is to risk exclusion from essential services or opportunities. When opting out comes at the cost of participation, “consent” loses its meaning.
In some places, this dependency already tilts towards dispossession. Identity systems that are poorly designed or badly governed can lock people out of welfare benefits and healthcare when records are wrong, documents are missing, or connectivity fails. Credit scoring models that draw on behavioural and social data can entrench existing inequalities, penalising those whose lives do not generate the “right” kinds of signals. Facial recognition deployed without robust oversight has been documented to misidentify people with darker skin tones at higher rates, raising questions about who bears the burden of error.
A deeper dispossession occurs at the level of self‑understanding. When one’s worth is continually quantified in terms of ratings, scores, followers, and other metrics, the temptation is to internalise these numbers as a measure of personal value. This is as true for a gig worker being rated by customers as it is for a young person tracking likes on their posts or a citizen worried about their “risk profile”. In such a culture, people are invited – perhaps compelled – to treat themselves as mini‑enterprises curating their brand and managing their data in order to remain competitive. The distinction between authentic character and optimised performance becomes blurred.
A final disadvantage, and I would argue the most dangerous, is that this complex of surveillance and scoring can be repurposed rapidly under conditions of crisis. Whether the crisis is framed as terrorism, a pandemic, financial instability or climate emergency, the infrastructures are already in place to tighten monitoring and restrict movement, speech or association. The temptation, for any authority, is to slide from asking “What must we know to keep people safe?” to “What can we know, now that we have this capability?” At that point, precautionary measures can become a standing architecture of control.
Ambiguous benefits: safety, fairness and automation
The more interesting terrain is where benefits and harms are entangled. Automated decision-making is often presented as a way to reduce human bias. If a hiring algorithm is trained to ignore gender, race or family name, then, in theory, it can make fairer decisions than a recruiter swayed by conscious or unconscious prejudice. A judge, given access to a risk assessment tool, might use it to calibrate bail decisions more consistently across cases. An insurer, using telemetry rather than crude demographic data, might price risk more accurately.
The difficulty is that models learn from historical data that already embed structural inequities. If access to credit, employment and education has been skewed for decades along lines of class, ethnicity, caste or gender, then training a model on those outcomes risks reproducing them at high speed and at scale. The bias becomes less visible, precisely because it’s encoded in mathematics and hidden within proprietary code.
So the question becomes: can we construct automated systems that genuinely widen opportunity rather than narrowing it? Can model training, dataset curation, evaluation and oversight be organised and managed in such a way that the historical injustices we know about are not simply baked into a more rigid order? Some experiments in algorithmic auditing, participatory design and regulatory oversight offer partial answers, but we don’t yet have comprehensive proof that these approaches can keep pace with the rate at which new systems are being deployed.
A similar ambiguity surrounds arguments about safety. Data‑driven monitoring can make some forms of harm less likely. But it can also create new forms of vulnerability. A society that depends heavily on centralised digital credentials and networked sensors for public health or policing becomes acutely exposed to technical failures and cyberattacks. The more we rely on automated filters to detect disinformation or extremism, the more we entrust the boundaries of acceptable speech to systems whose judgments are informed by opaque labels and ever‑shifting political pressures.
In short, the same infrastructures that promise to protect us can, under different conditions or in less benign hands, expose us to novel dangers. To recognise this duality is not a counsel of despair. It’s a reminder that the design, governance and cultural framing of these systems matter at least as much as the underlying technologies.
A global perspective
It would be a mistake to treat “datafication” as primarily a Western drama that might later be exported to the rest of the world. The reality is much more complicated.
In some regions with weak administrative capacity, digital systems are being adopted rapidly because they appear to offer a leap over outdated bureaucracies. In others with long histories of centralised rule, data infrastructures sit comfortably within existing traditions of state authority. Elsewhere, fragmented and often informal identities are being pulled into more formal registers under the twin pressures of global finance and counter‑terrorism rules. Everywhere, global platforms mediate communication and commerce in ways that cut across national boundaries while still being shaped by the legal and cultural assumptions of the jurisdictions in which they are headquartered.
At the same time, communities with strong histories of mutual aid, oral culture and decentralised authority bring different expectations to these arrangements. A system that looks “efficient” from the vantage point of a central planner may, on the ground, erode local practices of trust, reciprocity and conflict resolution. Conversely, digital tools can be appropriated by local actors to document abuses, organise resistance or develop alternative economic arrangements.
From a planetary perspective, data infrastructures are not only social systems. They are also material systems. Data centres consume enormous amounts of energy and water. Device production depends on mining, manufacturing and supply chains that reach into some of the world’s most fragile ecosystems and exploited labour markets. Satellites, undersea cables and wireless towers reshape landscapes and seascapes. To speak of “cloud” services is to use a metaphor that hides the metals, plastics, concrete and carbon that make them possible.
This materiality matters because it anchors our shiny digital stories in ecological limits. A future in which every interaction, movement and transaction for eight or nine billion people is logged, analysed and stored indefinitely is not only a philosophical question about autonomy. More fundamentally, it’s a practical question about energy, resources and waste. Can we imagine data systems that are not only fairer, but lighter – calibrated to what is genuinely necessary rather than to an insatiable appetite for prediction and profit?
The open questions
If we set aside overheated narratives about inevitable dystopia, we’re still left with a suite of unsettling questions:
· At what point does the accumulation of data and the automation of judgment tip a society into a condition where meaningful anonymity is no longer possible for anyone, anywhere? Are there thresholds of surveillance beyond which the formal presence of elections, courts and human rights charters becomes largely symbolic?
Can we build robust institutional safeguards – legal, technical and cultural – that prevent function creep, ensuring that tools introduced for narrow purposes don’t gradually expand into general purpose control mechanisms? If so, what evidence do we have from jurisdictions that have managed this balancing act over time rather than during a single political cycle?
What alternative models of identity and reputation might allow people to access services and build trust across distance without exposing their entire lives to continuous inspection? Are there viable large‑scale examples of such models in practice?
How can societies respect a plurality of cultural mindsets in their approach to data and automation? Is it possible to have global interoperability for travel, trade and communication without converging on a single, homogenised template for governance and surveillance?
Obviously, these questions reach beyond the usual debates about privacy policies and technical standards. They go to the heart of how societies imagine themselves and what they regard as a human being. In a worldview where a person is essentially a consumer, a data subject or a risk vector, pervasive measurement looks rational. In a worldview where a person is also bearer of intrinsic worth, mystery and unpredictability, the compulsion to capture and predict every behaviour looks more like a pathological dilemma.
Pros, cons, and the space for agency
If we strip the issue to its bare bones, the emerging datafied order offers at least four interwoven advantages: it can enhance efficiency, extend inclusion, personalise services and mitigate certain risks. At the same time, it threatens to entrench opacity, deepen dependency, amplify existing inequities and normalise a culture in which quantitative profiles outweigh lived relationships.
The crux is that these tendencies are not mechanically determined by the technology itself. They are shaped by worldviews, expressed in world‑systems, and interpreted through the mindsets of people who build, regulate, use and resist them. That means outcomes are not fully preordained, even if certain trajectories currently dominate.
The temptation, especially among those who see the dangers clearly, is to indulge in prophetic despair – to declare that “it is already too late,” that the surveillance architecture is complete, that our only options are minor acts of personal withdrawal. This story may satisfy a certain hunger for tragic clarity, but it underestimates both the contingency of large systems and the inventiveness of human cultures.
The less dramatic but more demanding stance is to treat the present as a period of intense contention in which multiple futures are still coexisting. The same facial recognition system that enables automated boarding at airports can, under different political conditions, be banned outright or restricted to narrow, audited uses. The same digital ID platform that currently functions as a passport to welfare benefits could, with different design choices and accountability mechanisms, become a tool of solidarity rather than exclusion. The same machine‑learning methods used to nudge consumption can also be redeployed to identify and reduce structural inequities, if that becomes the explicit goal.
Whether such reorientations are realistic at scale is itself an empirical question. Yet treating them as impossible becomes, in effect, a self‑fulfilling prophecy that leaves the field to those most eager to exploit the current trajectories.
Towards a different imagination
If there is one conviction that emerges from examining these developments across societies, it is that we must not address them with yesterday’s political categories. Framing the issue as a clash between “privacy” and “security”, or as a quarrel between “the state” and “the market”, leaves the underlying worldview untouched.
A more useful starting point might be to ask: What forms of knowledge and control are appropriate in a world where we are, for the first time, technically capable of tracking most human and non‑human activity in real time? At what point does the will to know everything become destructive of the very freedoms and relationships it claims to safeguard?
Another line of inquiry might be: How do different civilisations – with their varying spiritual, philosophical and legal traditions – imagine the limits of legitimate scrutiny? Are there resources within religious, indigenous or humanist worldviews that could help articulate a global ethos of restraint in data collection and automated judgment?
And perhaps the most practical question: What would it mean, in design terms, to build data systems that are explicitly biased towards human dignity rather than towards control and extraction? That ay well involve default data minimisation, robust rights to explanation and contestation, locally governed data commons, or architectures that keep sensitive information as close as possible to those whom it concerns. It might also involve accepting slower, less “efficient” processes in domains where speed and total visibility carry unacceptable moral costs.
None of these directions offers an easy blueprint. But if we don’t begin to formulate such alternatives we’ll find ourselves living inside someone else’s answers – answers written into code, contracts and infrastructures that could persist long after their original designers have forgotten why they made certain choices.
The real issue of concern is not whether datafied systems are inherently good or bad. It’s whether we’re prepared to interrogate the worldviews that currently drive their expansion, to expose their hidden assumptions, and to admit that other societal logics are possible. Only then can discussions of “pros and cons” rise above technical checklists and address what’s truly at stake for every inhabitant of this planet.