The defining political struggle of the computational age may not be over privacy, data ownership or even algorithmic fairness. These questions matter, but they begin too late. They assume that life has already been rendered observable, recorded and available for institutional use. The prior question is whether life should ever be captured in this way at all.
The deeper danger is not simply that we are being watched. It is that life is being progressively reconstructed as something that can be completely seen, classified, transacted and acted upon. The world is being redesigned as an executable environment: every movement a potential signal, every relationship a potential exchange, every action a potential record, every record a potential judgment, and every judgment capable of becoming an automatically enforced consequence.
This is more than digitisation. It is the convergence of the transactification of life, the rule-ification of life and the continuous surveillance of life. Transactification converts the continuous texture of existence into discrete, attributable events carrying value, rights, obligations, permissions or liabilities. Rule-ification converts situational judgment into predetermined categories, thresholds and protocols. Surveillance supplies the permanent field of visibility through which conduct can be detected and attributed. Automation then connects what the system sees to what the system does.
The result is a new architecture of confinement. Its walls are not primarily physical. They are built from compulsory legibility, persistent attribution and the elimination of the space between action and institutional consequence. Freedom is no longer merely constrained by rules. It is increasingly reformatted as movement through a landscape of pre-authorised possibilities.
When the rule begins to watch
Rules are not new, and neither is the institutional desire to make populations legible. States, markets and bureaucracies have always classified people, standardised conduct and translated complex lives into administrable categories. The computational shift is nevertheless qualitative. It combines unprecedented scale and granularity with persistent memory, cross-context linkage, predictive inference and the capacity to act in real time.
The rule no longer needs to wait for a recognisable moment of application. It can be embedded within the environment itself. A payment is blocked, access is altered, a price changes, a permission disappears or a risk score silently shifts. The decision becomes infrastructural. It no longer presents itself as a judgment made by an identifiable actor at a particular moment; it appears simply as what the system permits.
This collapses what might be called the social interval: the distance between a norm, the observation of an event, its interpretation, a judgment and its consequence. That interval has never been innocent. It has also contained prejudice, patronage, inconsistency and arbitrary power. The answer is not to romanticise human discretion. But the interval has also been where context could enter, where an explanation could be heard, where a rule could be challenged, where mercy and amendment were possible, and where the legitimacy of the system itself could become a political question.
When observation, classification and enforcement are fused, the interval does not merely become faster. It begins to disappear. The situation is classified before it can be interpreted. The exception appears as an anomaly; the anomaly becomes a risk; the risk becomes grounds for restriction. Appeal, if it exists, arrives after the system has already acted.
When representation becomes authority
Every institution requires partial representations of the world. The danger begins when those representations cease to be treated as provisional instruments and acquire authority over the realities they describe.
The identity record becomes the person. The credit history becomes trustworthiness. The behavioural pattern becomes intention. The performance profile becomes capability. The risk score becomes the future. A representation produced for one purpose travels into another context, joins other records and gradually hardens into an administrative double: an institutional account of who someone is and what they may be allowed to become. When the record determines access, description becomes destiny. The representation begins to act before the person can appear.
The person is then required to coincide with the system’s image of them. What cannot be seen is treated as irrelevant. What cannot be classified is treated as suspicious. What cannot be valued is treated as valueless. What cannot be processed is treated as if it does not exist.
This is not only an informational problem. It is an ontological one. Institutions acquire the power to decide what kinds of persons, relationships, contributions and futures are recognisable. They do not merely administer a world that exists independently of them; through their classifications and permissions, they begin to constitute the world in which action is possible.
The asymmetry is profound. People become increasingly transparent to institutions while the objectives, models and categories through which institutions see them remain obscure. The governed are rendered visible; the governing system recedes into infrastructure. Capture is completed when the representation can act upon the represented while remaining unavailable to challenge.
Existential freedom
Freedom cannot be reduced to fair processing within this system. A person is not free merely because they are represented accurately, classified consistently and offered a formal right of appeal. These protections matter, but they remain protections inside the architecture of capture.
Existential freedom begins from a more radical proposition: no representation can exhaust the being it represents. A living person is always more than their history, identity, behaviour, affiliations, transactions or predicted future. A living community is more than its demographic profile. A forest is more than its carbon account. A relationship is more than its ledger of exchanges.
To be free is to retain the capacity to exceed every institutional account of oneself. It is to contradict one’s previous behaviour, to refuse an assigned identity, to form relationships that have not been authorised, to act before one is fully legible and to initiate possibilities that could not have been inferred from the existing order. Permission selects among possibilities already defined. Freedom includes the power to generate possibilities that have not yet been defined.
Existential freedom is the surplus of the possible over the permitted. It is the gap between a living being and every model that attempts to describe it. Non-capturability is the preservation of that gap. A society capable of recognising only what it has already classified closes its own capacity to become otherwise.
This does not imply a solitary or self-sovereign individual standing outside all relationships. We are constituted through dependencies, histories, ecologies and other beings. Precisely because life is relational, no singular account can contain it. Every person is situated within relations that are multiple, changing and only partly knowable. Non-capturability protects this relational excess from being collapsed into a definitive institutional identity.
Beyond privacy and opacity
Privacy is necessary, but it is insufficient. Privacy generally assumes that information has already been captured and asks who may access or use it. Data ownership asks who controls the representation once it exists. Both intervene after the decisive transformation has occurred: life has already been rendered into data.
Opacity goes further. It protects the right not to be made intelligible through categories imposed by another; the right not to explain oneself continuously or submit every dimension of life to institutional interpretation. Yet non-capturability goes further still. It requires that no institution possess the technical capacity, legal standing or political authority to assemble a complete and actionable account of a life.
Non-capturability is therefore not simply secrecy. It is a constitutional limit on institutional knowledge. It preserves an irreducible incompleteness between a person and the records held about them. The claim is not merely that human beings are too complex to understand. It is that even accurate knowledge must not acquire unlimited authority over the being known. Some forms of institutional ignorance are not administrative failures but conditions of freedom.
An institution may know enough for a defined and legitimate purpose without acquiring the right to know everything else. A record may establish responsibility for a particular act without becoming a permanent account of the actor. A system may remember what is necessary for redress without allowing every past event to retain operational force over every future context. The aim is not the absence of knowledge, but the refusal of totalisable knowledge.
The right to become otherwise
The deepest consequence of permanent capture is the foreclosure of existential discontinuity: the capacity to break with what one has been and become otherwise.
Predictive systems transform the past from a history into a claim upon the future. Previous behaviour determines what opportunities are shown, what risks are inferred, what conditions are offered and what permissions are withheld. The future is progressively allocated according to the probability that it will resemble the past.
But freedom depends upon the possibility that it will not.
A free society must preserve the capacity to change without having every previous self continuously reintroduced as evidence. It must allow identities to be abandoned, beliefs to be revised, patterns to be broken and experiments to fail without becoming permanent reputational liabilities. It must allow people to be inconsistent with their data.
This is a right not merely to be forgotten, but to cease being governed by the perpetual activation of the past. Forgetting and memory are both necessary for justice. Institutions may need durable memories of their own decisions, public harms and unresolved liabilities. But people must not be permanently imprisoned by every trace they leave. Power should be required to remember its actions; living beings must retain the possibility of renewal.
The relational system is not yet the opposite
It is tempting to imagine that the alternative to a rules-oriented world is a relational one: situational, negotiated, adaptive and capable of recognising plural forms of value. This is partly true. Relationships hold forms of context that universal rules cannot. Commitments can be revised. Circumstances can matter. Repair can replace automatic punishment.
But a relational computational system can remain a system of total capture. Indeed, it may demand more information than a fixed-rule system. To personalise every agreement according to identity, history, circumstance and relationship, it must know more about everyone. A system of dynamic, situational negotiation could become a world of personalised law, differentiated permissions and continuously adjusted terms.
Nor does replacing one ledger with many necessarily solve the problem. Interoperable ledgers of finance, care, contribution, reputation, ecological effect and social value can be assembled into a still more complete account of a person or community. Plural measurement can become total measurement. A richer vocabulary of value can extend transactification into domains it had not previously reached.
The opposite of capture is therefore not a better ledger. It is a bounded ledger and a world that always exceeds it. Relationality must remain more than the information a system holds about relationships. Care must remain possible without certification. Contribution must remain possible without measurement. Association must remain possible without attribution. The ledger must be partial, purpose-bound and incapable of becoming the world.
A non-totalisable order
The alternative is neither rulelessness nor universal invisibility. Rules can secure standing, protect people from arbitrary relational power and make rights claimable without dependence upon favour. Records can reveal exploitation, establish liability and preserve evidence of institutional harm. Automation can expand freedom when it reliably provides an entitlement or removes an unnecessary gatekeeper.
The boundary lies elsewhere. Rules may govern defined actions without claiming to define the whole actor. Records may establish specific obligations without becoming portable judgments of personhood. Computation may coordinate parts of life without claiming exhaustive knowledge of life. Automation may assist provision without acquiring unreviewable authority over a person’s fundamental standing or possible futures.
A non-totalisable order would be organised around deliberately bounded knowledge. Representations would remain partial and tied to their purposes. Different contexts would not automatically compose into a universal identity. Inferences would not acquire the status of facts. Consequential judgments would remain visible and contestable. Some records would expire; some actions would remain anonymous; some spaces would contain no sensing at all. There would be legitimate domains in which no transaction was produced and no institutional memory created.
This is not an argument for impunity. Non-capturability must be directional. The greater an actor’s power to impose consequences upon others, the greater its obligation to be visible, accountable and intelligible. Institutions should be more legible to those they govern than those people are to the institutions. Systems should retain records of their decisions and the harms they produce. But the person should retain domains of existence that the system has neither the need nor the standing to know.
The principle is not maximum opacity. It is minimum necessary capture and maximum retained possibility.
The politics of the unfinished
The expansion of executable life is not technologically inevitable. It is produced by institutions seeking administrative scale, risk reduction, behavioural prediction, liability management, security, efficiency and control. Greater legibility makes finer classification possible. Finer classification generates demand for more sensing. More sensing enables automatic enforcement. Automatic enforcement makes still more granular rule-ification appear feasible.
This cycle will accelerate as agents, sensors, contracts and ledgers become capable of acting continuously and autonomously. The danger is not only that every organisation becomes agentic. It is that every object, relationship and exchange becomes an agentic point of enforcement: observing, interpreting and modifying the field around it. Governance could become ubiquitous without ever appearing as government.
The political question of the computational age is therefore not simply who owns the data, who writes the algorithm or whether the model is accurate. It is whether any system should be permitted to make a person, community or living world fully available for continuous institutional action.
We will need more than privacy rights. We will need rights against cross-context composition, against permanent predictive identity and against the conversion of inference into destiny. We will need protected spaces for anonymous presence, unmeasured contribution, unrecorded association and action that does not require prior legibility. We will need institutions capable not only of knowing, but of refusing to know.
Above all, we will need to defend the proposition that life is not an administrative object awaiting complete representation. It is unfinished, emergent and capable of beginning again.
Life must remain uncapturable
The danger is not a world containing rules, records or computation. It is a world in which rules see continuously, records travel without limit, models become identities and classifications act automatically upon the futures they claim merely to predict.
A free society cannot be one in which every legitimate possibility must first be visible to the system. It must preserve the right to initiate what cannot yet be classified, to enter relations that are not transactions, to contribute without being measured, to change without being permanently pursued by one’s previous form, and to remain more than the sum of what institutions know.
No system should be able to assemble a total account of a life. No representation should acquire sovereign authority over the being it represents. No past should be permitted to close the possibility of becoming otherwise.
We must build institutions that know enough to coordinate, remember enough to be accountable and remain ignorant enough for freedom.
For life to remain living, it must remain more than can be seen, known, valued, transacted or executed. It must remain uncapturable.
I strongly agree with the argument against the capture and quantification of life. Industrial society already measured individuals extensively, but the information age, social media, algorithms, and now AI have pushed this much further. More and more of ordinary life is being converted into data, scores, profiles, predictions, and measurable signals. In many ways, this is one of the deepest pathologies of social media.
What strikes me, however, is the asymmetry. While we have become extraordinarily good at quantifying individual life, we have done remarkably little to quantify power itself. We measure wealth, work, behavior, attention, risk, and increasingly even intelligence, yet political power remains described through remarkably old and often opaque categories. Sovereignty, in particular, still carries structural residues of divine authority: the highest power remains difficult to measure precisely, compare, trace, or subject to anything above itself.
Perhaps we need to reverse the direction of quantification. We should minimize, and wherever possible stop, the quantification of individual life, while directing far more of our analytical capacity toward quantifying power itself: who can impose consequences on whom, through what mechanisms, with what reach, and under what constraints. Life should remain as uncapturable as possible. Power should become as measurable, traceable, and accountable as possible.
The phrase ‘the decision becomes infrastructural’ is the hinge for me. When a permission disappears or a score changes without an identifiable decision-maker, due process is weakened before anyone argues about whether the outcome was accurate.