Discipline, Surveillance and the Carceral

Punishment used to be done to the body, in public, spectacularly. Now it is done to the person, in private, quietly — and the change is not simply humanitarian progress. It is a transformation in how power works, and it is the subject of one of the most influential and most over-applied arguments in modern social science.

This lesson sets out that argument, marks where it illuminates and where it overreaches, and turns to the harder question the theory tends to skip: what actually changes behaviour?

Two punishments, a century apart.

Foucault opens his account of the change with two scenes, and the contrast is the whole argument.

The first is an execution in 1757. A man convicted of attempting to kill the king is tortured publicly and at length — the account is deliberately unbearable to read — before an enormous crowd. The punishment is a spectacle. Its point is public: to display the sovereign's power on the body of the condemned, to answer a crime against the king with an overwhelming, visible, bodily vengeance. Power is shown.

The second is a timetable for a young offenders' institution, drawn up about eighty years later. It specifies every minute of the day: when to rise, wash, work, eat, study, pray, sleep. No spectacle. No crowd. No pain inflicted on the body. Just a schedule, applied continuously, in an enclosed institution, to produce a reformed and orderly person.

Between the two scenes, Foucault argues, the entire logic of punishment changed.

The old power punished the body and displayed itself. The new power trains the person and hides itself. It does not want to destroy the offender; it wants to reform him — to reach inside and reorganise his habits, his time, his soul. The scaffold gives way to the timetable, and the timetable is more thorough than the scaffold ever was, because it does not merely punish an act; it remakes a person.

And here is the move that made the argument famous. Foucault's claim is not that this was progress from cruelty to kindness. It is that a new and more effective form of power appeared — one that works not by spectacular force but by continuous, detailed, individualising training, and that escaped the prison to spread through the whole society. The school, the factory, the hospital, the barracks — all adopted the same techniques: the timetable, the enclosure, the ranking, the examination, the continuous observation. The prison is not an exception in a free society. It is the concentrated form of a discipline that runs through all of it — which is why the argument reaches so far, and why it needs careful handling.

What is discipline, as a form of power, and how does it differ from what came before?

Sovereign power — the old kind — is exercised from above, intermittently, spectacularly, on the body. It punishes transgression after the fact, visibly, to deter and to display. It is expensive, occasional, and it treats the population as a mass.

Disciplinary power is different in every respect. It works continuously, not intermittently — it is always on. It works in detail — on the individual, their movements, their time, their habits. It works productively, not just repressively — it does not only forbid; it produces capacities, docility, useful bodies (see 9.3.2 on control that produces effort). And it works by making people visible, because visibility is the mechanism.

The distinction Foucault draws is between power that is seen and power that sees. Sovereign power is seen — the spectacle, the display. Disciplinary power sees — it observes, records, examines, and by observing, controls. The direction of visibility reverses: the powerful used to be visible and the many hidden; now the many are visible and the power that watches is hidden.

And the device that captures this is the one everyone remembers.

The panopticon

The design, the principle, and what it does and does not explain.

The panopticon was a prison design proposed by Jeremy Bentham: a ring of cells around a central tower, arranged so that a single watcher in the tower can see into every cell, while the prisoners cannot see into the tower and so cannot tell whether they are being watched at any given moment.

Foucault seized on it not as a building but as a diagram of power.

The essential effect is this: because the prisoner cannot tell when he is being watched, he must behave as if he is being watched at all times. The surveillance becomes continuous in its effect even though it is intermittent in fact — and, crucially, the prisoner ends up policing himself. The watcher in the tower need not even be there. The mechanism works by being internalised: the observed takes over the work of observation, and control becomes self-control.

This is the deep point, and it is genuinely powerful. The most efficient control is not the one that watches you but the one you have so internalised that you watch yourself — which is normative control (see 9.3.2), the disciplined worker who needs no supervisor, the self-monitoring subject who has taken the standard inside. The panopticon is a machine for producing self-discipline, and self-discipline is cheaper, more thorough and more reliable than any external force.

And now the honest assessment, because this argument is applied far more widely than it can bear.

What it illuminates: the shift from spectacular to continuous power, the productive character of discipline, the internalisation of control, and the spread of institutional techniques (timetable, ranking, examination, record) across schools, workplaces and clinics. These are real and important, and they recur throughout Part 9.

What it overreaches on, and the criticisms are serious.

The historical claims are shakier than the theory. Foucault's account of the transition is selective; the old spectacular punishment and the new disciplinary one overlapped for a long time; and the reform was driven partly by exactly the humanitarian motives he downplays. The argument is a brilliant ideal-type, not an accurate history (see 4.3.8 — and it should be read as one).

The self-discipline is not total. People resist, evade, game and subvert discipline constantly (see 9.3.1 on the informal system, 9.3.3 on gaming the algorithm). The panoptic subject who has perfectly internalised control is a limit case, not a description — and treating it as a description is the overreach the theory invites.

And the argument has trouble with agency and with liberation. If power is everywhere, productive, and internalised, it becomes hard to say what resistance is or how anything improves — which is the standing criticism of the whole framework (see 5.7.2). A theory in which everything is discipline cannot distinguish a prison from a school from a hospital in the ways that matter to the people in them.

The disciplined reading, then: a profound account of a real form of power, over-extended by its author and far more by his followers, to be used as a lens rather than a law.

Surveillance since the panopticon

The concept updated, because the panopticon is now a poor fit.

The panopticon assumes a central watcher, an enclosed space, and a subject who does not know when they are watched. Contemporary surveillance fits none of these well, and the field has developed the concept accordingly.

Surveillance is now dispersed, not central. There is no single tower. Data is collected by many actors — states, employers, platforms, retailers, insurers — and combined in ways no single watcher controls (see 9.6.2). The metaphor of one eye in a tower fails; the reality is many eyes, many records, recombined.

It is now largely about data, not observation. Modern surveillance is not primarily someone watching you; it is the continuous capture of traces — transactions, locations, clicks, communications — and their analysis (see 9.3.3 on algorithmic management, 9.6.2 on platform data). The body in the cell has become the data trail, and the trail is more revealing and more permanent than any watcher.

It sorts rather than disciplines. A key updating: much surveillance is not aimed at reforming individuals but at sorting populations — into risk categories, credit scores, eligibility groups, targeting segments — and treating them differently (see 8.4.1 on categories that allocate, 9.5.2 on risk stratification). This "social sorting" is the dominant function, and it is closer to 8.1.1's categorical inequality than to the reform of a soul.

And it is often not experienced as control at all. People participate willingly — carrying tracking devices, posting their locations, accepting monitoring for convenience — which the panopticon model, built on an unwilling prisoner, does not capture. Some of it is coerced; a great deal is traded for convenience, and the trade is real, unequal, and mostly invisible to the person making it.

The honest position: the panoptic insight — that visibility induces self-regulation — survives and matters, but the panopticon as a model of contemporary surveillance has been substantially superseded by accounts of dispersed, data-driven, sorting surveillance that people frequently join voluntarily.

And the question the theory skips: what actually changes behaviour?

The discipline-and-surveillance tradition is strong on the form of power and weak on the evidence of what works — and a course committed to 7.9.1 has to supply the evidence the theory brackets.

On deterrence, the best evidence gives a consistent and specific answer. The certainty of being caught deters; the severity of punishment does much less. Raising the probability of detection reduces offending; raising sentence length, at the margins that policy usually operates on, does little (see 10.2.3 on mass incarceration failing to track crime). This is one of the better-established findings in criminology, and it contradicts the political instinct toward ever-harsher sentences. The panoptic mechanism — the sense of being watched — is closer to certainty than to severity, which is why it captures something real.

On surveillance technologies specifically, the evidence is mixed and more modest than either enthusiasts or critics claim. Closed-circuit cameras reduce some crimes (notably property crime in car parks) and not others (little effect on violence), with effects that are real, modest, and often displaced rather than prevented — crime moves rather than stops (see 7.4.3 on displacement). The technology is neither the panacea its advocates claim nor the totalising control its critics fear — it is a modest, uneven tool, which is the boring truth the dramatic theory obscures.

On what changes behaviour more broadly, the evidence points away from spectacular control and toward the mundane: certainty of detection, legitimacy of the authority (people obey rules they see as fairly made and fairly enforced — see 9.7.1), and the concrete removal of opportunity (the security improvements of 10.2.2, which prevented crime by making it harder rather than by disciplining anyone). None of these is the internalised soul-training of the theory, and all of them have better evidence.

Because the discipline-and-surveillance argument is the most-cited framework in this area and the most likely to be misapplied — and holding its insight apart from its overreach is the skill.

What to keep.

The shift from spectacular to continuous power is real, and it runs through every institution of Part 9 — the timetable, the examination, the record, the ranking are disciplinary techniques, and recognising them is genuinely illuminating.

The internalisation of control is real and important — the most efficient power is the one you exercise on yourself, and this is normative control, the self-monitoring subject, the panoptic effect. It is a genuine mechanism, visible in the disciplined worker and the self-surveilling social media user.

And the reversal of visibility is real — the many are now visible to a power that is itself hidden, and this is truer of data surveillance than it was of the tower.

What to discard, or hold loosely.

The theory as history — it is an ideal-type, selective, and downplays the humanitarian reform it explains away.

The theory as a total account — power is not everywhere and self-discipline is not complete; people resist, evade and trade, and a framework that cannot see this is missing most of what actually happens.

And the theory as a substitute for evidence — it tells you the form of control, not what works, and the evidence on what works (certainty over severity, legitimacy, opportunity-removal) is mundane, well-established, and mostly outside the theory.

Three questions to carry.

Is this discipline or sorting? Much modern surveillance does not reform individuals; it categorises populations and treats them differently — which is closer to 8.1.1 than to the panopticon.

Is the control internalised or external? The most powerful control is the self-monitoring the subject does for you — and where you find it, you have found the panoptic mechanism, whatever the technology.

And what does the evidence say changes behaviour? Not severity but certainty; not spectacle but legitimacy and opportunity. The dramatic theory and the boring evidence point in different directions, and the evidence wins.

One closing observation. Foucault's two scenes — the scaffold and the timetable — are a genuinely great piece of sociological writing, and the argument they open changed the field. But the timetable's power over us is not that it has trained our souls into perfect docility. It is that we have, mostly, come to watch ourselves — and that even this is incomplete, resisted, gamed and traded, in ways the theory's grandeur obscures. The most disciplined thing a reader of this lesson can do is admire the argument and still ask it for evidence — which is the whole of what Part 7 was for, applied to the most seductive framework in the Part.

Foucault opens with two punishments a century apart: a public execution in 1757 that displays the sovereign's power on the body, and a young-offenders' timetable that hides itself and trains the person. Between them the logic of punishment changed — from punishing the body spectacularly to reforming the soul continuously — and the disciplinary techniques escaped the prison into the school, factory, hospital and barracks.

Sovereign power is seen (spectacular, intermittent, on the body); disciplinary power sees (continuous, detailed, productive, individualising) — and the direction of visibility reverses, the many becoming visible to a hidden power.

The panopticon — cells around a tower where the watched cannot tell when they are watched — is a diagram of power: because you cannot tell when you are observed, you behave as if always observed, and end up policing yourself. The watcher need not be there; control is internalised, which is cheaper and more thorough than force.

But the argument overreaches: the history is selective and downplays humanitarian reform (it is an ideal-type, not history); self-discipline is never total (people resist, evade, game); and a theory in which power is everywhere struggles to say what resistance or improvement is.

Surveillance since the panopticon has outgrown the model: it is dispersed not central, data-driven not observational, sorting populations rather than disciplining souls (social sorting, closer to 8.1.1), and frequently joined voluntarily for convenience. The panoptic insight survives; the panopticon as a model of contemporary surveillance is superseded.

And the theory skips the evidence on what works. Certainty of detection deters; severity does much less — one of criminology's better-established findings, contradicting the instinct toward harsher sentences. Surveillance cameras have modest, uneven, often-displaced effects. And behaviour is changed more by certainty, legitimacy and opportunity-removal than by internalised soul-training — mundane findings, better evidenced, mostly outside the theory.

Keep the shift to continuous power, the internalisation of control, and the reversal of visibility; discard the theory as history, as a total account, and as a substitute for evidence.

Sovereign power — spectacular, intermittent punishment of the body to display authority.

Disciplinary power — continuous, detailed, productive training of the individual.

Discipline — the techniques (timetable, enclosure, ranking, examination, record) that produce docile, useful bodies.

Panopticon — the ring-and-tower design; a diagram of control through uncertain, internalised visibility.

Internalisation of control — the observed taking over the work of observation; self-discipline.

Reversal of visibility — the many made visible to a hidden watching power.

Social sorting — surveillance that categorises populations and treats them differently, rather than reforming individuals.

Dispersed surveillance — data-driven monitoring by many actors, superseding the central-tower model.

Certainty versus severity — the finding that the probability of detection deters and the harshness of punishment does much less.

Displacement — crime moved rather than prevented by a control measure.

Legitimacy — the compliance produced by rules seen as fairly made and enforced.

One — find the timetable. Take one institution you know and list its disciplinary techniques — the schedule, the ranking, the record, the examination. They are the discipline of the lesson, in plain sight.

Two — locate the internalisation. Identify one way you monitor your own behaviour as if watched. That is the panoptic effect, with no watcher required.

Three — sort versus discipline. Take one surveillance system and ask whether it reforms individuals or sorts populations. Most modern ones sort.

Four — test certainty against severity. For one rule, ask what would change behaviour more — a higher chance of being caught or a harsher penalty. The evidence favours the first; check your intuition against it.

Five — ask the theory for evidence. Take one dramatic claim about surveillance or discipline and ask what the evidence says actually changes behaviour. The gap between the theory and the evidence is the lesson.