The Labour Process and Control

An employer buys a person's time. What the employer wants is their effort, and effort cannot be bought — only elicited.

That gap between purchased time and delivered effort is the central problem of managing work, and everything in this lesson is an attempt to solve it: by measuring, by designing the machinery, by writing rules, by building careers, by managing feelings, and by getting workers to want to.

And it starts with the most famous story in the history of management, which is largely untrue.

The pig iron story, and what the archives showed.

In the founding text of scientific management, Frederick Taylor described a demonstration.

At a steel works, gangs of labourers were loading pig iron — ninety-two-pound blocks — onto railway cars, at an average of about twelve and a half long tons per man per day.

Taylor's team analysed the task: the exact weight, the exact walking distance, the proportion of the day a man could be under load, the rest intervals required. They concluded that a suitable man, properly directed, could load around forty-seven tons.

He then described selecting a worker he called Schmidt — a man he characterised as strong, thrifty, and mentally suited to the work in a way Taylor described in terms that have not aged — and instructing him minutely: pick up, walk, put down, sit, rest, on command, all day.

Schmidt loaded forty-seven and a half tons, earned a 60 per cent pay rise, and the demonstration became the foundational anecdote of modern management.

Historians went to the archives.

What they found does not support the account. The original records indicate that the reported rates and the sequence of events differ substantially from Taylor's published version. The task was not conducted as a single clean demonstration. The men who achieved the very high rates were a small selected group, several of whom did not sustain it, and a number of workers were dismissed or left during the process. Taylor's own contemporaneous documents do not match his later published narrative, and elements of the story appear to have been assembled after the fact.

None of which means his ideas were unimportant. They were among the most consequential of the twentieth century.

It means the founding demonstration of the doctrine that work should be governed by measurement was itself not accurately measured — which is 7.9.1's lesson in the least expected place, and it is worth knowing before assessing anything that follows.

And the more interesting point about Taylorism is not the stopwatch anyway.

The stopwatch was a means. The object was the knowledge.

Taylor's programme had four parts, and the first is the one that matters.

Gather up the knowledge held by workers — the accumulated tricks, judgements and rules of thumb by which a job is actually done — and reduce it to written rules and formulae.

Separate conception from execution. All planning moves to a planning department; the worker executes.

Use the resulting monopoly of knowledge to specify each step of the work.

And attach individual incentive pay to compliance.

Read the first two together and the point becomes clear. Before Taylorism, the person who knew how to do the job was the person doing it — which is 9.3.1's zone of uncertainty, and it is a source of power. A skilled worker who knows how long a task should take controls the pace, and management cannot dispute it because management does not know.

Taylorism is a transfer of knowledge from the worker to the firm, and everything else — the timing, the specification, the incentive — follows from it.

Which means the question this whole lesson turns on is: who knows how the work is done? And every control strategy below is an answer to it.

The deskilling thesis and its problems

Braverman's argument, which revived the field.

The claim: the tendency of capitalist production is to degrade work — to separate conception from execution systematically, so that the knowledge required to do a job is progressively removed from the person doing it, embedded in machinery, procedure and management, and the labour that remains is cheapened.

Two consequences follow. Labour power becomes cheaper to buy, because less skill must be paid for. And control passes to the employer, because the worker no longer possesses anything the employer needs.

The argument's strength is that it identifies a real mechanism with a real logic, and it explains a great deal about the history of manufacturing, clerical work and, latterly, professional work.

Its problems are substantial and should be stated in full.

It assumes a prior golden age of craft skill. The autonomous craftsman was a minority of the workforce even at his peak; most work in the nineteenth century was unskilled, brutal and closely supervised. Comparing modern work to the craftsman is 9.2.1's baseline problem.

Skill is very hard to measure and the measures disagree. Job analysis, formal qualification requirements, self-reported discretion and task complexity give different answers about the same jobs, and 7.2.2's problem is acute here — partly because "skill" is itself a socially negotiated category, as the devaluation evidence in 8.5.1 shows: what counts as skilled work is partly a function of who does it.

And on most aggregate measures, skill requirements have risen rather than fallen. Qualification requirements, task complexity and discretion measures have all trended upward across recent decades — though 9.1.4's credential inflation means rising qualification requirements are weak evidence of rising skill.

But the strongest correction is that the trend is not one-directional at all.

The better-supported finding is polarisation. What is automatable is not "unskilled work" but routine work — tasks that can be specified as a procedure, whether manual (assembling, sorting) or cognitive (calculating, filing, checking).

What resists automation is non-routine work at both ends: non-routine cognitive tasks (analysis, persuasion, judgement, management) and non-routine manual tasks (care, cleaning, food preparation, driving in unpredictable environments).

And routine work was concentrated in the middle of the wage distribution. So the employment structure has hollowed out — growth at the top and the bottom, contraction in the middle — which is a much better description of what has happened than either deskilling or upskilling, and which connects directly to 8.8.1's account of wage inequality.

Braverman's second acknowledged limitation was that he excluded worker subjectivity deliberately — he said so — and analysed the labour process as if workers simply had things done to them. Which is what the most important response to him addressed.

Why do workers work as hard as they do?

The question reversed, and the answer that reorganised the field.

Burawoy took a job in a Chicago machine shop — the same plant an earlier researcher had studied thirty years before, which made it a rare longitudinal ethnographic comparison.

And he inverted the standard question. Labour process theory asks why management must coerce effort. He asked: given that nobody is standing over them, why do these men work as hard as they do?

The answer was the game.

Machine operators were paid a base rate plus a bonus for exceeding a production quota. Around this, the shop floor had developed an elaborate practice called making out — hitting the quota, with its own tactics, its own vocabulary, its own status hierarchy, its own informal rules about which jobs were "gravy" and which were "stinkers", and its own etiquette about how much to exceed the rate by.

Making out was absorbing. It generated genuine satisfaction, real skill, informal competition, and something to talk about. Men who made out were respected. Men who could not were pitied.

And Burawoy's argument is the crucial move. By playing the game, workers were not merely producing output. They were consenting to the rules of the game — and in doing so, consenting to the conditions that made the game the only thing available.

The game constituted the shop floor as a place where the interesting question is "can I make out today?" rather than "why is the rate set here?" Conflict that could have been between workers and management became conflict between operators and the scheduling office, between shifts, between individuals. Consent was manufactured not despite the workers' agency but through it.

Two supporting mechanisms: an internal labour market, in which jobs are filled by promotion from within, which makes individual advancement the route to improvement and gives everyone a stake in the hierarchy; and an internal state — grievance procedures, representation, formal rights — which converts collective conflict into individual cases processed through a machinery.

Why this matters beyond a machine shop. The mechanism generalises to any work organised around a target, a score, a ranking or a personal best. Where the work supplies a game, the game supplies the consent — and the questions the game does not ask do not get asked.

The forms of control

Three types, historically successive and currently simultaneous.

Simple control — direct personal supervision. The owner or foreman watches, instructs and disciplines. Effective in small settings; it does not scale, and it makes the supervisor personally the object of resentment.

Technical control — control embedded in the machinery. The assembly line is the canonical case: the pace is set by the line, not by a person, so the conflict over pace becomes a conflict with an apparently technical fact. The supervisor becomes an enforcer of what the machine requires rather than the author of the demand.

Bureaucratic control — control embedded in rules, job classifications, evaluation criteria and career ladders (see 9.3.1). The most sophisticated, because it does not merely constrain but rewards: it offers a career, and a career gives a worker an interest in the organisation's rules being followed. Discipline becomes self-discipline in the service of promotion.

All three coexist, and which predominates depends on the work, the labour market and the workforce's power.

And a distinction that cuts across all three, and is the most practically useful in the lesson.

Direct control treats labour as a cost to be minimised and specified: close supervision, minimal discretion, easily replaced workers.

Responsible autonomy treats labour as a resource whose commitment must be secured: discretion, status, security, and responsibility for outcomes.

Firms use both, on different workers, in the same building. The strategy chosen tracks how easily the worker can be replaced and how much damage they could do — which means the same organisation typically operates a high-discretion regime for its core professionals and a specified, monitored regime for the people who clean the same building, frequently employed through a contractor (see 9.3.3).

This is not inconsistency. It is the same calculation producing two answers.

When the product is a feeling

Emotional labour, and why it belongs in a lesson about control.

In service work, part of what is sold is the worker's demeanour — and Hochschild's study of flight attendants named what that involves.

Emotional labour is the work of producing a required emotional display, and of managing one's own feelings in order to produce it.

Two forms. Surface acting: displaying an emotion you do not feel. Deep acting: working on yourself to actually feel it — summoning the memory, adopting the frame, reinterpreting the passenger as a frightened child rather than an aggressor.

Organisations specify feeling rules — what a worker should feel as well as display — and train, monitor and evaluate against them.

Three findings matter.

It is work, and it is tiring in a way physical accounts do not capture. The strain is documented in burnout, exhaustion and stress measures, and it is concentrated in roles combining high display requirements with low control over the interaction.

Deep acting has a specific cost. The worker who genuinely produces the required feeling has commercialised a capacity previously their own, and the reported consequence is a kind of estrangement from one's own emotions — difficulty knowing what one actually feels.

And it is unequally required. Emotional display requirements fall disproportionately on women and on lower-paid service workers, are rarely named in job descriptions, and are almost never paid for as a skill — which is 8.5.1's devaluation finding, appearing as an unrecognised job requirement.

A related extension: aesthetic labour — the requirement to look and sound a particular way, recruited for and trained. It converts embodied class markers into job requirements (see 8.3.3 on gatekeeping), and it is even less visible as a criterion than the emotional variety.

And when the thing to be managed is judgement.

Direct control fails where the work cannot be specified — which is 9.3.1's point about street-level discretion, and it applies with equal force to professional and technical work.

So the control strategy shifts to the person rather than the task.

Normative control operates by shaping what employees want. Strong culture, mission, identification, shared language, elaborate socialisation, and the presentation of the organisation as a community rather than an employer.

An ethnography of a high-technology firm documented what this produces. Employees worked extraordinary hours without supervision, monitored themselves against the culture's standards, and experienced their commitment as personal rather than demanded. The organisation did not need to specify the work because it had specified the worker.

And its documented failure mode is the burnout of the committed. The person who fully internalises the culture has no defence against it — there is no boundary to appeal to, because the demand is coming from inside — and the study's most striking material concerns people who had believed completely and were left with nothing to be disappointed in but themselves.

Which produces the paradox worth carrying. Autonomy and surveillance have both increased, and they are not opposites. Discretion over how to do the work, combined with intensive measurement of what the work produced, is the characteristic modern arrangement — and it transfers the burden of coordination onto the worker while retaining the evaluation at the centre.

On surveillance itself, a caution. The image of a total electronic monitoring that leaves no room for anything is overstated, and the ethnographic evidence consistently finds otherwise: workers develop counter-practices, gaming, informal norms about what the metric requires, and shared knowledge about which measures are watched.

Monitoring measures what is measurable, which is rarely the same as what matters — which is Goodhart's law (see 7.2.2) operating as the standing limit on control by metric, and it is why the informal system of 9.3.1 survives every attempt to eliminate it.

Because "who knows how the work is done" predicts almost everything else about a job.

Where the worker knows and the employer does not, the worker has autonomy, discretion, and bargaining power — and the employer's strategy will be responsible autonomy, normative control, and a career.

Where the employer knows and the worker executes, the strategy will be specification, measurement, replaceability and direct control — and the worker's power will come from collective action rather than from individual indispensability.

And the history of management is a sequence of attempts to move jobs from the first category to the second, of which Taylorism was the most explicit and software is the most recent.

Four questions for any job.

Where is the knowledge? In the person, the procedure, or the system.

What form does the control take — a supervisor, a machine, a rule, a career, a culture, or a metric?

Is there a game? If effort is organised around a target, a score or a personal best, ask what question the game prevents from being asked.

And what is being required that is not in the job description? Emotional display, aesthetic conformity, availability, identification. These are job requirements that are not paid for as skills, and they fall unequally.

One closing observation about the pig iron. The doctrine that work should be governed by objective measurement was launched with an account that the archives do not support.

That does not make measurement wrong. It makes it an argument that has to be assessed like any other — and it is a useful reminder that the demand for evidence is frequently made most loudly by people whose own case rests on a story.

An employer buys time and needs effort, and every control strategy is an attempt to close that gap.

The founding demonstration of scientific management — the pig iron story — is not supported by the archival record: the rates and sequence differ from the published account, the high performers were a small selected group, and the narrative appears substantially assembled afterwards. The ideas remain consequential; the demonstration was not measured as claimed.

Taylorism's object was not the stopwatch but the knowledge: gather what workers know, reduce it to rules, separate conception from execution, and specify each step. It is a transfer of the zone of uncertainty from worker to firm.

The deskilling thesis identifies a real mechanism and has four problems: it assumes a golden age of craft that was always a minority; skill is hard to measure and partly a negotiated category, as the devaluation evidence shows; aggregate measures generally show rising requirements, though credential inflation weakens that evidence; and the better-supported finding is polarisation — routine tasks, manual and cognitive, are automatable, non-routine tasks at both ends are not, and routine work was concentrated in the middle of the wage distribution. Braverman also excluded worker subjectivity by his own choice.

Burawoy inverted the question: why do workers work as hard as they do without supervision? The answer is the game — "making out" against a quota, absorbing, skilled, status-conferring — and playing the game constitutes consent to its rules, redirecting conflict sideways and making the rate itself unaskable. Internal labour markets and grievance machinery individualise what remains.

Three forms of control: simple (personal supervision, unscalable), technical (embedded in the line, so pace appears as a technical fact), and bureaucratic (rules and careers, which reward as well as constrain). And cutting across them, direct control against responsible autonomy — the same firm running both, on different workers, in the same building.

Emotional labour — surface and deep acting against organisational feeling rules — is work, is exhausting where display demands are high and control low, produces estrangement from one's own feelings in its deep form, and is disproportionately required of women and lower-paid service workers without being paid as a skill. Aesthetic labour extends this to embodied class markers.

Where judgement cannot be specified, control shifts to the person: normative control specifies the worker rather than the work, and its failure mode is the burnout of those who believed. Autonomy and surveillance have both risen and are not opposites. And monitoring measures what is measurable rather than what matters, which is why the informal system survives.

Labour process — the organisation of work through which purchased time becomes delivered effort.

Separation of conception and execution — moving planning to management and leaving execution to the worker; Taylorism's core.

Deskilling thesis — the claim that capitalist production tends to degrade work and cheapen labour power.

Job polarisation / routine-biased technical change — growth at the top and bottom with contraction in the middle, driven by the automatability of routine tasks.

Making out — the shop-floor game of achieving a production quota.

Manufacturing consent — the production of consent through participation in a game whose rules thereby go unquestioned.

Internal labour market / internal state — promotion from within, and grievance machinery converting collective conflict into individual cases.

Simple / technical / bureaucratic control — supervision by a person, by machinery, by rules and careers.

Direct control / responsible autonomy — specification and replaceability, against discretion and secured commitment.

Emotional labour / surface and deep acting / feeling rules — producing a required display, and the organisational specification of what to feel.

Aesthetic labour — the requirement to look and sound a particular way.

Normative control — managing what employees want rather than what they do.

One — locate the knowledge. For a job you know, ask who could specify the work in writing well enough for a stranger to do it. The answer tells you where the power is.

Two — find the game. Identify a target, score or ranking in your own work. Then write down the question it prevents from being asked.

Three — identify the control form. For three jobs, mark whether control operates through a supervisor, machinery, rules, a career, a culture or a metric. Most will use several.

Four — count the unpaid requirements. For any service job, list the emotional and aesthetic demands. Then check whether any appear in the job description or the pay scale.

Five — check a foundational story. Take any doctrine you have accepted and find its founding demonstration. Then find out whether anyone has checked it.