Work and Measure

There is a place where almost every adult meets the algorithm every day, and it is neither the courtroom nor the hospital: it is the workplace. The badge that records your entrance, the management system that assigns the shifts, the dashboard that displays the indicators, the system that routes case files and measures response times: long before artificial intelligence became a topic of conversation, work had already been, for decades, the most densely measured territory of ordinary life. The chapters of this Part have followed the datum from the subject to the profile through the enterprise's pipelines; this chapter reaches the person who spends a third of their life inside those pipelines: the worker. And it also reaches a phrase the Preface dropped on purpose, like a stone into a pond: the work that devours the hours.

The convergence of the book's three disciplines is, here, literal. The enterprise information system is this chapter's natural home: organised work is a flow of information before it is a flow of goods, and those who design such systems decide - often without knowing it - how much humanity will be able to circulate in the processes they draw. Big data is its fuel: the discipline known as people analytics applies to the workforce the same techniques of segmentation and prediction encountered in the chapter The Customer and the Profile. And wearable devices, born for health and well-being, enter the factory and the office as safety tools and, at the same time, as productivity sensors: the boundary between protection and surveillance runs, once again, through a design choice.

1. The measure as artefact: from the stopwatch to algorithmic management

The measurement of work was not born with computing. It was born, as a discipline, with the scientific management of Frederick Taylor (1911): the stopwatch beside the workman, the gesture broken down into elementary movements, the standard time as the unit of account of performance. The scientific organisation of labour was a revolution in productivity and, at the same time, the origin of a question never settled: what remains of the craft when it is reduced to the sum of its times? A century later, the stopwatch has become software. Workforce management systems assign shifts by optimising coverage; delivery platforms compute routes, expected times and priorities; warehouses measure picks per hour; call centres track the duration, outcome and tone of calls; people analytics tools estimate, from the whole of the data, the probability that an employee will resign, perform, be absent. The contemporary name for this ensemble is algorithmic management: the delegation to automatic systems of functions that used to belong to the shop-floor supervisor - assigning, evaluating, rewarding, reprimanding.

From the engineering standpoint, the pipeline is the one by now familiar to the reader: raw events (clock-ins, transactions, messages, positions), derived variables, aggregate indicators, thresholds that trip. And as always in this series, every link in the chain is a design choice disguised as a technical fact: what to count, over what time window, with what normalisation, with what threshold. A productivity indicator does not photograph the work: it defines it. It decides, silently, what part of human activity counts as value and what part - the colleague helped with a question, the customer listened to beyond the standard time, the error averted that no counter records - simply does not exist for the system.

2. Goodhart's law: when the measure devours the craft

The theory of measurement has known its own fundamental limit for half a century, and it bears the name of the economist who formulated it: Goodhart's law (1975), made proverbial in Marilyn Strathern's formulation (1997): when a measure becomes a target, it ceases to be a good measure. The mechanism is structural, not moral. Every metric is a proxy: an observable stand-in for a quality that is not observable. The number of calls closed stands for the quality of the assistance; lines of code stand for the progress of the project; pieces per hour stand for the workman's contribution. As long as the proxy remains a thermometer, it informs. The moment it becomes a target - tied to bonuses, sanctions, rankings - people stop optimising quality and begin, rationally, to optimise the number. The proxy comes unstuck from the substance it was meant to represent, and the measuring system, in perfect working order, now measures only itself.

Anyone who has read the chapter The Signal and the Symptom will recognise the structure: as the signal is not the symptom, the metric is not the craft. And the deep reason was given by a philosopher of science who was also a great chemist, Michael Polanyi (1966): we know more than we can tell. Every true craft rests on tacit knowledge - the hand that feels the right tolerance, the nurse who notices the patient "who does not sit right with her", the technician who recognises the wrong noise - which is unformalisable by its nature, and therefore unmeasurable by construction. Not because the sensors are lacking: because what the master passes on to the apprentice does not travel through statements. A measuring system can capture the outcomes of tacit competence; the competence itself is structurally invisible to it. And this is why the firm that manages only what it measures ends up, quite literally, managing a different company: the poorer one that lives in its dashboards.

3. The score and the merit: the worker before the platform

The most exposed form of algorithmic management is that of digital labour platforms, where the "boss" is, for long stretches, an application: it assigns the jobs, computes the pay, collects the customers' ratings and distils from them a reputational score that decides access to future work. The worker's score is the twin of the customer's score met in the chapter The Customer and the Profile, with one difference that changes everything: here the stake is not which advertisement you will see, but whether you will work tomorrow. European case law noticed early: as far back as 2020, an order of the Court of Bologna held discriminatory a ranking system that treated all absences alike, blind to their reasons - the strike no different from the whim. The blindness of the metric, which in marketing costs a missed opportunity, in work costs a right.

The European legislator responded by building, around the measured worker, a perimeter that the reader of this series will recognise as a coherent system of invariants. The General Data Protection Regulation already prohibited decisions based solely on automated processing with significant effects on the person (Article 22) and left to the Member States the specific protection within the employment relationship (Article 88). Regulation (EU) 2024/1689 drew the red lines: generalised social scoring prohibited, emotion recognition in the workplace prohibited except for medical or safety purposes (Article 5), and artificial intelligence systems for the management of workers - recruitment, task allocation, evaluation, monitoring - classified as high-risk (Annex III), with obligations of transparency and human oversight. Finally, Directive (EU) 2024/2831 on platform work spelled out the principle that sums it all up: decisions affecting the employment relationship cannot be left to the algorithm alone - the worker has the right to know how the system that evaluates him works, and to obtain human review of the decisions that concern him. It is Article 14 of the AI Act translated into the language of labour: the measure may inform management; it may not, at any point in the pipeline, replace the employer who answers for it.

4. Dignity and the organisation: measuring in the service of people

It would be a mistake, however - and this book does not make it - to conclude that measurement itself is the evil. Anyone who runs an organisation knows that without measure there is no fairness: it is the transparent metric that protects the quiet worker from the showy colleague, that makes the arbitrary judgement contestable, that brings a department's overload to light before it turns into burnout. The problem is not the measure: it is the measure without governance. And governance, once again, is made of design invariants that management engineering knows well: metrics declared and explained to those subject to them, never secret; indicators always interpreted by a human being who knows the context before any decision that touches the person; effective contestability, with a reachable human interlocutor; and respect for a temporal boundary - the right to disconnect - because a system that can reach the worker at any hour must not be allowed to.

Here the technical dimension of the chapter meets, as always in this series, an older wisdom. The traditions that accompany this book handed humanity an institution that no productivity consultant would ever have invented: the seventh day. The Jewish Shabbat - received and transfigured by the other faiths - is the oldest right to disconnect in history: a time withdrawn by divine law from measure, production and performance, the same for the master and for the servant. And the Gospel gave it the reading that overturns forever the hierarchy between man and his instruments of organisation: "the sabbath was made for man, not man for the sabbath" (Mk 2:27). One could write it on the frontispiece of every enterprise information system, with the metric in place of the Sabbath. The social doctrine of the Church said it in the language of the industrial twentieth century: work is for man, not man for work (Laborem exercens, 1981). A dashboard of indicators is an excellent servant and a terrible master - exactly like every other technology encountered in these pages.

5. The professional remainder: what no time card records

In the end, this chapter too delivers its remainder. The measure counts what work produces; it does not see what work means. It does not see the workman who teaches the new hire the gesture that appears in no procedure; it does not see the clerk who catches a colleague's error before it becomes damage; it does not see the care, the loyalty, the initiative, the pride in work well done - all the silent capital that keeps organisations standing and that no line of a management system has ever contained. It is not a defect of the sensors; it is the usual ontological frontier: the craft belongs to the person, and the person - the book has learned this chapter after chapter - exceeds every profile of himself. Including the company's.

And there is one last thing the measure will never see, and it is the most important: meaning. A man can spend eight hours performing tasks identical to his neighbour's and live them in another universe - because with those hours he supports a family, repays a debt, builds a skill, serves a community, offers his toil to God or simply keeps his word. The productivity of the two will be identical in the dashboards; the reality of their days, incommensurable. The time card records at what hour a man enters his work; it will never know what of the work enters the man. That is why the chapter's last word is not against the measure but above the measure: let everything that is needed be measured, let what is measured be governed, and let it be remembered - from the management system's first commit to the contract's last line - that a company is made of people who work, not of work that has hired people. The remainder, here too, is not calculated. It is respected.

References cited in the text: Taylor F.W. (1911), The Principles of Scientific Management; Goodhart C.A.E. (1975), Problems of Monetary Management: The U.K. Experience; Strathern M. (1997), "Improving Ratings": Audit in the British University System, European Review; Polanyi M. (1966), The Tacit Dimension; order of the Court of Bologna, 31 December 2020; Regulation (EU) 2016/679, Articles 22 and 88; Regulation (EU) 2024/1689, Article 5, Annex III; Directive (EU) 2024/2831 on improving working conditions in platform work; Gospel of Mark 2:27; John Paul II, Laborem exercens (1981).