How Does Data Turn Into Power?

“Data is the new oil” has been repeated for years. What the analogy wants to say is clear: raw data, apparently ordinary, can create great economic value once processed. But the comparison with oil misses something important. Oil diminishes where it is extracted; data multiplies as it is used and gains new meanings when combined with other data. Moreover, oil does not predict what we will do. Data is not merely a resource to be sold but an instrument for classifying people, shaping the future and taking decisions on others’ behalf. Its turning into power begins precisely here.

Individual records usually look innocent. Age, postal code, phone model, hour of purchase or the video watched tells nothing about our life on its own. Yet when thousands of small signs come together, a profile forms about who we are, what we need, what we fear and when we can be persuaded. This profile does not reflect us fully; it may even be wrong. Still, if a bank, an employer, an insurance company, a platform or a public institution acts upon it, data acquires a real force over us.

The road from knowing to deciding

Knowledge has always been related to power. The state counted its population and built systems of taxation and conscription; factories measured workers’ time and controlled production; companies knew their customers and widened their markets. What distinguishes the digital age is that measurement has spread across almost the whole of daily life and that results can be processed very quickly. A record that once sat in an archive can now be the input of a system deciding in an instant.

When a platform knows what we like, it does not merely offer us suitable content; it also decides where to direct our attention. When a financial institution derives a risk score from our behaviour, it does not merely recognise us; it determines under which conditions we may borrow. When an employer collects performance data, it does not merely measure work; it establishes a new authority over who will be promoted and who discarded. Data turns into power only when it is connected to a mechanism of decision.

The classified human being

Data systems cannot work without dividing the world into categories. They form classes such as safe–risky, relevant–irrelevant, productive–unproductive, genuine–suspicious. The problem is that life does not fit exactly into these boxes. People change, exceptions exist, the same behaviour means different things under different conditions. But for the system to work, uncertainty is reduced. The story that describes us is converted into a few variables and a probability score.

How these classes are built is a political question. A system trained on the records of people who were treated unequally in the past can present old inequality as a neutral prediction of the future. Living in a particular neighbourhood, graduating from a particular school or having a gap in one’s work history can become a disadvantage even in models that carry no intention to discriminate. The machine does not say “this person is bad”; it merely computes a low score. Yet behind the score the whole weight of history may be present.

Holding data is power; keeping others without it is power too

The advantage of large data holders is not only that they know more. It also matters that they can prevent others from reaching the same knowledge. When a platform builds a unique archive of a society’s behaviour, small competitors, researchers and public institutions cannot see the same world. The company may notice before anyone else what people want, which news is spreading or how a crisis is developing. While society is deprived of the knowledge produced about itself, a private institution converts that knowledge into strategic superiority.

Data thus comes to resemble a new movement of enclosure. Traces produced in common within daily life accumulate in private stores. A city’s movement, a society’s conversations, the language use and cultural production of millions become the raw material of corporate models. Value is produced together; ownership and the right of decision gather in a few hands. The user is part of this order not because they “gave their data” but because social life itself has been turned into a source of data.

When prediction begins to construct reality

Data-driven systems do not merely predict the future; sometimes they create the future they predict. If a neighbourhood is policed more heavily because it is deemed risky, more incidents are recorded there; the increased records strengthen the prediction that the neighbourhood is risky. If a piece of content is promoted because it is expected to be popular, it genuinely becomes popular. If a worker is given fewer opportunities because low performance is expected, the resulting low performance appears to confirm the first assessment.

In this loop data ceases to be a photograph of the past and becomes an instruction for the future. When a system does not measure its own effect, it believes it has discovered “objective reality.” Yet the classification changed the opportunities people met, and then collected the changed outcome as data once more. Power here lies not only in what is known about whom, but in knowledge’s capacity to make the world resemble itself.

Common data, common decisions

Limiting the power of data is not a matter of multiplying privacy settings. Telling people “don’t share your data” places the whole responsibility on the already weaker user. Leaving no data at all while using today’s essential services is usually impossible. Besides, the issue is not only personal privacy; conclusions drawn about communities can produce discrimination beyond the consent of individuals.

We need to ask more fundamental questions: which data should never be collected? How long should collected data be kept? Does a person affected by a model’s decision have the right to an explanation and to object? How should the economic value produced from a society’s collective behaviour be shared? Who should supervise the institutions that govern data? How can data used for public benefit be held in common without becoming a new apparatus of surveillance?

Data commons, data cooperatives, independent auditing, data minimisation and communities having a say over their own data may be parts of this search. None is a simple solution. Holding in common does not mean opening everything to everyone; careful boundaries are needed between privacy and public benefit. The aim is not to make data a new sacred property but to democratise the decision-making power built upon it.

Data is not power in itself. A record sitting on a hard disk governs no one. It becomes political when the knowledge to interpret it, the infrastructure to run it, the institution to turn it into a decision and the force to impose the outcome come together. Beside the question “is our data safe?” we should therefore place a more uncomfortable one: whose future is being built with our data? If the answer is only better service for companies or more effective administration for the state, then we live in an order in which we produce a great deal about our own lives and have very little say over them.