WHAT is the relationship between “science” and “data”?

Conceptualising science in terms of “data” is a mid-20th-century framing that tracks the growth of computing power. Science encompasses both the design of data collection and its interpretation — in the transformation of data to meaning.

Data points represent some sort of map of reality. There are whole scientific fields and philosophical debates about the collection of those data points, which can never be a passive expression of the underlying universe, because choices have to be made: what to collect, and what to leave out.

In most Western societies, where science has taken over from religion as the supreme authorising technology, governments use scientism to back up their claims and decrees.

The famous phrase “lies, damned lies and statistics” refers directly to the absurd use of data to justify the most indefensible positions.

But the rhetoric of data remains. Even Trump’s more outlandish claims about vaccines, for example, are presented through the lens of these discussions.

The ontology of the 21st century requires us to engage with debates about data. Data has meaning, its authority emanates from its witness to reality, we are compelled.
This reliance on data has taken on a new power in the age of big data.

Though the increasing collection and collation of data has been a feature of the scientific and digital ages, the sheer quantity of data harvested from each of us daily now is beyond human comprehension.

Technologies permit huge data gathering in the form of GPS trackers, heart rate monitors, accelerometers, transaction monitoring.

As it stands, a lot of data is for sale — this is precisely why it is gathered. Data science and analytics companies buy the data from the companies that harvest it, and sell “analysis” and consultancy back generally to the companies that want to optimise their business strategies using justificatory numbers.

As we submit to the changes, recommendations and decisions wrought by algorithmic feedback, we are the first global cohort to experience real robotic control.

The early internet was full of the hope of devolved informational structures. With access to data on society, people would wrest power back into their own hands. But outside of limited examples like Wikipedia, most of the data that is taken from us is held in the hands of corporations and therefore unavailable to the public.

Government datasets are another exception. One example is the English Index of Multiple Deprivation, last updated by the British government in late 2025.

This is a staggeringly detailed mass of data combining many measures of deprivation — employment, income, health, experience of education, crime, environmental pollution and housing quality — into a single index.

The extraordinary part is that each of these different aspects is quantified for areas of England containing on average only 1,600 people. That means there are more than 33,000 individual locations. The index is combined with demographic data, showing where children and old people are particularly suffering deprivation.

The data is recollected, collated and released freely every five years. It is used intensively by local government and health authorities in decisions about allocation of resources and the focus of interventions.

Deprivation is calculated relative to society’s changing expectations: as the official summary puts it, “these social expectations may change over time, but relative deprivation remains.”

The data is free and simple to access: https://deprivation.communities.gov.uk/ has an interactive map, with colour coding and charts to show the interaction of different metrics of deprivation, including the ability to compare any two areas.

The data is relative, with every different patch of land scaled as deprived relative to each other. Half of all patches are, by definition, below average. The data makes visible what many of us know: England is a patchwork of regional inequality, including at the most fine-grained of scales.

Cities have deep divides in deprivation, and the relative scale of the map reflects a reality in which people experience radically different qualities of life. Incidence of premature death, exposure to pollution and violent crime — all vary wildly.

In particular, the index of income deprivation affecting children shows the egregious separation of resources that shapes the next generation’s life chances.

There are inevitably changes and improvements that could be made to the data — its gathering and sorting, the questions it addresses, the ways that it weights some metrics of deprivation with relation to others. In the end, human misery cannot be quantified by numbers.

This data fits with a liberal or managerial view of society. This “dashboard” containing semi-automated text will form the bedrock of much social policy, and affect the flows of local government attention and money over the next five years.

Very many well-intentioned people will use it to inform constructive action, albeit in ways that often come down to the least unfair way of distributing limited resources rather than anything grander.

But also, as the datasets are freely available, we can also predict that they will be used by insurance companies: by capital for capital.

It is naive to think that political problems can be ameliorated simply by more or better data. Without changes to the structures of power, data simply presents us with a map of reality — a map whose high resolution can encourage us to focus on changing small details, rather than altering things so radically that a different map is required altogether.

However, if people take such data into their own hands, for their own ends, the transformation of data into meaning could be a truly powerful science.