Big data, little politics
Governing has always been a data-intensive task. Crises and pandemics remind us once again how important data is to make the right decisions and make the best forecasts. The big data It is a technology that will not only modify the efficiency in the provision of public services or the precision of strategic planning, but also the relationships between citizens and public power, as well as between politicians and the administrative system. The United Nations has spoken of a “data revolution”, thanks to which an objective, neutral and irrefutable knowledge would be generated, which would result in a more rational and apolitical government action, a public service that does not speculate with mere hypotheses or be slave of ideology. We would move from policy-defined evidence to evidence-based policy.
It is not surprising that expectations of democratization have been triggered in this way, which are presented as surpassing the old, ideological, subjectivist and arbitrary politics. Some suggest that the big data it has thrown thinkers like Adam Smith or Karl Marx into the dustbin of history, since markets and classes are aggregates, “averages”, like any social phenomenon, made up of millions of small transactions between individuals. What if the major categories of politics were nothing but constructions that have very little to do with the actual behavior of societies, words that hide rather than reveal who we really are?
The provision of data is an indisputable procedure for the improvement of government action; more questionable is the extreme enthusiasm that this new possibility provokes in what could be called “dataism”, a secular belief in the anodyne qualities of data that would lead to an ideology beyond any ideology and whose paradigm would be “no politics, just data (politics, no; just data) ”. Considered as measured objectivities, a sense of justice and fairness is expected from the data, a way of deciding without having to decide, a great opportunity for depoliticized legislation.
The first question that should be asked is whether we are dealing with a depoliticization in the best or worst sense of the term, that is, whether it diminishes power as an imposition or simply metamorphoses. What would be the new power relations generated by data analysis? As large amounts of data exceed human capacity to analyze it, more and more automated algorithms have to be used to identify patterns and support decision-making, increasing reliance on such technologies and intensifying power asymmetries.
The big data it is a political issue to the extent that the circuits of production, distribution and consumption are political, that is, places where access, control and capacity are unevenly distributed by asymmetric power relations. There has even been talk of new social classes in the data society based on who produces it, who has the means to collect it, and who has the skills to analyze it. The impact on power relations in its various forms is all the greater the more the government, public administration and expert knowledge rely on data control. There is a growing power differential between those who collect and analyze data versus those who simply feed it. But it is also that the data is not an apolitical reality; its collection, analysis and use depends largely on certain decisions. The more policies that are justified in data, the more important it is to know the assumptions, explicit or hidden, that underlie the decision to attend to this data and not to others, or the biases that they manifest. The nature of the information available always defines and conditions the problems that governments face and the way they do it.
All the appeal to the importance of data may be working as a mantra that makes us unaware of the need to carry out fair and sustainable data policies to configure such places equally. The discourse about data cannot be reduced to industrial and administrative needs, but must be open to questions of social and political expediency, including the possibility of stopping or rejecting certain technological applications. And we should not fall into the illusion of thinking that it would be enough to have the correct information so that all problems could be solved without resorting to political decisions, judgments and values.
Data analysis and its increasing sophistication appear to meet a demand for accuracy present in many sectors of society, especially in times of complexity and confusion. Politicians want irrefutable statistics, the media seek concise facts, judges aspire to identify irrefutable causalities, and people long for the certainty of numbers. Are we in a position to satisfy this demand through the technologies of the big data?
It is curious that the crisis of political representation, which has been invoked by many protests in recent years, has given way to an uncritical acceptance of the ability of data to represent us. Did not our political representatives represent us and instead our data does? If the mandate of political representation is questioned, monitored and revoked, the claim to represent through data what we really are and want should be accompanied by a reflection on the fulfillment of that promise, its epistemic limits and its political and political conditions. economical.
It should not be forgotten that the ability of data analysis to discover connections between elements is fundamentally based on correlations, not causalities. Just as there are exact but absurd translations, there are true but spurious correlations. Correlations are very useful, but understanding them as if they were causalities, that is, as if they made the interpretation exercise unnecessary, leads to fatal errors. We could recall in this regard the famous story that Google, using search statistics, detected a flu epidemic before the health control centers through epidemiological reports, but less than Google Flu Trends has also been wrong, probably because the successes of the experts are less news than their failures. The books about big data They also tell the story of a company that sent products for newborns deducting a pregnancy from the movement of a credit card of a man who, angered by this assumption, had to later apologize to the company when he discovered that his daughter was pregnant. What is not usually told is why that company and others have had to change their advertising strategy by also offering other products to protect themselves from erroneous or unethical “diagnoses”.
The policy of the big data It has given rise to a great number of fascinating promises, but we should not underestimate moments of uncertainty in their epistemological limits and spaces of freedom. As long as human systems are complex, contradictory and paradoxical, the data will generate knowledge that will continue to be refutable, human, too human.
Daniel Innerarity is professor of Political Philosophy and Ikerbasque researcher at the Universidad del País.