Data and algorithms in society

3. Data and algorithms in societyΒΆ

Two topics that receive particular attention in the discussion about the uses and effects of digital technology in politics and society are data and algorithms. Data collected with and on digital devices make new phenomena, objects, and behaviors visible to those with access to that data. At the same time, algorithms allow the targeted roll-out of rule-based interventions. Here data serve as input to develop rule-based internventions while also documenting their effects. At the same time, they make targets of interventions visible that otherwise might be difficult for interested actors to identify or make accessible for algorithmically designed interventions.

Digital technology has extended the amount and depth of data on human behavior and social systems. This has been seen by some as a measurement revolution for the social sciences and as providing many new avenues to knowledge as well as new business opportunities. Still, these data riches have to be translated into meaningful measures of phenomena of interest and society. Without meaningful interpretation these new data remain noise machines, potentially even hiding the signal one is looking for in order to gain understanding or improvement.

At the same time, new data and new measurement opportunities also mean new opportunities of social control through companies or governments by making new people, objects, or phenomena visible to onlookers and provide areas of contact by which to anchor targeted interventions in order to change dynamics or achieve specific outcomes.

Both these characteristics of data - translating data into meaningful measures and also thinking through the consequences of making more people and more of social life visible and manipulable - demand for a critical interrogation of this feature of digital technology through social scientists.

The other big topic for this session are algorithms. A computer algorithm provides a computationally executable series of steps with the goal of solving a given task or problem. These sequences can either be developed and programmed by humans, or they can be autonomously identified through machines themselves: rule-based algorithms or machine learning algorithms. These algorithms use data for learning patterns or for automatically intervening once specific patterns emerge that trigger a sequence of steps. Examples come from many different fields: for example in finance, algorithms are used in flagging a specific transaction as likely being fraudulent; in policing, they are used in assigning a person a high or low probability of being a criminal or not.

Algorithms are thus clearly a force in shaping people's option spaces in various areas of social life. This has given rise to broad concerns that are part of public debate but also of academic research. Concerns focus on the opaqueness of the uses of algorithms, their inner workings, and their effects; the fairness of their outcomes; and fears of unintended consequences once algorithms are rolled out in scale. This makes algorithms, their uses, mechanisms, and effects a prominent topic for social scientists engaged in understanding the use of digital technology in society.

This chapter will close by discussing uses of data and algorithms in two broad social fields: political campaigning and journalism. For both areas, I will briefly discuss uses of data and algorithms as well as presenting a small sample of studies, trying to address connected questions through different empirical strategies. But first, we need to get a better grip on the nature and uses of data and algorithms in general.