Originally published in Section Culture: Newsletter of the ASA Culture Section. Fall 2019. Vol. 31 Issue 2.
Emerging Technologies Research Lab, Monash University.
There has been a growing focus on questions of repair and maintenance across disciplines and interdisciplinary fields, including human geography, anthropology, sociology and science and technology studies (STS). Underpinning this is an acknowledgement of the incompleteness of ongoingly emergent states of things and processes and the environments that we and they are part of. Ideas of repair and maintenance are important because they can serve as disruptive and processual concepts. They invite us to focus our attention on how things keep going, rather than on how things are, and offer us a mode of understanding how humans, materials and processes are co-implicated in the making of our everyday environments. Research into repair and maintenance has often focused on how humans manipulate material things (for instance in geography or anthropology), technologies (in STS), or dialogue (in conversation analysis), and has also often pitched narratives of repair against those of innovation.
During the same period interest in digital data has grown in sociology and across the social sciences. This is represented in the growth of big data, and the ubiquity of personal data technologies, such as for body activity monitoring or health tracking. The advent of big data and personal data technologies can be seen as supporting an innovation narrative. That is, they are commonly seen as technological solutions to societal problems. For instance, big data and the predictive analytics associated with it are commonly associated with the ability to know in advance how certain people are likely to behave. Sociologists are rightly concerned about how such analytics might be used in predictive policing and other punitive modes of activity. Personal data technologies – such as wearables that monitor steps, heart rate, and caloric intake and expenditures – have been pitched as possible ways of changing unhealthy behaviours in people with health problems. Here sociologists have likewise been rightly worried that such technologies might lead to new and intensified modes of surveillance (e.g., by insurance companies).
These solutionist hopes for big data and predictive analytics in terms of personal data and health interventions are themselves consistent with narratives that suppose that technological innovation produces finished and complete products. Within them we might suppose that data itself is seen as complete and reliable. That is, if we are to use big data analytics to predict what people in society will do, then it would seem important to be able to believe that the data used can be depended upon. Likewise, if people are to use personal data as a basis from which to improve their health, then surely they would need to know that these data are accurate?
Inspired by the recent research on repair and brokenness, and the way that this challenges our assumptions about things and processes in society, a group of us decided to ask similar questions about data. Our interrogation of this question formed part of the work of the Data Ethnographies Lab, with participants in Australia, Sweden, Denmark and Spain. We debated and explored the question of ‘broken data’ through a series of three ethnographic and autoethnographic examples drawing on Sarah Pink’s work on personal self-tracking data, Robert Willim’s work with sound data, and Minna Ruckenstein’s research on how big data analytics are performed in practice. In each of these areas we found examples that showed us how data were not necessarily objective, complete, fixed, or finished. To understand this we proposed the concept of ‘broken data,’ as a way of thinking about digital data that aligns it with narratives of repair and maintenance, and to suggest that digital data does not necessarily support the narratives of technological innovation or solutionism with which it is often associated. Indeed, Pink’s study of self-trackers found that users were not always particularly concerned with the accuracy of their data. While the data might have had gaps, or did not necessarily represent the activity with which it was associated in the app, this did not always matter when the participant’s objective was to use it to create coherent narratives about themselves and their daily trajectories of movement. Ruckenstein’s work showed how big data might be broken due to, for instance, various anomalies or infrastructure failures, how this limited the data, and how it needed to be repaired so that it could be used in analytical processes.
Thus through the concept of broken data, we were able to emphasise that digital data should not be treated as providing certainties, but rather, needs to be understood through the stories of the ways in which it was produced. These are likely to be trajectories not of the making of completely finished products, but in fact incomplete, repaired, and maintained in ways congruent with the particular strategic purposes of those who use them. That is, while digital data might, on the surface, appear to support narratives of technological problem-solving, once one scratches below the surface digital data reveals itself to be similar to other socio-technical things and processes: that is, incomplete, sometimes damaged, and repaired and maintained by humans.
Pink, S., M. Ruckenstein, R. Willim, & M. Duque ‘Broken Data’ (2018) Big Data and Society. 5(1) https://doi.org/10.1177/2053951717753228
Fors, V., S. Pink, M. Berg and T. O’Dell (2019) Imagining Personal Data. London: Bloomsbury.