The Uses and Users of Social Media Data Mining. A review of Post, Mine, Repeat. Social Media Data Mining Becomes Ordinary by Helen Kennedy

With Post, Mine, Repeat (2016), Helen Kennedy offers a critical contribution to public debates about datafication, the uses and ethics of social media data mining and a report on a number of research projects that she has conducted with colleagues over the last years. The core question of the book is how is social media data and its analysis understood by different types of actors, ranging from city council officials and museum managers to marketing analysts and lay users of social networks. Kennedy cautions that the term ‘ordinary’, despite its legacy in cultural studies to address the level of ‘ordinary life’ , here refers to “organisations whose operations contribute to structuring daily life” ¹ These actors explore uses of data mining and analysis to solve problems, create publics for their causes or simply look for ways to stay up to date under increasing demands of efficiency.

Contrasting the managerial and often evangelical discourse on big data (and its alleged possibilities) with the perspectives of users informs Kennedy’s thinking about datafication. Within a cultural studies perspective, Kennedy makes a strong argument for critical inquiries of data, data mining and the role it plays in public discourse and quotidian life. She urges us to be both critical and to regard data mining and analytics as conduits to problem-solving.

“Thinking with both critical and problem-solving perspectives is essential, I suggest, for addressing the questions of what should concern us about social media data mining and whether there are forms of data mining that can be enabling of the non-powerful.” ²

At the core of the book is thus an engagement with questions of power and agency, that Kennedy tends to discuss in favor of different levels of agency without losing sight of the structural factors that determine in what ways social media data is structured and made available to users by owners of social networking sites. Kennedy argues that “the focus on structures of datafication has meant that consideration of the possibility of acting with agency against data power has been relatively absent.” ³ The challenges of acting in such a way, are not simply posed by technological infrastructures or analytic skill. Post, Mine, Repeat argues for a contextual understanding of data mining practices, in which the individual motivations of actors, ethical considerations and organisational as well as legal structures are not secondary but essential for understanding the relevance of data and data mining ⁴. It poses the normative question “can social media (and other) data mining ever be considered acceptable or be used in ways that we find acceptable?” ⁵. This question then addresses how data is understood by actors and researchers and not merely how it is gathered, processed and managed.

Structure and Organisation of the book
The book is organised into nine chapters. The first three chapters argue for the necessity to contribute in an engaged and informed way in debates about the mining of social media data and to develop ethical understandings of how such data is created and used. Kennedy positions her contribution within a spectrum of antagonistic positions, which have tended to skew public and academic debates to extremes. On one end of the spectrum are the possibilities of ‘big data’, managerial control and surveillance, neo-liberal (self-)optimisation and rampant infringements on privacy. On the other end of the spectrum are arguments for user empowerment, new forms of social activism and political participation. Kennedy reviews these positions extensively, to locate her “phenomenological” approach. ⁶ to social media data mining and the actors engaged in it: “I give priority to empirical detail because in discussions of data power to date, the emphasis has been on structuring forces. The empirical detail in this book is intended to complement critical commentary on structures of data power, not to dismiss or overlook it.” ⁷

Consequently, the body of the book (chapters 4 to 8) presents empirical findings from interviews, focus groups and experiments with different types of actors. Chapter four documents a form of ‘action research project’ ⁸ where Kennedy and her colleagues worked with UK city councils and museum managers to explore ways of mining social media data. Chapter five presents findings from a range of interviews with employees at social media insight companies, detailing how actors negotiate the ethical and practical challenges of providing convincing and reliable data to their clients. Chapter six looks at how data mining practices and analytics change organisational structures and work routines through interviews with communication professionals from, e.g., universities, non-profits and media companies. Chapter seven then reflects perspectives from users of social networks on ethical challenges of data mining. Summarising the results of three focus groups in the UK, Norway and Spain, this chapter shows that users contextually evaluate the appropriateness of data mining, demanding fair use and transparency. In Chapter eight, Kennedy explores academic uses of data mining and presents examples of ‘doing good with data’ , e.g. through data activism or the open data movement.

Given the strong presence of these studies in the book, the final chapter is then an empirically-grounded theoretical contribution to question and evaluate the latent ‘desire for numbers’ in data studies, a desire that is both instrumental and rational, but that tends to sideline the ethical dimensions of dealing with data in ‘ordinary’ contexts. At first sight, this conjunction of empirical findings with critical discourse may seem unusual “and some people might not like it”. ⁹. But the framing chapters are not simply adjunct or introductory to the empirical studies. Neither are the studies only illustrations or rebuttals of critical positions. One may regard this composition, although the term has a conflictual legacy, as dialectical. Kennedy (and her colleagues) interrogate and confront reflections of data and power structures on the ground floor of practical experience through the perspectives of different groups of actors. At the same time, these experiences frame and inform the scholarly evaluation of critical positions on data. What emerges as a result, is less a provocative position that seeks to distinguish its academic brand value but more an informed critique both of the values and misunderstandings of the role data mining (and data itself) play in today’s society below the radar of very prominent actors in the field. In the end, neither the studies nor critical discourse frame each other, but each in their own right reveal specific forms of knowledge that Kennedy presents to her readers as ‘things we think with’ about data mining, to use a term from Sherry Turkle. ¹⁰

Things and Concepts to Think With
Throughout the book, Kennedy makes use of a number of concepts that frame her perspective on social media data mining. Readers will readily acknowledge that her summaries of these key debates in media and communication studies serve as excellent introductions and contribute to the formulation of a relevant and timely research agenda. One strength of her argument is the extensive discussion of different forms of agency – among workers, users and in technology – that Kennedy invokes to “consider the extent of the dominance of structures, the possibility of agency, and the spaces in between” ¹¹.

Engaging with the work of Andrew Feenberg, Kennedy questions ‘the logic of machinery’ and seeks to uncover potentials of resistance within structures of power. ¹² Drawing on ‘social media logic’, she is interested in how the individual ‘progammability’ of social media ¹³ ties in with ‘intimacy interaction’ ¹⁴ and negotiations of context collapse between personal and public communication ¹⁵ within a framework of monetisation and economic interest. Kennedy frequently invokes Nissenbaum’s concept of “contextual integrity” ¹⁶ to argue for a more nuanced understanding of how public communications on social media platforms need to be considered within the quotidian (and often personal) contexts in which they appear: “social media feel private, personal and intimate, even when they are not” ¹⁷. Her discussion of ‘social’ versus ‘institutional privacy’ is here especially productive to go “beyond privacy and surveillance” in evaluating what kind of data may be used appropriately in data analytics and how users themselves perceive of their involvement in such research. ¹⁸

On the subject of datafication, Kennedy refers to the work of Porter on Trust in Numbers and Grosser’s arguments on the “metrification of sociality” to argue first of all that perceiving the world through data means to manage rather than understand it. ¹⁹ Data analytics are regarded as “measures of impersonality” ²⁰, which cater to a managerial demand of efficiency, scalability and surveillance. The availability of analytics software and seeming ubiquity of data pools creates a continuous “desire for numbers”, which are used as validation of policy decisions, as documentation of public outreach or as regulation of precarious working conditions. As she discusses in the closing chapter, the desire for numbers creates the expectation that social interactions can be measured and predicted, even though practitioners in data mining are aware of the shortcomings of tools, platforms and measurements themselves: “The datafication of the ordinary and the everyday makes this a widespread phenomenon—what was once qualitative is now measured quantitatively, so quantities are desired in relation to more things, to that which was previously qualitative” ²¹. Associated to this desire for numbers is the belief that mining will, when it is done correctly, “produce results”, feeding an “expectation that data would be found” ²². This desire for numbers becomes especially apparent in Kennedy’s report on public sector officials (chapter four) expecting data mining to yield new insights into the constitution and opinions of their audiences, while also expecting researchers to instruct them on best practices for mining and analysis.

Summary
Readers of Post, Mine, Repeat will find the dialogue between empirical research and critical thinking especially rewarding. One great benefit of the book is that it engages both theoretically and practically with questions that all too often are relegated to the margins of the debate, to mere praxis, or are overshadowed by the exploits of large corporations or well-endowed research centres. What emerges from Kennedy’s critical discussion of datafication are a number of ways in which social media data mining has become ordinary. The sheer volume of data that is nowadays produced and analysed suggests that data is ordinary in the sense that it is ubiquitous. As such, data is interpreted as a resource for public and private ends, as a gratuitous, free-flowing stream of information and opinion that was not previously available. For the public sector officials presented in chapter four, those who were not acquainted with data mining before, such analyses are ordinary in the sense that they yield banal or irrelevant results. For professionals working in social insight companies (chapter five), data mining techniques are commonplace but pose ethical challenges as to what types of data are appropriate to use and how clients will interpret the results of analyses. For communication professionals (chapter six) data mining is part of an everyday experience that has restructured how they conduct their work and that continues to challenge the organisational frameworks in which they operate. To users of social media platforms, data mining is ordinary but invisible . Although they support certain ways in which data mining may be beneficial for them or the public (targeted advertising or research purposes), the processes and technologies used in mining their personal data remain opaque and intransparent.
For academics, Kennedy seems to suggest, there is now a troubling new sense of the ordinary associated with studies of datafication: Mining social media data on a large scale is increasingly regarded as the benchmark of good research. While a few research institutions can actually provide the intellectual and technological resources to create huge datasets of online discourses and issue networks, more and more scholars working with qualitative methodologies are suddenly urged to legitimise their approaches, whether phenomenological, theoretical or philosophical, in face of the increasing metrification of research and the datafication of research output. Against such new digital divides among researchers, Post, Mine, Repeat makes a powerful statement for a cultural studies perspective on doing social research in a networked environment.

References
Burgess, Jean. 2006. “Hearing Ordinary Voices: Cultural Studies, Vernacular Creativity and Digital Storytelling.” Continuum: Journal of Media and Cultural Studies 20 (2): 201–14. https://dx.doi.org/10.12924/mac2013.01010002
Feenberg, Andrew. 2002. Transforming Technology: A Critical Theory Re-Visited. Oxford: Oxford University Press.
Grosser, Benjamin. 2014. “What Do Metrics Want? How Quantification Prescribes Social Interaction on Facebook.” Computational Culture. A Journal of Software Studies. 9 November 2014. http://computationalculture.net/article/what-do-metrics-want. Accessed April 4, 2017.
Ito, Mizuko, Judd Antin, Megan Finn, Arthur Law, and Annie Manion et al., eds. 2010. Hanging Out, Messing Around, and Geeking Out: Kids Living and Learning with New Media. Cambridge: MIT Press.
Marwick, Alice E., and danah m. boyd. 2011. “I Tweet Honestly, I Tweet Passionately: Twitter Users, Context Collapse, and the Imagined Audience.” New Media & Society 13 (1): 114–33. https://dx.doi.org/10.1177/1461444810365313
Nissenbaum, Helen F. 2009. Privacy in Context: Technology, Policy, and the Integrity of Social Life. Stanford: Stanford University Press.
Porter, Theodore M. 1996. Trust in Numbers: The Pursuit of Objectivity in Science and Public Life. Princeton NJ.: Princeton University Press.
Silverstone, Roger. 1994. “The Power of the Ordinary: On Cultural Studies and the Sociology of Culture.” Sociology 28 (4): 991–1001.
Turkle, Sherry. 1997. Life on the Screen. Identity in the Age of the Internet. New York: Touchstone.
van Dijck, José, and Thomas Poell. 2013. “Understanding Social Media Logic.” Media and Communication 1 (1): 2–14.
Williams, Raymond. 1993. “Culture is Ordinary.” In Studying Culture: An Introductory Reader, edited by Ann Gray, and Jim. McGuigan, 5–14. London: Edward Arnold.

Notes