Philosophy of data science
The “Impact of Social Science” blog of the London School of Economics has, in the past few weeks, published a series on “Philosophy of data science“. Each installment is an interview conducted by sociologist Mark Carrigan with a key contributor to the social science reflection on data.
Rob Kitchin proposes a definition of big data that is more complex than the ones often found in the business literature — seven traits instead of the usual three or four “Vs”. But all share an important point: volume, in itself, is not enough. Interestingly, Kitchin defends a view of data-driven science which is more open to existing theories and knowledge than sheer empiricism, or (worse!) end-of-theory views à la Chris Anderson.
Evelyn Ruppert insists on the social dimensions of big data, somewhat overlooked by the common emphasis on data analytics, and suggests that at the moment, there is insufficient engagement with the cultural, economic and political consequences of data. This is one reason that prompted the creation of the journal that she edits, Big Data & Society.
Deborah Lupton is fascinated by the metaphors surrounding big data, such as “floods” and “flows” – natural metaphors that we use, perhaps unwittingly, to domesticate a phenomenon that appears somewhat threatening. Interestingly, she points out that the interest for big data has also renewed interest for “small” data.
Susan Halford discusses the changes in methods training that are necessary for social scientists to be able to engage with big data, and how this requires enhanced exchanges across disciplines. This is definitely a challenge that should be taken up, if we want social science to thrive in the era of digital data.
Noortje Marres similarly insists on the opportunity (and challenge at the same time) of inter-disciplinary collaborations, and how digital data make them necessary, encouraging sociologists and data scientists to interact more, and to better understand each other.
Sabina Leonelli is a historian of science and reviews the way in which what is understood as trustworthy data, changes over time and across places. Again, she reiterates the need for social scientists to cooperate with each other, and with colleagues in other disciplines, to be able to engage with big data and contribute to the full development of their potential for new knowledge creation.
Finally, Emma Uprichard expresses worries as to how big data affect methodologies in ways never seen before, potentially threatening a wealth of expertise, knowledge, norms and best practices that had been established with small data. She calls for methodological pluralism in social science, and insists especially on the importance of theories and research questions: these, not methods or data in themselves, can provide real answers to the (big?) social issues we ultimately care about.