Differences between computer and social scientists and perceptions of data might more subtler than often thought

Submitted by Isto Huvila on Mon, 05/28/2018 - 10:32

I ended up in an interesting discussion together witha group of colleagues about the different perspectives to data and more generally, to the modus operandi in computer sciences and social and cultural sciences when I was recently participating as a panelist at the workshop What can be known from the web? Source criticism beyond bots, agents and trolls in social and cultural web research organised by prof. Gertraud Koch (University of Hamburg) at the ACM Web Science 2018 conference at VU Amsterdam. The discussion, or the workshop really, started already during the lunch break and continued at the workshop proper and was in many ways enlightening in bringing about common assumptions of how computer scientists and social scientists think and what they think of each other. 

It is typical to hear social scientists and cultural studies researchers to accuse computer scientists of essentialist beliefs about the nature of data -- that it is a one-to-one representation of something -- and similarly, computer scientists wondering the unspecificity of social scientists that refuse conclude anything more than that things are awfully complicated, or to criticise computer systems of their shortcomings without proposing any viable alternatives or ideas how to fix them. 

Firstly, it is questionable whether the boundaries are not that much between social sciences and (hard) computer sciences. There are hard social scientists and soft computer scientists. Secondly, it is doubtful whether the epistemological beliefs are always that different in the end. Much of the difference can indeed be traced back to the 'computer-science-minded' aim of using data as a shortcut to find a good-enough explanation of a phenomenon in order to develop something to address an identified problem within its confines, and a 'social-science-minded' goal of providing an as complete as possible explanation of the phenomenon itself. Similarly, when it comes to data, the difference in the view might not be that much between different disciplines than with the increasing tendency (related to a large extent to the possibility) to appropriate and analyse different types of 'stuff' as data rather than as something else. Before the proliferation of computational tools, working with data was difficult, time-consuming and labour-intensive that effectively made many other options more attractive. At the present, the situation is pretty much the opposite. More or less all kinds of stuff (even if not everything) is easy to treat as data -- for both good and bad. In this sense the datafication is not really a question of a transformation of everything to data but of treating and using stuff as data to a much larger extent than before.