Big Data is a selling point. Bigger is ostensibly better. Big enables smart, which allows the efficient management of things, people, and markets. A few years ago, I published an article that argued that big data was merely a grounds generating technology for what is otherwise argument. This is still my go-one argument about Big Data. What I have been thinking about more recently, is that the questions that Big Data ostensibly provides answers to may not be answerable. More specifically, questions about justice, power, and freedom are often described with the term aporia, a difficult, undecidable, impasse. The means by which we tend to deal with these blocks are not simply imperfect, but impossible in themselves. Counting systems, constitutions, collations, and consubstantiation are not nearly enough to deal with the null. Bayesian stats aside from providing more robust means of calculating also provide a means of navigating to the lodestar, just to the right of the event horizon of the unknowable (more on this later).
In this short essay, I make the case the Big Data attempts to replace real, abductive reasoning with pseudo-inductive reasoning, and that challenges and responses to the efficacy of this reasoning form a troubling cynical loop.
Big Data is probabilistic
Big Data analysis is not neutral, objective, value free, or apolitical. As a performative, to declare something Big Data endows it with a talismanic potential to resolve an aporia. Translating the results of large scale data collection and analysis into meaningful ethical, aesthetic, and rational argument is difficult work. There is no essential truth about the universe that is revealed by scrapping real estate listings except that land use policy is complicated and political. This is an important finding, but not one that required a great deal of analysis to know if one started with honest prior knowledge of history. If one starts from the perspective of a space alien who elected not to use wikipedia, this could be a meaningful finding,
Claims to the objectivity of scraping and analysis rely on a special sort of mendacity that briefly precedes recognition as a hack. Watching the self-congratulatory moralizing of austerity advocates in the context of Greek devaluation is instructive here, making circular points to an agreeable audience is like watching a slow motion catastrophe that should have been prevented, like Adam Sandler’s deal with Netflix. Economic rationality has an aesthetic, mathematics does as well. Beauty, as the most basic translation of the aesthetic, is never far from the mind. This is not to say that there is some form of thought beyond style, Roland Bleiker should have dispelled this notion for you.
This is not to say that some elements of the Big Data project are not useful. Detecting things in large datasets can be quite helpful if those things had not been detectable previously. Testing hypotheses using natural experiments may be far more powerful than disproving a null hypothesis. Maps (both literal and figural) created with scrapers and networking software can provide resources for developing new qualitative understandings. Visualizations and other intermediate documents (hopefully I have some research forthcoming here) can enhance practices of ethnographic noticing. There are many ways in which new computational tools can be very helpful in both qualitative and quantitative inquiry, with and without hypotheses. Although none of this is magic. There is no “research” button in the software, and if one is treating a cross-tab all function in R as such a button, they should re-evaluate their life choices.
The computational turn in critical.cultural studies will depend on the deployment of scrapers, mappers, and statistics in the service of arguments. Confectionary digital humanities projects have become as cloyingly sweet as poorly executed applications of Foucault might become stale. Critical/cultural studies can deploy these resources to enhance meaningful political projects. In one sense, this balances the argumentative equation, both sides have data on their side. The alternative is to critique the idea of data from such a totalizing position that many other arguments are lost. Rolling hard against the idea of Big Data takes quite a bit of effort, instead of deploying the critique of the human sciences to awkwardly cleave social scientific research away from the humanities, it seems equally productive to deploy those same tools to balance the equation, to spoil the fetishistic certainty of Big Data with the uncertainty of big data. Fight data with data. Strategically reverse power relations, just because someone has done some information artistry around your issue does not mean you are required to accede to their judgement.
Underlying this perspective is the idea that the knowledge is probabilistic. Pierce’s abduction can be quite useful here. The hinge of rhetoric is probability, sublimating the probable into the certain is the work of all great rhetoricians, confidence tricksters, and reality television stars. The Perfect information is impossible, and the search for it can become problematic in itself. Just as in the case of Bayesian priors, we can start with assumptions to be evaluated.
Big Data as Cynicism
Cynicism creeps in, it is as Sloterdijk reminds us, a form of enlightened false consciousness. Instead of turning back toward the abductive, the aesthetic, and the political, one cultural move is to double down. If some data wasn’t enough, get more data. As the dialectical pair of induction/deduction recedes toward the horizon the status quo will be preserved. The appeal of Big Data is the purification of the inductive. Chris Anderson’s fever dream was that if enough data were to be collected that it might effervescently displace representation and argument. People have quite rightly responded to this non-straw-person for years. Keynes discussed this in his fragment aptly titled pure induction, which tends to think dimly of the idea of simply finding truth just sort of hanging out, doing whatever.
Just as Anderson seems to sublimate pure induction with the Lacanian Real, another psychoanalytic idea is useful for understanding the cultural script. Big Data shifts form desire to drive — the search for data becomes worthwhile in itself. Anyone who has tried to argue with someone opposed to vaccination knows that a cynical self-loop is joust about impossible to break. New information is only a part of the conspiracy.
As a strategy for retaining the status quo Big Data is well positioned. The critique of Big Data that does not deploy Big Data is in no position to win the debate — after all, if we keep seeing more data we can perfect the model, in the mean time retain the status quo. Big Data hucksters almost always have some normative line, a recommendation that they claim follows naturally (rather than abductively) from the dataset. This is the danger of not challenging data on the grounds of data, if the fantasy of pure induction and hasty normative generalization are unchallenged, they are seductively simple.
The tools of inquiry are as specific as the periods they explore, for texts produced reading texts that articulate desires to consumer products, the cultural studies of the 1980s would be well tuned. These were the means by which dominant cultural texts were constructed, making them fairly clear choices for deconstructing them as well, no Derrida joke intended. Revealing the operation of hegemonic groups already has a certain aspect of a weak theory of false consciousness, the practice of cultural studies in this sense is not merely the process of explicating case studies in articulation. At its best, cultural studies focuses inquiry on questions of power. Adding computational techniques to the project of cultural studies can update the theories of culture that drive research, accelerate the pace of cultural research, and cancel out the deformations in the argumentative economy driven by asymmetric deployment of computational techniques by hegemonic groups.
A computational critical/cultural studies is not a loss but a gain.