I would like to thank Antoinette Rouvroy (@arouvroy) for her comments in response to an essay I wrote recently about the connected discourses surrounding postmodernism and the nascent discussions of "Big Data." While in no way was the essay perfect in an academic sense, it was my hope that it would at least start to lend a hand in a larger assessment of a phenomenon that has great implications for culture, epistemology, and historiography. Rouvroy's criticisms are right on point, I think, because they scrutinize Big Data in the same manner. Though ultimately she seems to reach a different conclusion, a conclusion which I welcome and hope to consider as a way to sharpen my own position. 

While the comments were in twitter form, I think I grasp their point. However, I think they merit more than 140 characters in response, and I would like to continue the dialogue below.


Point one:

Despite what I think is a valid reading of the essay, I don't know that I necessarily equate analytics with a quest for certainty as much as I want/meant to point to the method of data collection and integration that analytics engages in, and what that says about the value given to empiricism. Big Data is a hyper-empiricism because, now that we can finally use so many more data, we potentially reanimate the critiques leveled at narrative during the end of Modernity. This began to seem, to me, to be very similar the loosening of historical agency that either lead to or was an outcome of postmodernism's imperative to take into account new types of cultural production and/or anti-teleological developments.

At its heart this essay was about historiography, and for me the shifts from modernist accounts to postmodern accounts carried the same ideological changes. A quote from Charles Harrison, historian and member of the conceptual art group Art & Language, brings this into light. For Harrison Modernism prized, or in fact required, “a critical difference and development with respect to other recent and approved ‘major’ work in the same medium—which tended to be sculpture or painting.” This again reminded me of issues of data integrity that come with traditional relational databases, something many Big Data frameworks attempt to circumvent through innovations in storage and organization. 


Point Two:

The above is a good point, and an example of how Big Data is at this point a catch-all term for both a research method (medicine, sociology, etc..) and a storage infrastructure (e-commerce, computing etc..) among other things. While this difference is obvious on its face, the difference also extends to the ideological questions I tried to raise, and so I think it makes some sense to separate them. In its manifestations in e-commerce, advertising, or risk models, etc.. I think Big Data does implicitly promise a sort of decisioning power that is assessed at the individual level. I don't deny that it contributes "to [a] multiplicity of impersonal behavioral patterns," though this aspect of its application has looser ties to the shift from relational databases to schema-less databases that is the heart of the discourse I am investigating.

One comment in response to this piece (written elsewhere) said "big data is just statistics with lots of data." While in part this is true, when you begin to ask about what kind of data are being used, then the structures required to process and store these data start to mimic the developments that accompanied the end of modernist teleology and narrative. Also, it would be a mistake to assume that the "just statistics" doesn't carry its own assumptions about knowledge. 


Point Three:

You noted the problems with the word "truth" above, and here I think it relates to your third point. In the end the word "truth" is scary and loaded even if you are talking about the denial of it existing as a knowable entity. So I suppose its use here was muddling my point: really I am concerned with the means and not the ends of Big Data. And perhaps here we could start to see a funny shape emerging here. When people talk about Big Data the marketing speak focuses on the ends, where most all of the other stuff is about the methods and tools, literally the breakthroughs in data science. So the layman's descriptions traffic in a language of "pinpointing", "discovery", "enhancement", that uses certainty as a form of currency, even it somehow knows it won't deliver it. It explains its worth by favorable comparison to legacy systems that could not "handle" everything "out there." What is "out there" is a another question about the promises its end users make for themselves. Postmodern accounts were, like Big Data, an answer to the modernist account that simply had a logic embedded in it that did not allow it to "scale up".


Final Point:

This last point is a strong conclusion built off of the initial statements about multiplicity and neutrality. We agree that Big Data carries with it a critique of existing assumptions and, in turn, knowledge structures, what with the way it holds out the promise to "reveal" and/or counter accepted notions about our world, a business, or a group of actors on a historical stage. 

AuthorMike Pepi