Having an unpleasant reaction to Lisa Gitelman’s bad shrimp analogy

Digesting Lisa Gitelman’s admonition about data and rawness, most of her argument went down fine: after all, disciplinary cuisines aside (171), what she most stridently calls for is the disclosure of epistemological and methodological concerns, or recipes, and for us to discard the notion of any particular palate—whether one makes purportedly subjective or objective claims for a living—being truer than another. However, the jumbo shrimp analogy (168) stuck like a bone in my throat, in spite of how cheerfully Gitelmen provides and then discards it. I found it ill-conceived, and here’s why: “raw” is an absolute term, and while it may be an inaccurate and blinkered way to refer to data, it doesn’t really bear comparison to “jumbo,” which is a relative term for shrimp. These shrimp found here are more hefty than the already-large shrimp found near, which are in turn more generously-proportioned than the pedestrian, moderately-sized shrimp found over there—thus, “jumbo” in the first case. 

Mark Twain famously attributed to Benjamin Disraeli the quip fond to many a wag: “there are three types of lies: lies, damned lies, and statistics.” To cite either as the author, while often done, seems superfluous—it is such a truism that it doesn’t matter on whose authority we have it. And Gitelman and Manovich agree; both argue that the essential problem with data analysis, no matter how clever, is that sampling facts is a ticklish business. Carelessness and perfidiousness in the case of data collection and analysis end the same, and so whether by accident or design, prejudiced samples produce less generalizable results. The operative principle, then, is nuance, and not only acknowledging that data is always “cooked” but learning how to select it, season it, how to render most effectively the particular characteristics of interest. Thus, I would argue that in his call for “wide data” and in hers for frank disclosure of disciplinary predilections, Manovich and Gitelman in fact do apply the wisdom of fishmongers and think about scales of data as useful metrics. The jumbo shrimp, after all, is only oxymoronic in one iteration—elsewhere the langoustine and the tiger prawn produce none of the consternation apparently plaguing the American shopper.

I think it bears pointing out that the issues Gitelman has with jumbo shrimp and raw data seem to come down to the notion that somehow we’re being tricked into thinking the thing at hand it something other than it is: so, the shrimp is not a shrimp, it’s a jumbo shrimp, and data isn’t contingent observation about the material world, it is unbracketed, unadulterated truth. And while conflating industrial aquaculture and the empirical or positivist bent may work on many levels—after all, industrial aquaculture and many of its devastating consequences originate in notions that the world is entirely knowable and always improvable—there are some close to the surface in which that shorthand just doesn’t.

Advertisements
Having an unpleasant reaction to Lisa Gitelman’s bad shrimp analogy

The Wide Cooker

This week, Lisa Gitelman’s “Raw Data is an Oxymoron” and Lev Manovich’s “The Science of Culture?” introduce the data studies as a contested and multifarious field of study. Gitelman’s piece seems to address an imagined scholarly community that presumably treats data like a fresh pair of glasses: a clarifying, objective, and unquestionably useful tool to see the world. The thrust of Gitelman’s argument is to implore scholars to pay attention to the “pre-cooked” hermeneutics of data-vision and to point out the hegomonic ways in which data and other empirical tools have evolved over time. Additionallyshe draws a throughline between science studies and data studies by pointing to how objectivity in both cases is mythical.

On this last point, Gitelman and Manovich both treat data studies as a vibrational nexus between science and the humanistic study—an idea that has really strongly resonated with me in all of my own research. As I think about the possibility of using data visualization as a critical lens from which to consider either climate discourse or the circulation of fetal images, it becomes imperative for me not to pick a scientific view over a humanistic one and vice versa. I really want to hold these two tensions together: the long, globalizing view that data can offer (which will possibly be generalizing but maybe also generative) and the potentialities of data to speak to human actants, or what Manovich refers to as the “individual and particular” (Manovich 2016, 9).

Gitelman says that the imagination of data is always “an act of classification,” (Gitelman 172) and I believe that to be mostly true. But what I sensed after reading Manovich’s Cultural Analytics manifesto is that classification doesn’t have to constrict our study. Manovich’s call for a “wide data” argues against the categorization of data into discrete dimensions or variables. And while he doesn’t state it explicitly in this essay, I think what he means is that there is a use for views of data that are continuous rather than discrete. It reminds me of something Alan Liu once said about his own work in the digital humanities, where rather than trying to classify and plot the sentiments we encounter from large data sets, we can construct heat maps and look for hot spots.

I can think of ways that a wide cooker could be useful in both my projects. Maybe I can mine a lot of instagram images under a hashtag like #globalwarming and allow clusters of affiliation to form in ways that they could not if I just plotted them temporally or by geolocation. Or maybe I cull information from several pro-life websites and try to simulate my own anti-abortion page using textual and visual data. I also think Professor Sakr’s mosaics would actually be a great example of “wide data.” What other kinds of “wide cooking” could we come up with?

The Wide Cooker