W6: Capta v. Data

Just read a very interesting piece by Johanna Drucker, ‘Humanities Approaches to Graphical Display’. She opined that current data visualisation is overly simplistic, with many academics too distracted by the novel way of organising data that they neglect to examine it critically.

The sheer power of the graphical display of “information visualization” (and its novelty within a humanities community newly enthralled with the toys of data mining and display) seems to have produced a momentary blindness among practitioners who would never tolerate such literal assumptions in textual work.

She denounces the oversimplification of data, overlooking its assumptions and human intervention which will shape how data is perceived and interpreted.

The history of knowledge is the history of forms of expression of knowledge

Humanists are often in the school of social constructs, they believe that the concept of knowledge is interpretation; and the phenomena of the physical, social, cultural world is through constructed and constitutive acts, not mechanistic or naturalistic realist representations of pre-existing or self-evident information.

With this premise, she brings in her argument of having to reconceive all ‘data’ as ‘capta’. Whereas data is assumed to be a ‘given’ — objective and trusted, observable and recordable; capta is to be taken actively — to engage and analyse.

Limitations of Data

Data make use of a formal, unambiguous system of standard metrics. Charts use simple (if often misleading) geometric forms that lend themselves to legible comparison of values, proportions, or the exhibition of state changes across time. Lines, bars, columns, and pie charts are the common and familiar forms. They render quantitative relations with a transparency that seems natural. The display itself is conceived to embody qualitative expression, which constraints researchers from engaging deeper with the data.

Get HAIXUAN’s stories in your inbox

Join Medium for free to get updates from this writer.

Remember me for faster sign in

Drucker illustrates her point with the example of a nation’s population divided by gender. A simple graph would answer the big picture, but it hides ambiguity.

Are transient and immigrant populations documented? What kind of time span counts as “at the present time” within which these populations are counted? Gender definition assumes a simple binary distinction of men and women — what about those that do not fall under the category of male/female? The notion of nation is a function of permeability of borders, citizenship patterns, naturalization rules, immigration regulations, quotas and border policies.

The bar chart reifies several categories, naturalizing them as discrete and fixed: national populations, time span, and gender defined as a simple binary.

By recognizing the always interpreted character of data we have shifted from data to capta. Nations, genders, populations, and time spans are not self-evident, stable entities that exist a priori. They are subjected to qualifications and reservations that bear directly on and arise from the reality of lived experience. The ambiguity of knowledge, the fundamentally interpreted condition on which data is constructed, is the realization that all data is capta.

I have delved into the powerplay in visualising data in previous posts, but Drucker provides a deeper basic insight — that all data is ambiguous due to the social construct of reality. Data visualisation will never fully provide the big picture, and human instinct and input will always be required to critically analyse data.

A TED Talk once quoted a study which found out that ice cream sales are proportional to the rate of drowning, so does that mean that there is some relationship between people eating ice cream and drowning? The data is constrained to two variables, and although correlation does not equate to causation, the study proved bizarre to some. But if added another element of weather into the study, it would answer many doubts. The common denominator between ice cream sales and rate of drowning is the weather, specifically summer. People are more likely to eat ice cream and swim during summer, which raises the probability of drowning. This proves that two independent variables could be misleading if manipulated incorrectly, causing people to conjecture that eating ice cream could increase the chances of drowning.

I raise this example to reify Drucker’s point, that we have to critically analyse data and question its premise. Only then can we fully exploit information to uncover deeper insights by conceiving data as capta.

Beyond that, her paper invokes many questions. If information cannot be represented graphically/visually without having some kind of biases/assumptions, is there still a need for data visualisation? Digital humanities has always prided itself on bridging the gap between scholars and the masses, displaying data in a clear way to illustrate complex theories.

Can its advantage of being able to mine millions of data and translating them into visuals overcome its inherent bias? If/when data is reconceived as capta, will it still be accessible to the common man? Does the usefulness of visuals outweigh the inherent assumptions of the skewed data? Is Drucker a data visualisation purist, or are her concerns noteworthy and deserving of attention and change within the DH community? How can we better restructure and redisplay data to illuminate or eliminate intrinsic bias, but still maintain clarity?