In many respects, the Summer 2020 Hedgehog Review – Questioning the Quantitative Life – presents a series of different perspectives on a singular dichotomy: the quantitative versus the qualitative. There is a hidden sense in this that neither is quite the objective, complete picture, something having been lost in translation. Accompanying this dissatisfied position is a feeling that data – the quantitative sort – is somehow nefarious, dangerous, and compromised. Leif Weatherby in his piece Data and the Task of the Humanities argues that language is not unequivocal, and therein lies its power (that mere data lacks) – ‘what the linguist Roman Jakobson called the ‘poetic function’ of language.’ Language has a nuance, a subtlety that data lacks. Language is a qualitative presentment of experience; as Weatherby continues, ‘[t]he very porosity of language means that datafication will never capture it entirely.’
However, if data is no more than an approximation of its subject, is this not also the case with language? Let’s consider how both are formed. Data is based on observation: a person counts the number of cars in a warehouse and enters the number in a spreadsheet. Similarly, a machine may visually interpret those cars and spit out an API return in the form of a quantity: there are twelve cars. Yet each is potentially flawed. Let’s leave aside the potential for base error – a miscount, for example. What if one of those cars is actually a van? What if one of them belongs to a warehouse worker, and is not part of the inventory? There is a subjective element to the assessment of what constitutes a car for the purpose of the counting exercise, whether that be in the mind of the counter (I know cars have four wheels etc.), or the machine and how its models have been trained to recognise cars. Data, therefore, is as porous as language, at least in its calculation.
There are these two elements of information as we experience it: first, its assembly, by some third party – a person or a machine; and second, its disgorgement, whereupon it registers with us individually, descending upon our ears, or our eyes, and being processed by the brain. The brain in turn analyses the information in the context of memory and associated cognitive functions. Perhaps it is our experience of the medium that the source of distrust lies: if we hear some information from a machine, or if we hear it from a person (a warehouse steward, for example), we may assign to it a value in terms of accuracy, bias or fairness, that varies based on the messenger. This then is where our tyranny of data potentially surfaces.
It is not in the validity of the information per se; information is information: it is not directly observed, and therefore in terms of our personal cosmologies and ontological sensibilities it is less than real: it is interpretation. The extent to which we may choose to rely on the information so derived may vary based on the application context. Questions of ethics and morality begin to come into play.
A challenge for our modern civilisation, then, is to determine whether in some sense we see data as more real than language, as a higher form of communication. Indeed, when considering the realm of the real, of truth, we soon arrive at a theological station: what is it that we truly believe?
Earlier in the piece, Weatherby refers to graphics, quoting Johanna Drucker, making the point that the visual representation of data is as important as the data itself. Some of the work of Edward Tufte leaves one with the strong sense that data becomes more than mere data when it is visually rendered in particular ways. This is not just the case of maps (witness Donald Trump’s sharpie-adjusted hurricane map), nor statistical fakery (such as selective x-y axes intersections to present a more compelling story, and seemingly false data relations), but it is an acknowledgement that numbers alone do not constitute computational communication. Marey’s Paris-Lyon train timetable being a case in point, there is more value in the data for its rendering. This then transcends the mere recording of the data, and its presentation – or more accurately its representation – enriches the communication in anticipation of a specific context, that of a reader wanting to know which train to catch.
The fact of representation reminds me of Velazquez, and Las Meninas. Velazquez was capturing an image here about representation itself, the subject arguably not in the painting at all, and yet the painting representing the subject quite well. At the very least, it was acknowledging the poverty of a physically inert, static artifact – a painting – as in some way showing the reality of the King and Queen of Spain. Painting, data, language – each is its own medium, which attempts in some way to capture in some way, through a hopefully shared ontology, some element of the world around us.