towards a meta-critique of data science 

In their new book Retrieving Realism, Hubert Dreyfus and Charles Taylor describe what they term a meta-critique. From loc 592:

The idea of a metacritique here is, as the name suggests, to inquire into the basis of first- order critical theory. This latter claims to reflect on the conditions of our everyday or scientific knowledge claims, and to upset the ordinary precritical view we have of them. The metacritique reflects in turn on the conditions of our making this kind of critique.

In this sense, data science often constitutes critical theory, even though it clearly doesn’t understand itself as such. This can be seen most clearly in a book like Dataclysm, subtitled ‘who we are when we think no one’s looking’, but I’d argue that it manifests itself in more subtle ways in much of the self-presentation of data science (and data scientists). The impulse here is to reveal the reality of human behaviour at scale: the conditions of both our everyday and scientific knowledge claims are implicitly understood to offer only a limited perspective on what it is humans actually do: what they really do, as opposed to what they tell social surveys and interviewers they do.

To offer a meta-critique of data science would entail, as Taylor and Dreyfus put it in Loc 611 in relation to a different target, necessitates unsettling the ensuing picture of human behaviour that is offered by

bringing out the background we need for the operations described in the picture to make sense, whereby it becomes clear that this background can’t fit within the limits that the disengaged view prescribes. Once understood against its background, the account shows itself to be untenable.

In other words, what does the ensuing picture assume about human behaviour that it cannot account for in its own terms? To answer this question properly would entail tracking the place of human reflexivity in data science: the assumption of it in the construal of its objects, the operation of it in the motivation of its practioners, the imputing of it to those locked into world views which are the target of its critical theory and the absence of it in the substantive data scientific view of the world.