SUMMARY
As has often been noted, BACHELARD's counter-intuitive orientation to scientific inquiry, with its rationalizing insistence on relational anti-essentialism, has profound implications for social research methodology. The question remains how this orientation might inform the actual practice of research. In this article we present a pragmatic response, one that emphasizes the need to scrupulously avoid the use of essentialized categories. Doing so involves much work and constant vigilance, for which technique is an absolute requirement. Our reading of BACHELARD therefore insists that productive research requires the artifice of a methodological technology that wrenches research from self-evidence whilst avoiding its ossification in theory. We argue that this continuous disruption and rebuilding of forms of thought is necessary but often neglected in social research; often simply because suitable technology is unavailable. By developing work by DOWLING (1998, 2009, 2013), we then suggest one that is. This is demonstrated by contrasting a diagrammatic technology known for only breaking weakly with established categories—BECKER's classification of deviance—with a relational space that achieves the rational artifice required (one in fact more consistent with BECKER's own pragmatic project). The value of the artifice a relational space achieves is then illustrated in the empirical context of digital file-sharing.