Against Cleaning

Practitioners, critics, and popularizers of new methods of data-driven research treat the concept of “data cleaning” as integral to such work without remarking on the oddly domestic image the term makes—as though a corn straw broom were to be incorporated, Rube Goldberg-like, into the design of the Large Hadron Collider. In reality, data cleaning is a consequential step in the research process that we often make opaque by the way we talk about it. The phrase “data cleaning” is a stand-in for longer and more precise descriptions of what people are doing in the initial phases of data-intensive research.