CONCEPT
Datafication
Viktor Mayer-Schönberger and Kenneth Cukier's term for the transformation of a phenomenon into a quantified, computable format—distinct from mere digitisation—whereby aspects of the world that previously existed only as lived experience become data substrates for analysis, correlation, and machine learning.
Datafication is not the same as digitisation. To digitise a book is to convert its existing text into bits; the text existed before, and the bits merely carry it into a new medium. To datafy a phenomenon is to render an aspect of the world—one that was previously experienced but not recorded in computable form—into a quantified signal that did not exist before and can now be tabulated, correlated, and acted upon at scale. Location had always existed; the smartphone datafied it, converting every person's movement through space into a continuous stream of coordinates. Friendship had always existed; the
social network datafied it, converting the soft, ambiguous bonds between people into a structured graph. The concept, coined by Viktor Mayer-Schönberger and Kenneth Cukier in
Big Data (2013), identifies the hidden epistemology of
large language models and every other AI system: these systems do not learn about the world. They learn about the datafied trace of the world,