Multiple layers of backstories sustain every data point with contextual narrative informed by stylized theoretical assumptions, observational contingencies, and historical conditionalities. Data is entangled with informationally rich contexts, and without grasping this dynamic contextual information, the data are of diminished value, if not meaningless. Restated, the abstraction of data, the effort to make data “objective” and susceptible to computational analysis, involves the stripping away of context and is, thus, destructive of information, and value.
Conventionally, this contextual information is systematically removed by ignorance of its value or to achieve "scientific" precision. We remove context to edit out noise, and to reveal the interplay of discrete, measurable variables. This is of course foundational to the scientific authority of the social and behavioral sciences, but it comes at a “cost” insofar as it actively restricts or erases what might constitute dynamic configurations of contextual and emergent knowledge and highlight new risk. This reduction to a single variable or backstory removes the contextual narrative and our ability to make decisions.
We continually justify the same actions and outcomes as we cannot see where the removal happens. We define "good data" when it has contextual narrative and rich information on how and what the data was collected for and to prove what hypothesis. Conversly "bad data" has no contextual narrative. Wrong or data with errors is more obvious.
