A librarian who mislabels a shelf does not lose a book; she loses the conditions under which it can ever be found again. That is the real stake of metadata, and it is a theoretical stake before it is an administrative one. Creator, date, version, relation, status: each field is a claim about which differences in a record deserve to survive contact with time and with machines. Most schema design proceeds as though these were bureaucratic conveniences, filled in after the intellectual work is finished. The reverse is closer to the truth. A schema decides in advance what kind of object something is allowed to become, and therefore what questions can later be asked of it. Call the moment when a format starts silently vetting which distinctions get admitted FormatAdmission: it is not a filter applied to knowledge from outside but a condition built into knowledge's own container. Systems that ignore this discover, usually too late, that they cannot retrieve what they never agreed to describe.
Grafton, A. (1997) The Footnote: A Curious History. Cambridge, MA: Harvard University Press.
Gieryn, T.F. (1999) Cultural Boundaries of Science: Credibility on the Line. Chicago: University of Chicago Press.
Abbott, A. (1988) The System of Professions. Chicago: University of Chicago Press.
Bateson, G. (1972) Steps to an Ecology of Mind: Collected Essays in Anthropology, Psychiatry, Evolution, and Epistemology. San Francisco: Chandler Publishing.
Bourdieu, P. (1984) Distinction: A Social Critique of the Judgement of Taste. Translated by R. Nice. Cambridge, MA: Harvard University Press.