Provenance & Forensics

A folder of artefacts and no log.

The normal condition, not a special case.

Residue, not proof

  • mtime before creation time: copied here, not authored here
  • an embedded application version: an edit placed within a range of years, sometimes an organisation’s licence
  • flattened styles: passed through a fixed-layout format, whatever the extension claims
  • a departed colleague’s name in a filename: who held it, and roughly when

Nobody curated any of it

Which is exactly what makes it useful.

And all of it is defeasible

Timestamps can be set. Metadata can be stripped by a tool with no opinion about it.

A reconstruction that doesn’t say what it’s unsure of is a story.

Archivists got here first

Provenance: keep records grouped by who created them, not reorganised by subject.

Chain of custody: record every transfer, so a gap reads as a gap.

A custody log read backwards from now: two entries recorded by a system, two inferred from residue, and an unknown gap before the earliest of them, stated rather than filled in.

1898

Three Dutch archivists publish a hundred rules for arranging archives.

Rule one, in effect: an archive is a product of its creator, and regrouping it by subject destroys evidence you cannot rebuild.

Endorsed internationally in 1910. Still the working principle.

Both say one thing

The history of a thing is part of the thing.

It cannot be reconstructed reliably once discarded.

Git did not invent an approach to provenance. It mechanised one that had been written down for a century.

Next week

Provenance answers what happened.

It doesn’t answer what to keep.

Sources

  • Handleiding voor het Ordenen en Beschrijven van Archieven S. Muller, J. A. Feith and R. Fruin, 1898; translated by Arthur H. Leavitt as Manual for the Arrangement and Description of Archives, 1940. The hundred rules that codified provenance and original order. Print; no stable public copy to link.
  • Archival Theory and the Dutch Manual Eric Ketelaar, Archivaria 41, Spring 1996, 31–40. Traces the Manual back to French respect des fonds and the German Registraturprinzip, and forward into practice. archivaria.ca
  • Digital Forensics and Born-Digital Content in Cultural Heritage Collections Matthew Kirschenbaum, Richard Ovenden and Gabriela Redwine, CLIR publication 149, 2010. What happens when archivists adopt the evidentiary methods of forensics: including the ethics of reading residue a donor never meant to hand over. clir.org