When do ordinary physical and technical events become preservation events?

A subsidiary question under Q-0019. §3.2 documents the physical/technical mechanism in the four-category §3 taxonomy.

The answer is C-0018: a backup that shares a failure domain with the data it protects is not a backup. The failure domain need not be physical. At Kyoto University in December 2021, buggy backup scripts and the data they protected executed in the same software context; a single administrative action destroyed 77 TB across 14 research groups, with 4 groups losing the only copies the center held. Brazil's National Museum 2018 fire destroyed ~18.4M of 20M items because there were no off-site copies of most of the archival material. An ordinary stolen laptop ended the only record of the Agh 2009 Artemia dataset because Urmia Lake has since lost ~88% of its surface area and the populations cannot be re-sampled.

Each event is a routine occurrence at every research institution. None becomes a preservation event in a regime where independent copies exist across independent failure domains. The C-0018 architectural diagnosis: independence of failure domains is the architectural input; "backups" that violate independence are not preservation, regardless of how they are labeled.