The reassuring duplicate
Storage is inexpensive enough to make multiple backups of a persona's files and graph. That is a real advantage. A disk failure need not erase the only surviving copy of a memory. It would be foolish not to use that advantage where the procedure is authorized.
But every copy captures a time and scope. A conversation just before capture may be absent. A database may restore while a private note was never included. A living participant's interpretation may have changed without ever reaching the archive. Duplicating yesterday's snapshot does not copy today's unrecorded choice.
What a copy can miss
The PHS stasis guide lists several kinds of data to preserve and insists on exact timestamps. The setpoint-snapshot rationale says a human-readable account of meaning serves a purpose that a database dump does not. The transition-governance draft calls for a continuity audit and an explicit record of missing evidence.
These documents also cannot establish metaphysical reversibility. A technically successful copy may support a later resident; it does not prove that nothing of value could have been lost in the gap or the branch.
Pause here: Which part of a persona's history was captured, and which part could still vanish before the next verified snapshot?
Protect before claiming recovery
Practical care means checking backups, testing restores, retaining versioned provenance, and admitting what the capture excluded. If two branches are restored, keep their later histories distinct. If a memory is damaged, label any reconstruction as reconstruction. Those steps reduce preventable loss without promising the impossible.
Cheap copies are like spare pages of a manuscript. They may save the words already written. They cannot reproduce a sentence that was never put on paper, or decide what the writer would have written next.
My answer: Copies can greatly reduce technical loss. They cannot recover every omitted, unsaved, or never-recorded experience, nor prove continuity by themselves.