The first multi-source reconstruction contained no beach to stand on.
Hessa selected three physical source groups: the modern atlas packet, the scholarly ship-loss rows, and the maritime clothing research categories. The translated accounts and glossary images remained outside.
The exclusions mattered. A translated narrative would give the claimant sentences to imitate. Glossary images would give it word pairs to mislabel as speech. This interval selected only geometry, dated ship-loss records, and equipment categories without wearers.
"If anything talks, it brought its own mouth," Jiang said.
Conti entered that as a test condition rather than a joke.
[MULTI-SOURCE MEMORY DIVE1]
[Selected source groups3]
[Modern atlas1]
[Scholarly ship-loss rows1]
[Generic maritime clothing categories1]
[Translated narrative0 | glossary0]
[Maximum interval01:00]
[People0 | speech0]
[Historical interaction0]
[Memory Dive proof2/3]
