What If Written Language Was Never Invented?

The Kéita memory houses specialized, she learned over her first month, in legal precedent — a function that in a literate society would have been served by courtrooms full of filed documents and case law. Here, a senior memory-keeper named Adama held, by Naomi's best estimate after months of cross-referencing, something in the range of eleven thousand distinct legal judgments spanning four centuries, each one recitable on demand, each one cross-checked against at least two other memory-keepers trained independently to hold the same material.

"You keep asking how we know no one has changed a judgment to suit themselves," Adama told her, with the patient amusement of someone who had fielded this exact question from every outside researcher who had ever visited. "You would ask a librarian how she knows no one tore a page from her book and replaced it with a different one. The answer is the same for us as it would be for her: because more than one copy exists, held by people with no reason to agree with each other, and we check."

The checking, Naomi came to understand, was the entire architecture. No single memory-keeper's recitation was ever treated as authoritative alone. Judgments were cross-verified in triads, recitations compared syllable by syllable, discrepancies treated with the same gravity a literate society might reserve for a forged document — because, functionally, that was exactly what an unverified discrepancy was.