Ayana.

At thirty-six, a power surge during a storm corrupted Ayana's hard drive. Inside it were one hundred and fifty hours of recordings from the Kouré oral archive project. Twelve years of work. Testimonies from elders, midwives, herders, guides. Real voices of real people who had described how the last giraffes of West Africa nearly disappeared and how the community took part in their recovery.

Ayana didn't scream. She sat still for ten minutes. Called a technician. It took four days to confirm what she already half knew: eighty percent was recoverable. The remaining twenty — thirty hours from the project's early years — was gone.

Thirty hours. Voices of people who are no longer here. Sessions from the first trips to Kouré, when she still had no method and no professional equipment. Recordings made with whatever was available, under conditions that can't be replicated because the sources have died or lost their memory. No copy. No transcript. Nothing.

After the failure, she set up a double backup system: local disk and a server at the Alliance Française in Niamey. Never again a single point of failure. But every time she does a backup — and she does them often, almost obsessively — she thinks about the thirty hours that aren't there.

The one hundred and eighty hours that did survive are the result of twelve years of trips to Kouré, sixty kilometers on dirt roads, three rounds of tea before each recording, sixty-seven sources who decided to trust her. The metadata system Ayana designed cross-references source, date, location, topic, and relationships between testimonies. Each recorded hour takes between three and five hours of night-time transcription in the archive room of her apartment in the Plateau neighborhood. Metal shelves, notebooks sorted by date, headphones at two in the morning.

Bibata's recording — the midwife who told the story of the giraffes' recovery from the perspective of the women of Kouré — is safe. It survived the hard drive because Ayana had started making individual copies of the most important recordings before the failure. Not as a system. Out of instinct. Or fear. The difference, by that point, doesn't matter.