“We think.”
“No. We know that part.” I bring up the log timing Addy identified. “Megan responded. The system accepted her input late or recorded it late. Either the aircraft was already fighting her, or the log was altered to make her response look delayed.”
Jeffrey’s face tightens.
“She didn’t ignore the warnings,” I say. “She fought the aircraft.”
He looks at the screen.
Then at me.
I see the moment he stops trying to slow me down and starts following the evidence.
“Show me the timestamp discrepancy again,” he says.
I do.
The old report displayed pilot input thirty-one seconds after the first escalation warning. Too late, according to the investigation. Enough to support pilot error.
But the raw trace, buried beneath a reconstructed maintenance file, shows a manual correction command initiated nine seconds after the warning.
Nine.
Megan would’ve been fast.
The processed report delayed the record by twenty-two seconds.
Four years ago, the official investigation relied on processed diagnostic exports because the raw aircraft telemetry was partially corrupted during impact recovery.
At the time, the discrepancy looked like sensor degradation.
Now it looks curated.
Jeffrey leans in, reading. “That’s not impact damage, and it’s not sensor noise.”
“No.”
“Someone cleaned the log.”
“Yes.”
He swears under his breath.
I bring up the current demo corruption. “The same technique appears here. Standard diagnostics show the failsafe present. Under the trigger condition, it’s bypassed. Reporting remains clean until stress simulation.”
“Which is why the patch looked stable. Because it was hiding the same way.”
“Yes.”
Jeffrey drags a hand down his face. “Jeez.”
He turns toward me. “Who had access to Megan’s aircraft system before the flight?”
“Meaningful access?”
“Meaningful.”
“Three people.”