The contractor's name was Kestrel Avionics. Six months earlier, Evelyn had worked there. She had left after discovering that a test anomaly had been relabeled as harmless sensor noise. Her supervisor had told her she was too young to appreciate how much depended on public confidence. Evelyn had replied that confidence built on missing data was not confidence at all.
Her resignation had cost her friends, professional opportunities, and a comfortable salary. It had also cost her the simple belief that competent people would do the right thing when shown the truth. She carried a printed copy of the findings in her satchel because she no longer trusted every electronic channel.
Now, standing beside seat 2A, Evelyn recognized the Kestrel logo on Martin Hale's cuff link. It was a tiny silver bird, its wings folded into a geometric V. The discovery did not change the seat number, but it changed the stakes of the conversation.
“Are you affiliated with Kestrel?” she asked before she could stop herself.
Martin glanced at his sleeve. “One of our holdings. Why?”
“Just curious.”
“Then perhaps you know we're hosting an important meeting in Geneva.”
The flight attendant Amelia carried a small notebook in her uniform pocket. She used it for details the official service log did not capture: a passenger with a peanut allergy who preferred to confirm ingredients twice, an elderly couple anxious about their connection, a broken armrest that had been reported three times. That morning she wrote down the seat numbers, the time, and each instruction she received. She did not yet know what any of it meant. She only knew that good memory became unreliable when frightened people later told different versions of the same event. After the confrontation, a supervisor advised her to leave the matter out of the cabin report because it was already being managed. Amelia thanked him, kept her expression neutral, and wrote down that advice too.
Evelyn kept her expression still. She had been warned that Kestrel executives would attend her presentation. No one had mentioned the Hale family, which provided financing to several companies that built components for the contractor. A coincidence was possible. She had spent enough time analyzing failure patterns to know that coincidences were not evidence.
Evelyn's first encounter with Kestrel Navigation had occurred three years earlier at a demonstration in a windowless laboratory. The system was supposed to combine readings from several sensors when satellite signals degraded. A young engineer named Theo Mercer had guided her through a simulation showing a plane crossing a line of thunderstorms. He spoke with genuine excitement about redundancy, describing how one sensor could correct another's drift. Then, when a fault was introduced, the projected path on the display jumped half a degree before settling. Theo called it a visualization artifact. Evelyn asked to see the raw output.
The raw trace showed something less comfortable. Under a narrow set of conditions, a corrupted input could be classified as a recoverable anomaly even while it influenced the system's confidence estimate. The computer eventually rejected the bad data, but it took longer than the presentation suggested. Theo was not dismissive when Evelyn questioned him. He admitted that the model still needed work. The vice president leading the demonstration interrupted to say that the scenario would never arise in real flight. Evelyn asked how they knew. He smiled for the visitors and moved to the next slide.
After the meeting, Theo caught Evelyn near the elevators and handed her a page of handwritten notes. He asked her not to attribute it to him until he had completed more testing. She agreed to preserve the question, not to conceal findings. Over the following months, they exchanged technical messages about drift thresholds, recovery windows and the difference between a design goal and a measured result. Theo's later messages became shorter. Eventually he wrote that the company had reorganized validation and that he no longer had access to certain test archives. Then he stopped answering.
Evelyn was not an engineer of the kind who spent her days designing individual circuit boards. Her strength was translating between design teams, operators, auditors and decision-makers. Her father had insisted that people at the top of an organization needed to understand which questions they could not answer themselves. Evelyn found the most trustworthy experts were the ones who identified the limits of their knowledge. The least trustworthy were often fluent in certainty. She learned to make a separate column in every notebook for assumptions, and another for conditions that would falsify those assumptions.